I use DeepSeek every day (via VSCode Insiders and Zed Editor). It's very affordable and, while it's slightly behind Claude (not sure how far behind Fable), it suits my working style well. I'm not using unsupervised multi-agent workflows and don't need a library of skills files - I'm writing most of the code and leaning on AI to help with mundane tasks - like;
- generating types for APIs
- generating boilerplate based on existing code
- improving existing code (adding error handling, timeouts, things like that)
- Writing SQL repository boilerplate / queries
- Creating implementations against hand written tests
- Helping me understand and implement APIs from third party libraries
- Writing documentation
I've spent like $2 in the last month and have used over 100 million tokens.
It's doubled my productivity and unlocked work that I could not have done before.
As an Australian, I'm not sure that I care about the safety of my data when it comes to LLMs. US companies already stole scores of data to train their models on and it's hard to imagine they suddenly grew some integrity. I'll care when regulators step in, until then it's out of my control so I'll just use the best price-to-productivity product available.
In terms of price-to-productivity, nothing will beat DeepSeek right now. For what you have described, all of the existing frontier models will perform well (probably about the same even).
If you expanded the list to very hard research tasks, Fable was so far ahead of the others that it doesn't even deserve debate. If you are a researcher doing something involving scientific computation or mathematics that wasn't rejected by the guard rails, and you were using Fable, that week was probably your most productive week ever. A couple of my PhD students effectively finished their current projects in that period by getting Fable to chew on it for 30 hours straight (not sure how I feel about that).
This was my experience too. I asked fable to implement a (quite complex, novel) CRDT engine. It did fantastic work for the 3 days I had access to it. The spec it wrote is exactly what I want. It used prototypes and examples to figure out some hard problems and answer a lot of complex design questions. Claude Opus has been lumbering along over the last week trying to turn what it did into a useful library. As far as I can tell, by trying to vibe together pieces of work fable did without understanding it properly. Opus makes mistakes constantly, misunderstands the spec, and it makes terrible engineering decisions. Earlier today it claimed it proved something was impossible. I asked it to think that through and it immediately backtracked and apologized. Was it right then, or is it right now? Claude has no clue.
I'm kicking myself for not using Fable more while I had it. Now that I've used fable, I'm not sure I even want opus any more. It might be more efficient in the medium term to just program everything myself until I have access to a similarly capable model.
I feel like Deepseek, opus, etc are only good at problems that have already been solved 100 times on github. They're like the iPhone 3G. Its exciting they exist at all. But subsequent versions make them seem like cheap junk.
I had a similar experience. For complex maths in my 3d engine around IK, made more progress in those 3 days than I would’ve done in weeks.
I recently discovered though that ChatGPT 5.5 Pro is almost as smart but only available to me via the ChatGPT app. I’ve been having it read my code and having Opus 4.8 use the ChatGPT app to collaborate. It’s a step down but its a temporary stopgap for complex problems.
It's been pretty cool (and a bit scary) to hear similar experiences from many other researchers and developers just how impressive Fable was for their particular workflow and the productivity it seemed to unlock. Personally, I have found that I can't let any of the other AI agents touch the code that Fable produced for me, since they consistently fail to understand the very delicate choices that it made to ensure optimal performance.
I'm an engine programmer in video games on a big project - recently I'm mostly just fixing bugs. And Fable has been absolutely phenomenal in those 2 days we had it lol. It was genuinely the first time I didn't really have to reproduce a bug to fix it - Fable was able to really understand the code and interconnections between the systems to reason why it might have happened and fix it. Opus really struggles with that in my experience, or it munches code for 2 hours only to come up with a completely wrong explanation, like, wrong if you think about it for more than 2 seconds. TBF I only had 2 days to play with fable, but it was incredible in that time - can't wait to have it back.
Something I wonder about in this domain is if this is the fault of the language.
Game development is usually in C++ or C#. With C++, bugs are a nightmare to find. C# less so, but there are still memory leaks possible.
I've been writing Rust for a few years now and it has been absolutely phenomenal. I get the performance of C++ and the only thing I think about when developing is the logic itself.
There aren't any mature game engines that use Rust (Bevy is rapidly improving, but it's no Unreal/Unity) - though you can compile Rust to a dll so it can be consumed by Unreal/Unity.
Due to the language constraints of Rust, I have found that LLMs need to work _way_ less to figure out bugs and render out code. The compiler gives very specific error messages and if it compiles, it works.
When we started getting access to Claude last year, it was barely competent at C++. Like someone who read a book about C++ but never actually wrote any, or had to debug any issue. It would confidently declare things which any C++ programmer could tell you were bollocks straight away.
But....now it's able to reason about things like memory management, thread safety and second order effects when multithreading very very very well. It's honestly incredible how well it can understand C++ code and figure things out. Opus is already pretty good(although far from perfect) but Fable was just something else. I could give it a description like "It was reported that only sometimes(1/20 repro) when the user opens the inventory and closes it, they can't move the camera afterwards" and it was able to find out both the inventory camera control and input code and pinpoint the exact race condition that could lead to this situation.
To train people to become researchers. If the definition of what a researcher is changes over time (due to AI or otherwise), then the nature of the PhD will also change over time.
I would say to push frontier of human knowledge forwards. Its not a high school anymore where you learn to learn. PhD is already a place where you should deliver value. (By "value" i don't mean commercial product)
> I would say to push frontier of human knowledge forwards.
Not sure if this is what you meant, but to be clear: pushing human knowledge forward is a practical prerequisite of a PhD (given it's pretty hard to convince people you're adequately trained to do such a thing without actually demonstrating it), but it's not the value of the PhD itself. The value is producing a researcher -- that is, someone who has the skills to continue accomplishing this in the future.
The difference is that if you happen to expand human knowledge by a stroke of dumb luck (or smart AI...) without actually having acquired skills to continue doing so in the future, then you're not really earning a PhD.
Depends on the field. There aren't really all that many research tasks that can be completed in a week. One of my favorite stories from the past few decades was the finding of accelerating expansion of space. It took three decades of scheduling telescope time, looking at exactly the right places in space at exactly the right time, to find enough supernovas to have data to even analyze. Not a whole lot of non-trivial science is just "read existing literature, think hard, and produce text."
But honestly, there's a lot that is (particularly for simulation like stuff which can be implemented in code). We should expect to see a bunch of extra output here, which might improve scientific productivity (even though the real bottlenecks are gonna be the academic publishers and their fixation on holding on to their copyrights).
In most cases it seems (at least to me and colleagues) to be turning out that picking best intelligence is a better option than picking better intelligence / dollar, assuming you can afford the cost. At least on interesting problems. If you’re doing generic web dev work, probably not the case.
What are interesting problems? Deepseek and some other open source models do decently well in the gpu kernel benches. https://kernelbench.com/hard Most people think they’re working on “very hard” problems and they really aren’t
Absolutely agree with this. As Louis Rossmann recently pointed out, it's the difference between a correct answer and a wrong answer; the correct answer is worth a good amount, while the wrong answer is worth nothing. Under this metric, for harder tasks, the most intelligent model is best per dollar.
That's why experience + cost effective model is IMO the best combination.
Experience allows you to design the skeleton where the implementation details are often inconsequential. There are relatively few scenarios where an LLM would need more guidance to render an outcome, but even dumb local models can do that.
Building a simple UI component vs an efficient multi-threaded bidirectional socket implementation (both examples of things I did recently with DeepSeek flash).
Angular can only be written in one way so the UI was trivial. I know the architecture for the socket implementation and the trade offs for various approaches, so I sketch out the implementation and get DeepSeek to complete it (error handling, keepalive messages, timeouts, etc).
I don't think I would have saved much time if I just asked Fable to "make the socket implementation" but even if it got it right the first time, it would have only saved me a few minutes given that's how long it took to write any way.
Even in more conventional applications, Node.js / React CRUD applications - "write a graphql query for blah" "add an endpoint to run query" "add validation to endpoint" are all trivial for DeepSeek flash. In most of these cases, I have found you're constrained by context window size because these are rarely well architected applications.
> Experience allows you to design the skeleton where the implementation details are often inconsequential.
Expanding on this thought a little more: it is possible to set up scaffolding that make incorrect implementations inconsequential. If the LLM can detect when the implementation is wrong, it can retry with the errors feeding back into the loop. This is shored up by up-front investments in tests, API definitions, strong types, linting rules, etc. the various cheap, fast Flash models do not need to 1-shot solutions if capable of autonomous reiteration.
This is a great point. In the past this is how I learned to architect software in general as you are never the only contributor and often must accommodate the contributions of contributors who are early in the careers. Good software architecture reduces the damage possible by low quality contributions.
I have had great success with contract driven development workflows in this regard.
LLMs being effectively very knowledgeable jr engineers, slot right into those patterns.
Another interesting thing is that Rust _really_ shines with LLMs. The compiler gives very helpful messages which the LLM bounces off very effectively. If it compiles, it works, you just need to double check the logic is correct.
By contrast, with something like Go, you'd need to be on the lookout for nil pointers, race conditions and so on. Node.js / TypeScript is less dangerous, but you still have memory leaks that aren't possible in Rust and lack pattern matching so must be vigilant about type discohesion.
This is mostly true and a month ago I would have agreed wholeheartedly. However, there are still situations where almost every line of code has to be deliberately chosen in a delicate way. Think research code or settings which demand a high degree of optimization. Opus or GPT-5.5 are absolutely miserable to guide in these settings.
"Well, don't use an AI for those purposes!"
That used to be true (and is now again true I suppose). Fable was a peek at something different, it seemed to be actually able to start to tackle these kinds of problems. That saves a lot of time, since checking these programs is often far, far easier than writing them. Experience is needed to check and scaffold still, but something like Fable becomes a prerequisite for these settings.
People are wasting tokens with perfectly fine models by not planning effectively. Anyone who says Opus is not capable of doing what Fable can probably are not planning effectively and just praying to the one shot gods.
What fraction of your work is "interesting problems", and what field do you work in?
> If you’re doing generic web dev work, probably not the case
I have a feeling the bulk of most people's work is "generic $X work." My take is people should figure out their mix of interesting vs boring work, and optimize accordingly. Flash models also tend to much faster,
Yep. After moving Deepseek for a long time I also moved to Mimo 2.5 pro. It's not only similarly cheap, but it's very good with 1m context window and gigantic tasks. Produces very lean code.
IMO US vs Chinese data vacuuming is a false choice. One of the main benefits of these open weights models is that you can get the privacy and cost savings by hosting your own model in cloud infrastructure.
Open weights models are only 4.5 months behind closed weight ones.
The fact that US considers propping its flagship technology by blacklisting competitors demonstrates how small the US competitive advantage really is.
Hosting DeepSeek Pro yourself is gonna be wildly expensive though?
You have a wide choice of providers available, so if you can find one you trust you can get inference without data harvesting and it's still very cheap. But dedicated HW is insanely inefficient.
Opus estimates you can do it for $13k/month if you get committed pricing on the HW.
The USA has an entire economic system to financialize your personal data and obfuscate your privacy. China can just sniff on their citizens API Calls, and use that to distill models.
They could just be flagging that they're considering blacklisting as a way to get some more Trump coin sold or other bribes. Or indeed to get further money from USA companies - which way will Trump decide, only the numbers in his family's bank accounts can give us a clue.
Using laws to ban competitors is just economic warfare thats all. Its got nothing to do with "national security", thats just the reason they give us normies. You should be safe in Australia since they actually need to be friends with China.
> You should be safe in Australia since they actually need to be friends with China.
As a non-Australian, I enjoyed very much reading about Australia in a book by Tim Marshall titled "The Power of Geography". I didn't quite realize just how vulnerable they become with China's ambitions and expanse in the South Pacific due to their reliance on vital sea lanes for trade with its Asian partners. After reading that book, I can appreciate your comment much more.
Economic safety can be guaranteed in many ways. It's not economic safety that these moves are for. It's for blocking anyone from preventing them from maximizing the value the Western oligarchs extract from both their own citizens and the entire world.
What? Every country has the right to freely decide which companies are permitted to operate under its jurisdiction, and exercising that right isn’t “warfare” of any kind.
I think the country that talks the most how the market regulates itself and needs no state intervention isn't the most qualified to put tariffs and block other countries from trade. And yet here we are…
How are you accessing their API? Through OpenRouter, or direct? Are you using DeepSeek v4 Pro? $2 seems a lot cheaper than my own experience accessing them through OpenRouter for over 100 million tokens, but I am using OpenRouter to access v4 pro.
Cache hit rate dominates your total cost calculation for long agent session, and it largely depends on the provider. Deepseek's native deployment is probably much better than third party in this regard. For v4 pro it's a whopping >100x price difference between normal input vs. cached input tokens.
I am using Flash and accessing the API directly via vscode insiders and occasionally Zed (it's buggy but I keep coming back to it because I want it to succeed).
Unless you need enterprise multi-model management, I don't see the point in OpenRouter as it just adds cost overhead and you can just self-host an open router alternative (LiteLLM, Bifrost, etc). Running an LLM gateway locally is kind of nice as it allows you to normalize your configurations against your internal gateway - but I haven't really needed to.
Deepseek direct is atleast 2x cheaper than other providers serving it. Maybe their caching strategy is just significantly better or they’re subsidizing api pricing. I think it’s the former. $2 is closer to 50 million v4pro tokens in my experience
Ensure you're hitting the DeepSeek provider via OpenRouter - they have the massive cached tokens discount. If you're hitting any other provider you're paying an order of magnitude more.
Pro is substantially more expensive than flash. In addition, there's wide variance in price with DeepSeek themselves providing the cheapest tokens last I checked (but they train on them). Caching policy also varies by provider. TTL can be as low as 5 minutes or as high as 24 hours and reading from the cache might or might not reset the timer. Whether or not you get a hit makes (IIRC) a 10x (edit: it's actually 50x) price difference in the case of DeepSeek themselves.
I second this. I’ve been using it a lot (with OpenCode) for personal projects. It’s intelligent enough at a tiny fraction of the cost of Claude or Codex.
The fact that America has yet to figure out the importance of soft power (and TRUST in particular) in the AI/information age is mind boggling to me.
The CLOUD act, FISA rulings, the Snowden leaks, and now the aggressive tech oligarch push to weaponize the unholy combination of AI, MAGA, and social media algorithms in an absurd and patently obvious attempt to impose (or maintain?) "world domination" seems likely to cement their downfall.
"Authoritarian" China's low-cost and open source approach looks downright democratic by comparison.
It seems increasingly clear to me that AI is forcing a reckoning in terms of how we interpret authority and control in light of new ways that information is evolving. The old political labels seem grossly insufficient to describe the present reality and the ones whining about "democracy", "freedom", and "liberal values" increasingly sound (and act) like bitter, out of touch old men desperately clinging onto a world that is rapidly outgrowing them.
Last time I checked, the closest approximation of liberal values are "liberty, equality, fraternity". They seem to think that "liberty" should only apply to them, "equality" is a threat to their consolidated power, and "fraternity" is something that should be weaponized to turn communities against one another to distract from the ways the oligarchs have been and continue to abuse and consolidate their power.
I can second this. Deepseek is great for mundane work day tasks.
I use it via claude code, just pointing the api to deepseek.
It's also not a clear "Opus 4.8 >> DS 4 Pro", I've done 16 tasks in 4 days across the two, and while Opus was indeed on average better, both models performed well being able to handle most of my workload.
In fact DeepSeek was _significantly_ better on 3 task out of 16 and Opus was _significantly_ better only in 2 out of 16.
So why I still claim Opus 4.8 to be the winner? Because the few times that DS failed or got off the rails, it failed much harder and needed several prompts to be realigned on the actual tasks.
Another thing at which Deepseek is significantly behind is code reviewing. Opus is more intelligent/thorough, Deepseek will sometimes generate bogus or low quality feedback.
And the last thing at which Opus is better, period, is vibe coding. If you want to implement features end-to-end it handles ultracode flows quite better. I don't vibe code at work, but I do so with personal projects.
But the cost concern is real. I've spent sub 2$ in 5 days of work using DS 4 Pro, which is on average just 4 queries to Anthropic.
Give me a slightly better DS 4 (it is still in preview and training isn't finished) and I may ditch Anthropic for good.
Have you tried glm 5.2? It’s better than ds4pro and very close to opus4.6. Alot more expensive than ds4pro/mimo but significantly chesper than opus 4.6 for 90% of the quality
> As an Australian, I'm not sure that I care about the safety of my data when it comes to LLMs. US companies already stole scores of data
I'm not sure I understand this. I'm not defending the US, but isn't your data being in more hands worse?
Also, isn't Australia in a more contentious situation with China? Them being more allied with the US and all? Not to mention the whole nuclear sub issue. Having data stolen is shitty either way but isn't data taken by an adversarial country a worse situation?
If I spoke freely about how I feel about the US right now and the direct personal impact of policies implemented by the recent administration, I'd be denied entry into the country to visit friends later this year.
It is inconsequential if the US or China have my data, both will misuse it and I am powerless to protect myself from that fact.
Not using LLMs presents a bigger threat to my career than protecting my data.
Correct - As an Australian I feel free to say anything. But as an Australian who might like to travel again to the USA at some point in the future, I do not feel I have freedom to share openly online.
Whereas, China does not ask for my social network logins, or for me to be pre-approved to travel there. So that is unlikely to be affected.
> If I spoke freely about how I feel about the US right now
If you read my comment as defending the US then you've misread. Also, you're probably just as pissed as 60% of Americans
> It is inconsequential if the US or China have my data
Sure it does. The way each distributes data between government and industry has some differences. So too does the different disinformation campaigns each country is running against Australia.
But the main point is really that 2 > 1. 1 country scraping your data is bad. 2 is worse.
To be fair, most of their votes don't actually count.
And even with our westminster system, One nation has swept the right wing. Considering how many seats they will likely get in the next election, my high horse has shrunk a fair bit.
Why do you think I like them or support those actions?
Do you support everything the Klingon Empire does? That doesn't seem like the Worf I know. The Worf I know can distinguish the people of a civilization from the leaders of their societies. To conflate them just creates the problems you are criticizing
The same fact is true in a democracy. A democracy specifically is set up so just a single person doesn't rule. That's called a dictatorship.
But you turn a democracy into a dictatorship through tribalism. So forgive me if I get mad at you trying to place me into a nice well defined box. You're helping erode the democracy I (and so many others) desperately want to maintain (and improve).
Sorry that you're being blamed. People have a pretty negative sentiment towards Americans in general right now because we have been watching helplessly in horror as our best friend tears things down.
I have friends in blue states who didn't vote because it wouldn't have changed the outcome given the electoral college - it's unfair to blame them for not participating as they were never able to in the first place.
Not that democrats are angels. You guys are damned if you're blue / (extra) damned if you're red - given the state of both options.
Consider that USA's government logic is: "if we don't like a government we will kill civilians until they rebel and put a government we like". See Cuba for example.
So, by their own metric, they aren't helpless at all and should appreciate other countries aren't as savage as they are with civilians, instead of complaining.
Look Worf, I'm just one human. You don't know me. You don't know what I'm doing or not doing. But I know you're making our situation worse and I know you're not living up to the ethics of your namesake
As an Australian.. politically I need to worry about business data touching China. It will come up at a Risk Advisory Committee meeting as a serious issue.
In actual personal practice, no. China having my data presents no actual impact to me, America will do things that impact me.
China is mostly interested in geopolitical stuff and getting an economic advantage, plus they have no jurisdiction in the US. Your data in the hands of the US government however could potentially land you in prison.
Most Aussies aren't really worried about China... China remains our largest trading partner, and polling shows less than half of Australians think the AUKUS alliance (which includes the nuclear subs) makes our region safer.
Latinamerican here.
When you talk about "adversarial country" I think of the USA (they can kidnap a president, kill people on boats without a trial, etc) and not China.
YMMV for different regions.
China is not digging through my social media in order to find a reason to cut my grant funding. China isnt going to pull me over for speeding in montana only to examine my phone to see whether i am maga enough to get off with a warning. And china isnt at o'hare security scanning for anyone with skin darker than freshly fallen snow. China may be evil, but it is a far away evil that doesnt have a physical impact on my day to day. The other evil is much closer to home. Even if it is not the biggest, the crocodile closest to the canoe is always more concerning than the one still on the bank.
While having data/code stolen isn't ideal, there is a certain point where you need to assume it's already out there. There's actually more probability of harm from shady US companies imo, because people are less suspicious about data sovereignty
Historically speaking, the US might even be a larger risk to Australia than China is. The US alliance goes back a long way, and so does the opposition to US influence. Since the Whitlam government, MPs are generally fearful of US retaliation, rendering much of our politics hostage to US influence. Of course, some PMs have openly embraced the US so this feeling isn't universal. But many of our issues are directly tied to the US.
China has been far more beneficial to Australia by comparison, with the downside being the encroaching influence of CCP propaganda. Many of our strengths are tied to our relationships with Southeast Asia.
Paul Keating has famously declared US as an "aggressive ally", "our colonial masters", AUKUS as our "worst international decision", and that "our future is in Southeast Asia". This was under Biden too.
So the situation is much more complicated, and the feelings on the ground right now is that the US are not our friends (of course, the CCP is not great either).
The same Kim Dotcom who was arrested in Germany, extradited from Thailand to Germany, and most recently has been parroting Russian propaganda points about Ukraine? The same one whom the New Zealand legal system has allowed to exhaust every legal avenue and remains free to this day?
There are much better examples to use to showcase the US's extrajudicial/international reach than this guy.
Not jailable. But I'm absolutely free to disregard moral agency of mr. Kim Dotcom and stop paying attention to whatever he says or what happens to him as a result of his support for warring Russia.
Laws are ultimately defined by the boundary and edge cases.
The godly person, doing no wrong, that through unfortunate circumstances, is brought before the court does not advance our understanding of a law.
It's the messiest case, by unscrupulous persons, in unconscionable circumstances that ultimately decide if the execution of that law gives the public more or less rights than on initial interpretation.
Yes, but public minds are not made up by edge cases. Surely a better example could be used rather than a divisive criminal.
Also, not sure how anyone except bad actors or sloppy readers could have gotten that I was defending the US here instead of criticizing using Kim Dotcom as an example in characterizing the US's actions.
> Also, isn't Australia in a more contentious situation with China?
China does not run 'rendition flights' to kidnap other countries' citizens and hold them in gitmo style prisons for decades. China cannot compel and does not compel random countries to extradite foreign citizens to China to persecute them and lock them up. Guess who does.
I've been using deepseek for some development at home and it is really good for the price. It is at the point where i am ok with using it as a tool that i can rely on and not an expensive gadget with flaky uptimes.
I'm in favour of sending as much data as possible to DeepSeek for them to train on. I'd happily zip up everything useful I have and send it to them if they asked for it.
I'm not interested in explaining to you something immediately evident that you clearly disagree with. If you don't feel that trillion dollar companies training commercial products on the free labor of human writers/artists/engineers/etc. without their consent or compensation isn't "stealing" you're welcome to that opinion, but I believe that's what the GP was implying.
I'd be fine if it cut both ways. It shouldn't be legal for works distributed by people under licenses that would prohibit derivative work whilst if I reverse engineered a Nintendo Switch 2 with an LLM I'd be sent to the underworld.
Noteworthy that Z.ai, maker of the just released near-frontier GLM 5.2, has already been on the Entity List since Jan 2025[1]. Being on the Entity List does not mean all trade is forbidden. Broadly speaking it means American companies and individuals are not allowed sell them goods and services, but they are still allowed to buy from them and pay them.
AFAIK the Chinese AI companies barely depend on US goods and services, except for nVidia GPUs which were export restricted anyway, so it doesn't seem to be very consequential (see Z.ai). For the RAM maker CXMT it could be a lot more problematic though.
How does that even work? If Z.ai wants to buy lets say GPUs for AI training, what's stopping them from going to a local reseller? Its not even circumventing the rules, its the natural thing to do.
For that matter, does (only) NVIDIA make datacenter cards? When I buy a gaming card, I dont buy from NVIDIA, I buy from an integrator, like Gigabyte, who work with a company like Foxconn to make the cards.
The GamersNexus documentary (https://youtu.be/1H3xQaf7BFI) on the semi-underground GPU trade in China, while a little amateurish in terms of depth and general atmosphere, is an interesting watch and may answer some of your questions.
Basically, those export controls make GPUs more expensive for affected parties in China, but don't effectively stop them from being acquired or used over there.
The export restrictions only apply to certain GPU models, which are the more recent powerful ones used for training tasks. So the H100, B100 etc. are banned along with 4090, 5090.
Nvidia has downgraded chips that aren’t banned. H100 is banned, H800 is allowed. A100 is banned, A800 is allowed. But the sale has a tariff attached.
That’s all how it’s supposed to work. In practice companies probably circumvent the restrictions.
> That’s all how it’s supposed to work. In practice companies probably circumvent the restrictions.
Ex-colleague of mine told me he used to work for this company in UAE (he told me this story 15+ years ago, so he worked there even before that). He said it took him months of working there before he discovered that their entire business was evading US sanctions against Iran-they’d order servers/etc from the US, tell the US vendor they were for use locally in UAE, then ship them straight across the Gulf. The UAE government presumably knew this was happening but chose to turn a blind eye; the US government likely did too, but struggled to tell which orders/purchasers were legitimate and which were sanctions-evaders, plus likely was worried about enforcement action causing issues in the US-UAE diplomatic relationship.
I’m sure there are similar businesses out there who specialise in evading US sanctions on China.
Can I tell you I can go downstairs now and buy a 5090 (not d or d2 version, the real 5090) in China for 4000usd (33000rmb)? They have it stock right now, they have since release. It’s a Japan version.
I didn’t ask about h100 or others but can’t see why they wouldn’t have.
Good context. And it shows the limit of this kind of control: the model at the top of the open-weights index this week comes from a lab already on that list, and that changes nothing about whether I can run it. Once weights are on Hugging Face the download doesn't care about an Entity List. Chip export controls bite. Weight controls mostly can't.
> except for nVidia GPUs which were export restricted anyway,
those export restrictions are a joke. when they were introduced, there was a sudden spike in NVIDIA GPU exports to surrounding Asian countries. and the US government knows this
CXMT memory maker will not be banned, because US AI labs are salivating at the idea of more RAM supply, and are lobbying hard to prevent restrictions
There is no single "it", the Singapore loophole is well known, and nothing has been done about it. Except by China, once , when it blocked the sale of Manus to Meta, even though on paper, Manus is a Singaporean enterprise.
What's the publicly stated/marketing reason for capitalist America to put companies on the Entity List? Genuinely asking. Because to me it screams 'we were only for the free market until there was no competition'
"First published in 1997 to inform the public on entities involved in disseminating weapons of mass destruction, the list has since expanded to include entities that engaged in "activities sanctioned by the State Department and activities contrary to U.S. national security and/or foreign policy interests"
So RAM chip makers when there's a RAM shortage must be 'contrary to U.S. national security and/or foreign policy interests' i.e. the US government is trying to squeeze its citizens on RAM prices.
> Other way around according to what GGP quoted, this would get RAM into the US but not out, reducing prices.
Only in a world where the other party has no agency. In real life the other party raises prices on their exports to compensate for the supply chain disruption and they still get the items.
Ultimately the consumer pays more, the extra goes the government, and the net impact is just obfuscated taxation and a reduction in both supply and demand that's bad for the economy and other living things.
Free markets generally only make sense when at the same scope of the ruling government. When country A can manipulate markets in ways that country B can't or won't, eventually country B will attempt to make trade rules that level the playing field.
Its also worth noting we don't really have free markets in the US anyway.
The definition matters for sure. IMO, markets (like speech) are either free or they aren't.
We have a lot of government intervention in markets today. From subsidies to tax incentives to regulations and import tariffs, markets are much more controlled than it seems on the surface.
A free market can't exist without heavy government interference (property rights to start with) but not all markers with heavy government interference are free, either
Market regulation and protection of property rights are quite different though. Protecting rights agreed upon by a strong majority of the population isn't really heavy government intervention in my opinion.
Free Market vs. planned economy was always mostly talking points, not a consistent ideology imo.
Even during the Cold War, American farms were heavily subsidized. The abundant supermarkets were not a product of free markets, they were a propaganda piece.
Today the US is pretty far from being a free market. Tax deductions are subsidies. Industry subsidies fund things on the front end, and bailouts are essentially subsidies after the fact.
And on top of that there are plenty of (good and bad) regulations which distort the market. For example it is illegal to import foreign insulin even if it would be cheaper. In most parts of US metro areas it is illegal to build multifamily housing.
Most of the things you list don't make a market non-free. A free market can still have government regulation and distortion. In fact it requires it otherwise it will be captured by large players in short order and become non-free as a result.
The insulin example I agree is non-free. More generally the entire medical sector is only somewhat free. However I'm not sure that's a bad thing given the stakes and the history of the free market as it applies to healthcare. The medical establishment itself is an only barely disguised guild system after all.
> A free market can still have government regulation and distortion
The question is "how much" rather than a binary consideration. With the size of the government in most countries, we are way past what would be considered a marginal influence.
> an economic system in which prices and wages are determined by unrestricted competition between businesses, without government regulation or fear of monopolies.
By definition, free markets do not have government regulation. If you have an alternative definition of “free market” please feel free to share it.
I think you're misinterpreting the intent of what you quoted. I also think the phrasing of that definition is quite sloppy, being prone to exactly the sort of misunderstanding we see here.
For a market to be stable, fair, and free of monopolies government intervention is required. I don't think that fact is at all controversial. So already the definition you've quoted there has, if read literally, set up a scenario that is impossible to realize.
A free market is one where regulations are broad and are applied to the players evenly. A subsidy that applies to a sector as a whole (ex solar panels) is an example of such.
Many of the agricultural subsidies are much more targeted than that. However the regulator needs to maintain the stability of the entire system as a whole, and to that end food is not some luxury good that can be subject to shortages without major consequences. A tradeoff has to be made, either for more market regulation or significantly less market stability. Market participants in danger of starvation not known for exhibiting reasonable, well thought out behavior.
Similar to free speech, the free market is better seen as a vague guiding ideal rather than an absolute and objective description of a system. It's illegal in the US to make credible threats of violence towards someone. Can that fact be taken on its own to mean that we don't enjoy freedom of speech?
I am not misrepresenting the intent. The definition says “without government regulation”. If their intent was that some regulation was allowed, they would have said that. It’s the job of a dictionary author to be precise.
If I define a “two legged stool” you can’t add a third leg just because two legs is unstable and doesn’t lead to good outcomes. Whether a market with no government regulation is stable or leads to good outcomes does not change the definition.
You are referring to a “mixed economy” which has some markets and some regulations. A mixed economy is more stable because it is able to reduce market failures like monopolies and externalities that free markets cannot escape.
You're arguing semantics and have rendered the term meaningless in the process. The purpose of a dictionary is to communicate meaning. The meaning of a word or phrase is the idea that it is employed to communicate in practice.
When people talk about a free market they aren't referring to some logically contradictory thought experiment. It is something that they actually wish to strive for in practice. Thus it is clear that some amount of regulation must necessarily be involved.
This is quite similar to freedom of speech, as I previously mentioned. The concept itself is an abstract ideal and the context of the law matters.
What you are referring to is a laissez-faire market which as history has repeatedly shown doesn't remain a free one for long. An adjacent comment pointed to property rights as an example. You can have a free real estate market without having a laissez-faire approach to property rights, yet the government involvement in that case is extremely heavy handed. The free speech analogy might be a city controlled by the mob where the government won't intervene but saying certain things will still get you killed.
I agree with you that in practice the US is a mixed economy, but then I already acknowledged that (among other things) our farm subsidies are highly targeted and the medical sector is regulated in a very invasive manner. However I do not agree with your suggestion that eliminating monopolies makes a market mixed. A monopolized market is not a free one because there is no competition.
> For a market to be stable, fair, and free of monopolies government intervention is required. I don't think that fact is at all controversial.
It is. Government intervention is what creates monopolies, because economic value stops being the ultimate metric that decides which enterprise is profitable over another; because it favours one entity over another, large corporations are not valued on their financials alone, but on the strength of their political connections - which in turn enables lobbying as an effective tool.
The entire premise of free markets is that the market is the ultimate judge of value. If you add the government and its heavy hand on the scale to the mix, it is no longer a free market. It’s pretty simple.
As an explicit example, major revenue streams that Tesla (used to) take advantage of are
1. the EV tax credit, and
2. carbon credits
so the richest man in the world/the US had significant tailwinds for a central business venture of his via either directly taking money from the government, or taking money from other companies due to the government requiring they give him money.
> The abundant supermarkets were not a product of free markets, they were a propaganda piece.
The US has abundant supermarkets to this day. It was overwhelmingly a product of the market economy and remains so. The US has between 45,000 and 75,000 supermarkets (75,000 if you include supercenter stores that also sell groceries). That's not counting smaller specialty food stores.
It's a product of consumer spending capacity (net disposable income), of which the US has an enormous amount and has for over a century relative to other nations.
The common thread here is that it is China. Before the 2010s and earlier, the US wasn't so concerned about China, but ever since then, China has been a big US concern for it being a technology and military rival
e.g. the Investigative Report on the U.S. National Security
Issues Posed by Chinese Telecommunications
Companies Huawei and ZTE report from 2012:
From that point forward, that concern has only grown. So you can view these actions as screaming "we were only for the free market until there was no competition" but if you want to genuinely know the answer to your question, the publicly stated concern is "China is a national security threat"
> What's the publicly stated/marketing reason for capitalist America to put companies on the Entity List?
“Capitalism” is (as a result of propaganda by its defenders after it was named and accurately described by its socialist critics) often mistaken for a dedication to free trade, but capitalism is a regime characterized first and foremost by society being organized around the interests of the capital-holding class, the first of which is the preservation of the situation in which society is organized around the interests of that class. The reasons companies are put on the Entity List is because they are broadly seen as a threat (long-term or immediate) to the continuation of that regime. That’s what the “foreign policy and national security interests” that form the official basis of the Entity List ultimately, generally, boil down to, in one way or another.
(They don’t always boil down to that, because why the US is basically a capitalist system, it is not purely one, and even in a more pure capitalist regime, individual influential decision-makers may have other interests that they act on besides the implementation and preservation of capitalism that end up getting reflected in policy.)
DeepSeek rose to fame through stolen IP from U.S., and created shell companies to bypass U.S. law. Typical CCP-inspired behavior. The free world needs protection from abusers.
I'm sure there's a ton of other abuses they've committed as they race to become China's preferred LLM.
I assume you're not confused about the fact that DeepSeek winning would be against US self-interest, and the thriving of OpenAI + Anthropic domestically and globally is very much in the national interest of the US.
Even if both groups did exactly the same thing, it would be irrational for the US to not bias itself in favor of its own businesses.
Your original comment implied that stealing American IP and creating shell companies is “CCP-inspired behavior”. This suggests there is some high minded Rule of Law based justification.
Now you are saying that stealing IP was never the problem, and actually the issue is anything that stands in the way of the US gaining wealth and power.
The US should consider legalising some of this stuff, if I look at a leaderboard something like the top-10 models are built by companies facing serious accusations of copyright infringement. I assume the Europeans are obeying the law or whatever which is why they've so far only achieved peak-2024 performance or whatever and are making no particular contribution to the cutting edge, unlike the Americans and the Asians.
Come again how these laws are promoting useful results? They seem to be economically crippling. The free world should consider embracing freedom from these laws as it seems that will bring greater prosperity.
that is open to debate. Any commercial activity with a sanctioned entity has a pretty broad interpretation. Companies might not want to take the chance even if they are "still allowed".
So this is the other side of banning American models for non-Americans? And how exactly do they plan on enforcing all of this? Great Firewall of America?
This is a complete joke. The malicious clowns behind this should be removed from power and prevented from ever holding any position of power in any form of governance system.
China is already going as hard as they can on their own GPUs. When has availability of non-Chinese tech in China meant China didn't ultimately come up with homegrown replacements?
It's just my general impression. They banned China from the ISS, China made their own space station. China's making their own x86 chips, their own GPUs.
As a fellow wheel reinventer, I admire their audacity. It's the sort of thing that makes me wish my country was like China.
Classic case of getting downvoted solely because it's a negative comparison to China. The letter for letter exact same comment in a thread only about the US on here, that doesn't mention China, would get upvoted.
You are, of course, right. All of the downsides with none of the benefits.
Well, it's not as if there's not an elite enriching and entrenching themselves here either.
Reading about Thiel's "Dialog" meeting, I wonder, do China have their own equivalent of such blatant corruption meetings?
Even if they do, do they discuss "compliance collars" for the security guards in those meetings? I have a feeling they don't.
The future the Chinese elite wants to own, at least seems to include us as something other than a zombie army at the gates of their post apocalyptic bunkers. That's a low bar I know, and not much to be hopeful about, but our elites aren't even meeting it.
>The fact those plans also have subclauses ensuring the party elite are made even more wealthy and powerful is less alluring.
In which major democratic western country don't the elites get wealthier? In fact, the top 10% asset owners saw the biggest post-covid recovery, which the rest stagnated or are in decline. Not defending China but are own oligarchs are just as savvy at enriching themselves while screwing us over.
I think the PPP calculations may suffer from the same problem as the USA inflation rate being nominally low. Housing, food and medicine are all ludicrously cheap outside of tier 1 cities.
You might want to compare Singapore with a city like NYC or London, not with a territorial state. It's pretty normal around the world for cities to be replenished mainly by people moving in.
(Of course, to be fair you then also need to compare GDP per capita against other cities. And they usually do a lot better than territorial countries that include a lot of hinterland.)
> A low TFR is indisputably the slow death of a society.
No this is a non-sequitur.
Low fertility only means that the affected society grows smaller. This is only equivalent to "slow death" when it persists for centuries until the society is actually dead.
Whereas a society replaced by its own children maintains continuity of community, nation, or broad grouping of people having common traditions, institutions, and collective activities and interests
Not only x86 chips, they are going in fairly hard on Risc V and Loongarch (MIPS/Risc V inspired ISA). Risc V is still growing trying to catch up to ARM, while Loongarch LA664/LA864 chips are much closer to x86 performance than other options. They still are many years behind but not as far as you would expect.
GPU's are still a fair way behind with Moore Threads S80 being a better example of their high end. I suspect they have some major driver issues because they current benchmark far below what that silicon should be able to do. https://en.mthreads.com/product/S80
There is also the pressure to have them innovate on older process nodes so they can make this stuff domestically. For instance Huawei is doing what they call 'logic folding' which is basically just stacking dies in a way that ends up reducing the overall size of chip features. Not sure how it addresses thermals but it is a cool idea.
They greatly increased efforts when the US restricted high end exports to them. Unless further restrictions accomplish something worthwhile in the short term they seem unlikely to be of benefit to the US.
You're aware that many countries are blacklisted from trade with the US already, along with certain segments of existing companies. It just means that enforcement comes with contracts, law, banking systems, etc.
The same way you can't just open a bank account anonymously.
It seems inevitable to me that this is going to end up with a KYC/AML style digital online ID.
Without that you are not using Claude 10. It would not be shocking at all at some level of capability that running open source models anonymously is made a crime.
To pretend there is not huge issues with a laissez faire attitude as these models progress I think is intellectually dishonest and performative.
The US is slowly becoming more like China. From talks of nationalizing companies to make US state owned entities to banning foreign competition. It's just so strange how you become the thing you fear.
Hah, if only. Then at least it would see some of the benefits too. Instead, it's making sure to only copy all of the downsides, with zero of the upsides.
This is the kind of flippant remark that does the gravity of the situation a disservice. Nobody burns witches out of the blue one day, they do it because they figured out that burning witches makes everyone afraid to do anything that could be remotely construed as witchcraft, and they all start being good little church attendees.
So I hate to break it to you but you're not fighting powers-that-be who are just bumbling around and decided to randomly be like China one day. You're fighting powers-that-be who did the analysis and decided that being like China was the right way to go, for the good of everyone, and they backed it up with math and logic. You're going to need math and logic to convince them otherwise.
> You're fighting powers-that-be who did the analysis and decided that being like China was the right way to go
Not at all. If they decided to be like China, then that would include the benefits. They're nothing like China. Xi isn't siphoning hundreds of billions to his personal account through $XI coin. He isn't giving the reigns of the country to his tech CEO buddies. He isn't destroying scientific institutions and looting the country for personal gain.
I hate to break it to you but this is the kind of surfacelevel comparison that doesn't hold up when considered for more than 2 seconds.
I agree actually, we're trending towards China in some ways and not in others. I was just pointing out that the ways we're trending towards China are driven by people who did a bit more analysis than OP is giving them credit for. Since your real disagreement is with OP, I'll bow out of this one.
Bro, you have "grass is greener on the other side" syndrome. China is a totalitarian state not comparable to the US. Take this from a non-US citizen. There maybe some backsliding or ppl may think things are worsening in the US, but US citizens always have the power to make things right, coz they are free and they can bear arms.
The propaganda about ourselves we sold the world has been much more effective than the rights we've been sold ourselves. I would prefer material rights and competent governance to either "freedom" (whatever the hell that refers to) or the ability to buy guns we're barely allowed to use.
I never said that "as of June 2026, the grass is greener in China". What I did say is that the US has rapidly been adopting all of the downsides of China with zero of the upsides. That's an observation.
> There maybe some backsliding or ppl may think things are worsening in the US
If you're phrasing this as "maybe" and "may think", you're not here to have a serious discussion.
You just need to think outside the box: the CCP is big tech, not Congress. The CCP (big tech in the analogy) would never dream of giving control to congress.
This is why they can run their society with such competency, though. All we got are the stupidest billionaires the world can imagine and their servile lapdogs in congress
I think America is becoming more like Russia, with a dictator who eliminates any billionaire who disagrees with His Dictatorship. The American groups who want to go back to the "glory years" of the 1950s are equivalent to the Russian groups who want to go back to the "glory years" of the Soviet Union.
This is exactly where we are headed: a has-been that sold itself down a river to a bunch of kleptocrats who only are about themselves and their sycophants.
I was referring to the US government buying a 10% stake in Intel and the ideas being floated of further investment in Intel and other companies by the Trump admin.
What VPN provider is this? I could use it because Chinese users of my apps often complain about not being able to download things from my western hosted servers.
I lived in China for a bit years ago and one the biggest issues accessing western websites weren't restrictions against the site. Most of the times the culprit was using CDNs or services from Google or Cloudflare which were restricted totally or partially.
I was working on a site around that time, learned about that fixed it for the chinese user base, after that users from china went up considerably.
CVS (drugstore chain) and Walgreens were among the first major U.S. merchants to accept Alipay via QR code payments, allowing Chinese consumers (and anyone with Alipay) to pay at these stores.
CVS also accepts alipay as a payment method, so they must have some kind of direct partnership. I don't see why they'd need an intermediary, although maybe they take the cash and just credit alipay some balance etc.
Hmm I have a couple Alibaba Cloud servers located in CN Hangzhou region and I just pay with my U.S. credit card, no Alipay needed. alibabacloud.com is meant for international customers anyway. I’m not sure if aliyun.com accepts international CCs, maybe you’re using that.
I also have at times ordered deliverables on Alibaba (not aliexpress). Being able to send funds in yuan over Ali network solves almost all problems with that.
If Chinese LLMs are successfully making people in the west defend China, then I think we have all the evidence we need to explain why they are giving away their models.
The next step of course will be to get people using that ungodly cheap AI on Chinese servers. Which will also be defended because "I would never trust an American Lab".
It's funny how you can look at obscenely overvalued Western AI companies, with capex's that are unsustainable, and along comes a system which pokes a gaping hole in this model, and the response is "Ya, but what do those sneaky Chinese have up their sleeve here?!"
Perhaps the question should instead be "Why are these Western AI companies getting insane valuations on dubious ROI, and how can these Chinese models run on a fraction of the infrastructure?"
Not saying you're doing this specifically, but I'd be careful with thinking that "company" in China means the same as "company" in America (or in the West more generally).
True, "Industrial Policy" [0] is much bigger over in China, and has been for a long time.
Although that might become "was" given the recent shift in US politics, where the Republican party of big-government has been heavily taxing Americans with tariffs [1], nationalizing US companies [2][3], and commanding the manufacture of specific goods [4].
>"Ya, but what do those sneaky Chinese have up their sleeve here?!"
I feel like that is way over simplified. I do not trust the American government. I do not trust the Chinese government. As an American, I believe the Chinese government has a longer and broader history of stealing intellectual property and far less checks and balances than the American government. The current American administration will be gone soon and maybe we can get some sanity back at some point. Overall, I trust my data in America more than China and I think that's reasonable. I am not naive. I'm aware of all the problems with my country and am quite vocal about it. In fact, I think it would be naive to think that the Chinese government won't have complete access to everything that goes through a Chinese server.
From what perspective? The American colonies repeatedly and flagrantly ignored foreign property and intellectual rights, e.g. via laws to protect domestic but not foreign authors. Samuel Slater was called Slater the Traitor in Britain for a reason.
One thing is that many of these Chinese companies have had to do more with less due to technology export restrictions. Doing more with less is not something that the big US AI companies think about much. The Chinese models are apparently much more compute and energy efficient.
That doesn't make any sense. If the cost is training, the cost is compute, not data. Distilling another model means paying for that model, which isn't obviously less expensive than crawling the web and curating a dataset. Regardless of where your data comes from, the training process costs the same.
this is a justification for why one frontier lab would have extremely high spend rates. There is >1 frontier lab though.
More explicitly: if the Chinese can get near-leading performance models at a fraction of the cost by stealing from Anthropic, why doesn't OpenAI? Lord knows they could use the extra cash.
>It's funny how you can look at obscenely overvalued Western AI companies, with capex's that are unsustainable, and along comes a system which pokes a gaping hole in this model, and the response is "Ya, but what do those sneaky Chinese have up their sleeve here?!"
The only way this is funny is if you are completely oblivious to how China usually operates.
> If Chinese LLMs are successfully making people in the west defend China, then I think we have all the evidence we need to explain why they are giving away their models.
Right, and if "The West" wants people to defend them, they better get in on the free action too.
In fact, they have no choice - tokens are soon going to be a commodity, if they aren't already. Most everyone is going to be happy paying 1/20th of the cost for 80% of the value.
Oh, yeah, before I forget, hear the worlds smallest violin, playing for those token suppliers in "The West" whose repeatedly stated goal is to replace human knowledge workers...
I think the next step is for China to start selling Huawei (and other) GPUs around the world, because everyone appreciates the risk of giving either of the two superpowers their data. The US is no longer in a position to pressure other countries to bad imports from Chinese companies.
This is the beginning end of the hyperscaler era, not a shift from US hyperscalers to Chinese hyperscalers. Taking Nvidia's extremely lucrative market is the goal.
Software ecosystem aside, their hardware is inferior to NVDA/TSMC's and will remain so. And even if it wasn't, China just doesn't have the fab capacity to both meet domestic demand and export enough to hurt demand for NVDA.
They're behind, but they're catching up pretty quickly. I wouldn't bet against Chinese companies dominating this market in the medium to long term.
Nvidia could easily end up being the Tesla vs Huawei as BYD.
It’s not like this is a new tactic, china has been very successful at wiping out competition by undercutting them using state subsidies in many sectors of the economy.
As always, the US has a government removing red tape to foster innovation, while China has a regime, unfairly picking winners, to hurt and subvert the West.
this is not an accurate picture of Chinese industrial policy. In fact, you could argue they have the opposite problem. Their industrial policy encourages too many companies to enter a space, where the resulting competition kills off profit margins for whichever companies end up surviving until the end. This is exactly what we end up seeing with extremely cheap Chinese goods.
If china anointed one company per sector, they would have no reason to be so cost competitive globally. There would be no structural cause for Chinese products to outcompete the rest of the world. You can see some of this in the US, where these kinds of anointed companies exist (say e.g. Boeing/other defense contractors, at least post the 90's). They are (famously) not particularly cost competitive. This is also exactly what you'd expect economically.
Please read up on what's really going on inside China when it comes to industrial policy and innovation. Barry Naughton has a nice series of books and papers that might disperse some of your incorrect preconceptions.
How so, with the Snowden leaks we learned the extent of American digital espionage in Europe, the US government puts pressure on Europe to prevent taxation or regulation of American business and even European citizens have become the subject of mistreatment in American airports based on their digital profiles. We can enter China visa free.
Given that you're big on data and don't like emotions, what have the Chinese materially done to us Europeans we ought to care about?
They don't sell weapons to Russia, as Wang Yi said in Brussels, if they'd put their full weight behind Russia the war would be over, this is the middle position for them.
Which, I agree with you btw, I don't like but I can understand rationally. We still buy oil and gas from Russia too. And with 20% of the world's supply casually going offline after America's own 3 day special miltary operation the Chinese would be insane not to.
What one cannot understand rationally to come back to the Snowden era in which Denmark spied on us in Germany on behalf of the US, which is insane to begin with, is being threatened a decade later with annexation of their territory. China is on some fronts a rival, hence the tension around Russia, but the US is as destabilizing and unpredictable for the world as Russia itself now.
Americans truly seem to have no concept that they're in the middle of their own post-Soviet meltdown and look like a rogue state to the world now, which makes China which is no saint more and more attractive simply by being a source of order.
It's hard to call either side good. It can be argued that the side which had the highest powers and oligarchs implicated in Epstein files, and which has also threatened attacks on European nations, isn't the better option. Also the nation which actively funds the war in middle-east.
If only the US was a little bit more like China, nobody would ever know about the Epstein files.
You think China doesn't get its hands dirty because of some moral superiority? The country is utterly brutal towards its own citizens, what makes you think that the lack of warmongering is anything but inability?
Compel? I am confused, all data in China is held in datacentres which the state has full access to, that is the terms of their operation and why some big tech US companies didn't want to operate in China. They don't need to "compel" anyone, the CCP has people at every large company supervising employees, and they already have full access to your data.
I am always completely baffled by these comments that not only get basic facts wrong but appear unable to conceive of a situation where the everything is subordinate to the state.
There is no negotiation, there is no due process, you give access to everything before you start or you can't operate.
Isn't this essentially true in the US too though? The feds can show up at your data center with a National Security Letter and demand access at gunpoint. And you have to give it to them, because guns, and you can't ever tell anyone about it because that's what it says in the National Security Letter and also because guns.
Your reply already shows the difference. In US you have the default expectation of privacy, until the feds get their eyes on you. Meanwhile in China the default expectation is no privacy, exactly as OP argued.
This system in action is best demonstrated during the lock-down period of COVID, when any random dude who contracted the virus would immediately have their personal life for the previous week/month published nationwide to the hour, and those who have overlap will immediately get a 14 day lock-down at home.
I have not seen surveillance done with such ease and breath elsewhere. And local PD already have access to such info, and there are scandals where police sell such info for profit.
Is it so different in the US? These are just the surveillance we know about, and they're not openly telling us about these, they're actively fighting public knowledge of such programs:
You absolutely should remember the Chinese war with the first of those; if you have trouble, a good way to remind yourself is that it was not long after the US one.
Yeah, so getting 1 out of 10 he mentioned, even if it's their direct neighbor (where disputes happen for all countries), ain't bad! This absolutely means they're the same /s
If I had wanted to say “China and the US are the same”, I would have strung together a set of words that looked a lot more like “China and the US are the same” than the ones that I actually posted.
Tibet, Manchuria, it should be somewhat obvious that a nation that is as ethnically diverse as China was not a nation borne out of lots of different people deciding simultaneously that they would like to create a country together.
What is modern China has only existed for 100 years or so. When the country collapsed there were ethnic divisions that were erased after the country was unified.
The hallmark of successful ethnic cleansing is people claiming that there were never any wars, that things were always this way. The same is true of Kaliningrad, the most German city, centuries of history as a leading nation within Germany and the HRE, now a completely Russian city. It is only in the West that you see any narrative around division, in places like China or Russia history is erased (and how could it be any other way, the cornerstone of Chinese politics is one nation, one people...there is no political value in this narrative in Western countries).
Could you give a recent example? Because your example is as pertinent as the overthrowing of Guatemala's democratically elected President in 1954 by the CIA.
There's nothing wrong with the US buying greenland, which has been done for territories around the world?
The US has a long history of protecting individual freedoms, China does not. There's an irony that you're aware of the misgivings of the US because we have free speech protections. You're probably less aware of the misgivings of the EU because they regularly arrest citizens for speech, and no awareness of the issues with China because they'll just disappear journalists in the night.
I think you're kind of missing the important bit; there is no threat of invasion and there never has been. People just want reasons to be mad at Trump, and so they make things up
Since the late 60s there has been explicit federal legislation. I assume you meant this as a rhetorical question but there is a definite answer because the US is a transparent system. Law is passed, the courts enforce.
The fact that you are unable to answer this question about China where it is often unclear why certain people are being targeted should demonstrate how big the gap is. For example, when Xi's first corruption crackdown happened, it turned out that it was being orchestrated in part by someone in their 90s who had left front-line politics twenty years ago (and much of why that happened is unclear, we are only just learning about things that happened in Chinese politics multiple decades ago so it will be a while before we know...we do know that Xi was then able to make unilateral sweeping changes to his own role shortly after that broke every convention of the last 5 decades)...it is difficult to compare this to anything that happens in any other political system. In China, you find out someone has been executed for reasons that are obviously not explicit months after it has happened.
It is genuinely quite difficult to compare to anything else. 6 people meet in a room, they have almost all the power, and maybe you will read about what they might have said 4 decades later in a book...that will never be published in China.
The inability to conceive of a country where these things would happen and you would have no idea that it had ever happened...and, perhaps more importantly, wouldn't care that you didn't know.
I think too many people are conflating Chinese providers with Chinese models - you can easily have Chinese models safely (well, relatively safely, I guess) hosted on US or EU infrastructure.
Case in point, Microsoft just announced it is toying with the idea of using DeepSeek as a cheaper model tier in CoPilot. They are hosting the model themselves.
Exactly this. For some reason this is constantly being overlooked/confused. It is very possible to deploy these models inside your own VPC on the big cloud providers and have it be completely secure. I would argue that is even more secure than trusting a model provider’s native API as your traffic is not staying inside your own controlled cloud environment.
I was reading the threads about local AI closely yesterday. Some people seem happy with it.
If I had the cash, I'd spend 6-10k on a strix halo with 128 GB and run it local with no internet connection. I think the Framework desktop is sold out but there were others seem to be still available.
Not OP, but IMO the Chinese are waging economic warfare by "commodifying the competition". US VC's and equity are massively invested in AI being very very profitable and as soon as it's not that becomes garbage debt on a balance sheet, and a lot of it. When money is invested it's expected to make a return. When a few people make bad investments they lose their bag, when many many people all make the same bad investment, you get a nasty recession. When the US goes into recession China is more emboldened to pursue their agenda because policy makers are distracted by shoring up support at home and not as interested in expanding their power abroad.
Free market economics is constant warfare. It's called competition, and if you're really doing free market economics, you don't get to shut down the market every time you're bad at it.
That sounds like a problem for the idiots invested in this garbage and who cut foolish sweetheart debt deals with the hyper-scalers. That's mostly rich people, because they haven't yet found a way to offload their bags onto retail (though they're certainly trying).
I'm confused. China has been in a fairly bad recession the last ~2 years (their housing bubble popping). Why are we assuming that recessions can only happen in the US?
I'm also confused how China managed to convince so many US VC's to spend an insane amount on a technology with "no moat". China exporting cheap AI hurts American hyperscalars. But it doesn't induce the behavior in American hyperscalars that causes them to be vulnerable to cheap AI. It seems kind of patronizing to point the finger at China, rather than acknowledge the American (potential) misallocation of resources.
Not sure I follow your point, are you confusing espionage with IP theft? The victims are entirely different in each case, I am sure china would not restrict itself to corporate espionage.
Soft power primarily (see my original comment),and less data for western labs to train on, less money for western labs to have.
The state can use their AI as a tool to bolster their standing, which is working well. Chinese AI labs don't have to worry about making money or affording more compute. The state will (and does) give them whatever they need.
Because the US is a democracy and there is blowback for pissing off the public. So allocating hundred of billions of taxpayer money to AI would be enormously unpopular. We have private markets that can spend their money however they chose.
China just has the state with one leader than thats it.
Not who you are replying to, nor do I agree with their take (no sign of irony, complete lack of self awareness, and just blatantly xenophobic. not suprised by it either). but they did already address this:
> it's called soft power, look at endless glazing, it works
So people in countries adversarial to the US, or even demographics at risk domestically like illegal migrants have nothing to worry about dumping their info OpenAI or Anthropic servers? But the Chinese are going to spy on people using open weight models running on private or third party servers? Do you hear yourself? Do you have any grasp of theory of mind at all? Is your vision of Chinese "ethnonationalism" white people picking cotton? Is that what you are afraid of?
The untold capital and effort, the disciplined, complex long game operation, all to get 10 million people a second asking "are there timezones in space?" or "excel bar chart colors?" or "do you think I'm beautiful?"
It's hard to truly grasp the enormity of the evil, really.
fully homomorphic encryption is a theoretical thing that runs a billion times slower than running the code normally, to make that more clear, it takes a year to run a millisecond's computation.
These bastards already prevent me from buying a BYD car, and a xiaomi phone, and they are adamant about me not using a chinese AI model. I hope they do not succeed.
I am using a Xiaomi phone. The hardware is fine, the camera is very good compared to the average, the software absolutely sucks, bootloader unlocking is not possible and it comes full of spyware and adware that you can't disable.
> what is the obvious thing I'm supposed to be getting from your ellipses?
Well, I'm unsure if it's a correct form of expression in English, but in my native language ellipses can often serve to express speaker/writer feeling of uncertainty about something or some kind of sadness/apathy/similar towards the situation being discussed. It's not about some "obvious thing you should get", sorry for misunderstanding.
> or are you just referring to hosted providers of those AI models?
Yes I am. I was not talking about BYD or Xiaomi; I think I focused too much on AI because that's the main topic in discussion here. That was my mistake and I apologize for that.
The ellipses in American English is often used as an awkward pause, or to indicate trailing off, or sometimes to indicate a sarcasm or passive aggressiveness. The last one seems to be a more modern usage. I have seen older generations use it as an almost soft question mark to indicate uncertainty, which always confused me because I assumed thy were being passive aggressive out of nowhere.
Deepseek is awesome for home projects, I have used it quite a bit and the 10 bucks I loaded is yet to run out. Quality is close enough to Sonnet that with a bit of extra focus, I am not complaining.
Part of the security risk also is the number of different models. I’ve been tempted to try some other models, but how many do I want to give access to SSH or even my repo? Obviously there are ways to work with this, but it’s gonna run through some people‘s heads.
Because they siphon off data to US intelligence, and if you claim they don't, you couldn't possibly know because the CLOUD Act can mandate them to do so without telling you or allowing you to admit it. Of course, if you're in the US this doesn't matter but for the rest of the world it does.
The US government exists to defend capital interests. It's why we can't buy BYD cars. It's why we can't import any cars unless they're 25 years old. It's why a Tiktok sale was forced. It's why the US is seeking to block states from banning prediction markets. It's why the federal government is seeking to block states from blocking data center projects.
As soon as DeepSeek came out I realized what was going on: China was going to make sure that no US company was going to "own" AI. It is an issue of national security. It's why the US essentially blocks US tech companies to maintain sovereignty.
I'm reminded of the browser wars of the 1990s that led to the antitrust suit against Microsoft. Microsoft used the "commoditize your complement" strategy [1] against Netscape. The US has blocked the export of not only EUV lithography but high-end chips to China. China doesn't want to be dependent on US platforms or policy.
So China is going to make sure there are open source models available and the US government is going to try and stop them to protect US tech companies.
The reason why some Capital Interests want to blacklist DeepSeek in the US is so that you are forced to buy Claude/GPT/Gemini, which will feed revenue into an industry that requires revenue (or it’s a big problem).
The reason why some Capital Interests don’t want to ban DeepSeek is so companies that utilize AI have more options, and running your own DeepSeek cluster acts as an independent cost comparison for enterprise inference contracts.
The raising AI valuation is giving more weight to those that want to blacklist DeepSeek. The AI Safety narrative is strong. I see a path where any institution with enough compute might be watched in a similar was chem labs are observed by the DEA.
> keeping open weights models available will keep current industry profit distribution where majority is captured by nvidia and cloud providers
With the Chinese manufacturing capacity, how long do you think this will remain true for? You can't spin up a fab in a year, that's true, but you can in five years, at which point all the protectionism in the world isn't going to help.
DeepSeek et al has contributed tremendously, not only by sharing open weights models, but also genuinely improving the SOTA in inference and training, especially given their compute-constrained situation.
if China starts making cheap GPUs for the entire globe, I will say THANK YOU and will continue my work on AI.
The collaborative nature of open-source (free software) is the ultimate good that benefits entire humanity. I value that more than any other benefit of protecting any company.
How much I care about Nvidia or OpenAI market cap? absolutely zero
How much I care about the poor, but smart and creative people on Earth being able to contribute to the state of the art of AI development? 100% more.
> The reason why some Capital Interests want to blacklist DeepSeek in the US is so that you are forced to buy Claude/GPT/Gemini, which will feed revenue into an industry that requires revenue (or it’s a big problem).
It sounds right, but I see zero evidence of this. I think you underestimate how many people in the current administration are True Believers. This does not, to me, seem like anything to do with Big Capital, but rather "America over China".
and it would be great to have an independent auditor have access to all the training material and good search tools, so that take down requests can be made by copyright owners
Seems like interests of US government and US capital (monopolize and corner markets, jack up prices, extract economic rent in perpetuity) run strictly against interests of the broader US consumers and overall global population
I don't use TikTok, but a cursory search shows that there's a #tiananmensquare tag that has a few thousand videos, including many about the protests and Tank Man. So while I haven't seen a TikTok about it, someone has.
I don't (could be wrong) think other American's seeing content on American TikTok is relevant to China's censorship of what Chinese people see in China
Versus the next obvious trap after Iran that dear leader stumbles into?
The China fearmongering doesn't really work in this case, our own government wants the ability to lie to and manipulate us about everything continuously.
Those data centers that "states" want to block: the CCP was directly tied to funding the protests and movements to weaken the U.S. technologically. Kevin O'Leary exposed it, in collaboration with the White House.
Politically they believe AI belongs to humanity, which is why they are basically the only ones left publishing research in the open. That's probably part of their socialist nature.
But also a financial one. They believe that models are commodities, that you can swap one for the other and that the only thing that matters are the applications built upon them.
So they want to make sure that the world, and their own companies, are not limited in their business and application by a protected US commodity.
They will keep releasing in the open no matter what for quite some time.
It's quite impressive how the latest years I have found more and more to empathize with China than many of the western counterparts.
But it's increasingly clear that since the last decade protectionism and nationalism is taking the place of globalization, even though globalization has been a terrific success in lifting billions out of poverty and making the US thrive.
> It's quite impressive how the latest years I have found more and more to empathize with China than many of the western counterparts.
I'm not sure that this is because China is suddenly a great place or political system so much as a reflection of many western nations speed running to autocracy in the name of manliness.
Yeah, for me it's the latter. Until recent years, it was at least possible to defend the US as having some good principles, despite how imperfectly defended or promoted they may have been. That in turn could make it worth defending. Now, it's just blatantly turned into everything it always stood against.
china's advantage is manufacturing capacity. giving away the model is a loss leader for the hardware business. they have a big gap on chip manufacturing. the only way to close that is by developing more efficient software. an open ecosystem is the best way to accelerate innovation of software development. since they are behind the US on model capabilities they aren't really losing anything by making the models open weights and being open about the performance enhancements techniques. but the open weights models are not necessarily what are running on the platforms or what they have internally. deepseek released v4 on their platform about a month before the open weights release, which I would guess was done to expose it to adversarial testing, so they could fine tune the removal of capabilities from the open weights model. (but i may very well be wrong)
Also, the open-weight local models are proving that the commodity can be delivered for most applications at a far lower price than frontier is charging.
I think you’re assigning magnanimity to a competitor that is lagging behind and has every, state backed incentive to capture the market the only way they can. By making the models dirt cheap to access. If the roles were reversed you wouldn’t see open source versions of Chinese models. Much like you don’t see them open sourcing their blade battery design.
Well yeah humanity in the sense of everyone except maybe Uyghurs, Tibetans, Mongols, ... I mean open models are really cool, but I have a hard time believing China is doing it purely for the greater good. The commoditizing your complement story sounds much more plausible.
realisticaly, the united states has no right or moral authority when it comes to human rights or rule of law given the last few years. given that, the admonishment of china's human rights violations against uighurs run hollow and hypocritical.
just happy to have more models that I could eventually play with if hardware ever becomes cheaper.
>> Politically they believe AI belongs to humanity
It’s really BS. Comunist China Party is mostly interested in control above all. Forget humanity, human rights, what is good or what is bad or even financials. The most important thing to them is to keep the people on a short leash. Of course once they feel they have that under control they think about the “humanity” stuff as well but that’s just extra.
1. Nobody gives a damn about that. Frontier models are for work, not asking pointless stuff you can find on wikipedia.
2. The models themselves are not trained for anything, it's a prompt-level filter when you use chinese models as providers. I've just asked Kimi 2.6 and it answered without issues.
3. US models sugarcoat a lot topics about Israel. So don't be naive thinking there's no propaganda or sugarcoating in other providers. But again, nobody cares about this use case.
The reason we can't buy BYD cars is because if we allowed it without restrictions, it would utterly and completely destroy the United States auto industry. That's terrible public policy, and we should not allow it.
Before anyone starts talking about the free market, there is no free market here whatsoever. The fact that BYD's cost structure is what it is is the direct result of Chinese industrial policy.
Unilateral surrender in a core aspect of statecraft, which involves maintaining our industrial power and skilled labor force, is absolutely insane. I hope my government never gets convinced by market fundamentalist idiots to do such a thing, any more than it already has, to our great detriment.
The Chinese don't make these kinds of idiotic mistakes, which is how they have amassed the power, wealth, and influence that they have.
> there is no free market here whatsoever. The fact that BYD's cost structure is what it is is the direct result of Chinese industrial policy.
Aside from countless other ways before and after this, the US government handed over tens of billions of dollars in cash to GM and Chrysler in 2008 and 2009.
The US is primarily powered by oil and natural gas, has massive domestic capacity, and locked in supply all over the Americas. China has a completely different energy mix and they move from a position of competitive disadvantage on ICE cars to competitive advantage on electric. Rapid electrification for China is all win, but for the US and our partners in the oil business, it would mean stranding trillions in capital investments that still have decades to run, so it just isn't going to happen.
Why do people say this? You can get a Shanghai-made Tesla Model 3 with CATL batteries in the US yet somehow, if a Chinese car, made with Chinese components in Chinese factories were to enter the market, the would spell doom for the entire US auto industry.
You just explained it. You can get a Chinese made car in China.
That's kind of the point. They are smart enough to protect and support their domestic industries. We also have to do that, it's necessary for sovereignty.
This is absolutely true. Remember that automakers greatly contributed to war efforts in the past. It is an indispensable domestic industry, just as much as energy.
Then there is the issue that BYD cars are presumably connected to servers in China and most probably backdoored. They are too much of a security risk. I would absolutely not drive such a car, without permanently disabling the onboard cellular modem.
Perhaps the U.S should change its industrial policy as well so that it can be competitive on the global market? To me it’s been clear that the car manufacturers both in the US and Europe were just milking their customers every year with a facelift as a reason to sell the same old car. I am glad that a 3rd player is in the market to challenge the “heritage” tax.
Don’t worry about the free market. China will definitely agree to free market terms after it captures the market like the U.S did and Britain before it. Then enforce strict free market rules and strict IP rules.
> The reason we can't buy BYD cars is because if we allowed it without restrictions, it would utterly and completely destroy the United States auto industry. That's terrible public policy, and we should not allow it.
Yeah, that was the argument against Japanese car makers, too.
A shitty system needs destroying sometimes. Competition from Toyota/Honda was critical in making US auto makers up their game.
It is terrible public policy to fall decades behind making expensive shitty versions of what the rest of the world has.
It's not like I don't understand the argument on the other side of this. I've heard it my entire life. It's been dominant since the late 1970s and 1980s.
It's just that it's wrong.
We need a competent industrial policy and support for skilled labor and policies that encourage domestic production.
I'm not sure if you've noticed, but our country has become fucked, overwhelmed by financialization, scams, monopoly rents and extraction, and all of the wealth accumulating to a handful of people, while we've become less resilient and, at this point, almost certainly have lost our place as the most dominant economy and industrial power in the world.
Of course it is part of an industrial policy. It is, however, not nearly sufficient, and if it's the only thing we do, it will become increasingly untenable and eventually fail.
But it's an essential first step to prevent our audio industry from just being summarily destroyed. Other steps are also needed to encourage domestic manufacturing and homegrown successes.
Also, I'm not sure why this is even controversial. Why do you think there's BMW and Hyundai plants in the American South? Tariffs are already heavily employed by us and every other industrialized country.
Since 2018 ~400 chinese EV producers were allowed to stop operations. An industry that exists because of subsidies and preferential loans as well as state induced trade barriers is not sustainable.
From 2020 onwards, the Chinese government began phasing out its EV subsidy program. This policy shift came just as the country entered a price war, led by aggressive pricing from giants like BYD and Tesla. The combination of reduced state support and brutal price competition proved fatal for many underfunded startups.
IMO the problem is that we've been given the excuse of market fundamentalism for the past several decades on the way down, as most everyone lost their middle class jobs, wages stagnated, etc. Now we're supposed to accept some last ditch attempt at protectionism based on directly blocking choices for consumers, when the US manufacturers aren't even really competing? It just seems like open hypocrisy. At this point the reasonable protectionist policy would be based around subsidizing American industry so that they become competitive options, not merely trying to keep the better foreign options out.
Every single load of bullshit shuffled into our faces has been presented as a benefit to consumers.
Google gives away their search and Gmail for free, don't you know? So it can't possibly be a monopoly.
And so on. It's just propaganda. It's bullshit. That's not the way that you determine whether firms have excess market power, and this fraud (called "the consumer welfare standard") was the deliberate choice of right-wing policymakers who were bent on dismantling antitrust policies and succeeded.
This doesn't imply the opposite though - something that is a detriment to consumers won't necessarily be a good thing.
I feel like you're conflating two things here. The "consumer welfare standard" was indeed a terrible scam, neutralized antitrust enforcement, and basically put us in the current situation where many "choices" we have consist of two megacorps that both suck. Just like the (IIRC) Bush-era FCC declaration that "competition" between DSL and Cable was good enough, and rolled back the CLEC/ILEC dynamic.. We're in full agreement there!
But the original point I was making was about things that actually did benefit consumers. Lower prices and more competition (from foreign companies) DO benefit consumers. The problem is that they harm the part of our economy whereby those consumers get incomes. So they're a race to the bottom.
And my point there is that it's a bit hollow to be racing to the bottom for thirty years, smashing one blue collar industry after another, and then when we finally get to cars it's like oh no, time for some protectionist policy. It might be great for the people still in that industry, but hypocritical to everyone else who saw their industry destroyed but are now prevented from having less expensive cars.
As I said, with things this far gone, the only thing that makes sense to me is directly subsidizing purchases from domestic industry (positive incentive), rather than continuing to prevent competition (negative incentive). The latter reeks of the same monopolistic captive-market consolidation we've seen from the destruction of antitrust in general.
automotive platforms are a key military asset
it's not like the pokemon dildo industry, if you stop building jeeps your abolity to bully third parties is diminished
> You think people laughing is an important metric…
I think if you're gonna argue "preserving the auto industry is a national security issue" you have to address the fact that an auto industry that relies on protectionism to avoid being competitive with the rest of the world will probably not be very effective at national security.
Otherwise, you wind up like Russia in Ukraine - people laugh at your failed efforts.
> an integrated industrial facility capable of producing vehicles in large quantities
Large quantities of vehicles don't do much good if those vehicles are shitty compared to the opposition's. Iraq's army under Hussein was one of the largest on the planet at one time.
"The nine M1A1 tanks of Eagle Troop destroyed 28 Iraqi tanks, 16 personnel carriers and 30 trucks in 23 minutes with no American losses."
"In doing that the scout platoon encountered another Iraqi tank position of thirteen T-72s. The lightly armored Bradleys, each equipped only with a 25-mm cannon and two TOW missiles, are intended for reconnaissance, not direct engagement with armored tanks. Despite a misfire, and having to reload the launchers in the face of the enemy, the two Bradleys destroyed 5 tanks before help arrived."
What does the Russian economy have to do with anything? First off, they're run by a kleptocratic oligarchy, and second of all they're still one of the top 5 most militarily powerful countries in the world. I don't even know what point you're making? Did they bail out Chrysler? Which side of the analogy are they even on?
This is just an example of the fundamental nature of asymmetric insurgent warfare and the nature of proxy conflicts. It's not like Vietnam was more powerful than the US economy either. You seem confused.
In an all-out existential battle Ukraine would have been wiped off the map in the first 20 minutes by nuclear weapons. This isn't an actual contest of industrial might versus industrial might.
"In an all-out existential battle" involving nuclear weapons, the United States won't be affected by the presence or absence of domestic car factories either. World War II could soak up years of total warfare effort from the belligerents, and still have factories and governments intact to send more soldiers and bombs toward the enemy. I don't think that can happen now that countries as poor as North Korea can make nuclear weapons.
it is not anymore, because US doctrine has changed after losing war in Vietnam.
US can no longer sustain massive motorized and armored forces, because it implies heavy casualty rate.
The doctrine changed to shock&awe and lobbing standoff munitions from far away, which we all saw in Iran (and how it turned out).
US strictly protects boomers at Big Three and their regional dealerships and the entire supply chain that makes money off of financing, extended warranty, selling overpriced parts, overpriced heavy vehicles, etc
US note remotely capable of doing a China playbook which is: _OLD_ IP. In exchange for allocating cheap land, building cheap factories/infra, staffing with cheap technical labour etc etc... the IP sharer just sits back and collect checks. The Chinese playbook actually offers value US (and west in general) not capable of providing.
We're kind of doing it with the tsmc fabs, but yea, there are civilizational problems in the west which goes beyond cheap resources, talent, and labor.
it would destroy it, but then new more competent US automakers would pop up, similar to tesla.
US Big Three are simply full of incompetent boomers who want to maintain monopoly using tariffs, chicken tax, and banning of competitors that actively harm consumers.
Suddenly US government thinks that capitalism and free market is not desirable... huh
> The Chinese don't make these kinds of idiotic mistakes, which is how they have amassed the power, wealth, and influence that they have.
I generally agree with most of what you said but not this. China's chief advantage is having a billion people. On average, they aren't that wealthy or powerful. And their leadership makes plenty of idiotic mistakes - look at their real estate market.
That's not the chief advantage, insofar as there is a difference between China, India, and Indonesia, which there is.
Their chief advantage has been a coherent, long-running national industrial policy and trade policy that encourages industry while keeping the financial sector from taking over the economy and ripping everybody off.
We used to do that too from the late 1930's to the late 1970's, which is why we were the dominant industrial power in the world at that time as well.
> We used to do that too from the late 1930's to the late 1970's, which is why we were the dominant industrial power in the world at that time as well.
I think there's another world event that happened in that time span that might better explain America's world-wide industrial dominance.
> Europe was devastated and bankrupt. Asia was devastated and bankrupt.
Well yeah. Because America's world wide industrial dominance soundly beat the shit out of everyone, due to deployment of a highly successful industrial policy.
Imagine if we needed to rapidly step up industrial output tomorrow to fight another global war and China was on the other side. How do you think it would go?
> Because America's world wide industrial dominance soundly beat the shit out of everyone, due to deployment of a highly successful industrial policy.
That industrial dominance came largely during the war, and was made possible by the fact that they weren't being bombed while it scaled up.
There's a huge element of geopolitical luck involved in the rise of the US.
> Imagine if we needed to rapidly step up industrial output tomorrow to fight another global war and China was on the other side. How do you think it would go?
Horribly! I think they're much more prepared for such a thing.
Well then we agree, that their industrial policy is working a little better than ours. Which was the original point.
They don't let western businesses overwhelm their domestic industry at all. For us to let them do it to us would be unilateral disarmament and suicide.
No I don't understand your objection. My argument is that preserving domestic capacity is necessary for our survival, and that limiting imports is necessary but insufficient to achieve that goal.
Banning stock buybacks would be another helpful step. Can you imagine being at the helm of a major US automaker as the transition to electric is happening and thinking you have no better investment to make in your own company than literally taking the revenue you're earning and sending it to hedge funds and Wall Street?
You're wrong about protectionism though. It is an essential part of industrial policy and heavily employed by every industrial powerhouse country including Japan, China, Germany, and yes the US. China uses it extensively and it's a core pillar of why they are now the center of world industry.
The long running argument to the contrary is better understood as propaganda by the financial sector.
I wouldn't consider India. It's been plagued by protectionism and tariffs and won't achieve anything close to China any time soon. The only industry of value for its people which is software services is now crumbling with AI created in US and China.
Edit: probably your point too and I misread
We're in late-stage capitalism here. The pitchforks are already out and spreading across the globe. Unless the big companies get broken up, this nation will split into either a police state or socialist state.
are you sure it has nothing to do with the fact that those cars are very heavy, potentially fire-making (battery firmware) computers on wheels that can be rooted (or are already) at any time by their manufacturers in China?
Is the idea that we're just one OTA update from them turning into bombs? Considering the quality of software in the auto industry, I would be about as worried about any domestically assembled EV.
and if you see the parent starting to replace Volvo engineers with Chinese nationals you will witness sudden change of heart by US officials; until then it really is just a financial fiction
Apparently, whether or not the engineers are Chinese is the deciding factor. as long as Nicklas Backstrom is designing the EV, its all good. Or in other words, Chinese scary!
> Where does 'it' end? Is the subsidiary not operating at the behest of their parent?
corporations are financial fictions. the risk I am discussing is literally the mass destruction and murder of civilians during a time of war. individuals carrying it out will be subject to sanctions (including elimination), only those who are protected by their government will carry out such instructions.
and therefore national security focuses on firms which have engineers that are protected thus.
do you have evidence that if they wanted to do that they would not be giving orders to European or American engineers? as I said, financial fictions don't matter in matters such as these.
if you have evidence that they have Chinese nationals working as engineers without oversight, you should submit it to the government.
no. first you need to design the pathway by which control of the computers will allow overriding control over all the relevant control surfaces, something Tesla for example doesn't do. for that you need engineers collaborating all along.
I don't understand what you mean. I assumed the scenario was the manufacturer bricking the car, starting it on fire, or making it crash. Bricking is pretty obviously trivial. Starting it on fire would require changes to the battery, motor, or charger firmware, no big deal. Making it crash requires control over steering, motor, and brakes, which any car with modern driver assist features already has. What part of this would a Tesla, or any other car, not be able to do?
> What part of this would a Tesla, or any other car, not be able to do?
Tesla computers are not capable of maintaining steering control when the user is physically moving the steering wheel. does BYD have similar restrictions? can you trust them?
this discussion has reached the limit of its usefulness. if you aren't able to see why having nationals in an adversary nation be in control of dangerous hardware on your soil is a big problem, nothing will aid you in understanding this.
That's not the point of disagreement. I completely see why that's a big problem. Where we disagree is the notion that Volvo isn't controlled by an adversary nation or doesn't have dangerous hardware.
If you've managed to come away from this discussion with the impression that I think it's no big deal to have adversaries in control of dangerous hardware on your soil, then I do have to agree that this discussion has reached the limit of its usefulness, because I said nothing of the kind.
I'm sure they think of them as a matter of national security, because they think of everything as a matter of national security, but a few analysts I respect say that the mood there is not nearly as AGI-pilled, and I have no trouble believing that.
China is far more focused on robotics. Deepseek is largely bootstrapped by the hedge fund that developed it. They received a grant from the government of China, and recently an investment. Imagine thinking text autocomplete is a matter of national security.
China will flood the west with affordable robotics and watch the West eat itself alive. They know Western capital owners are so greedy they'll screw over their entire society to chase a buck and replace labor..
Of course its a matter of national security if there are military applications. The point of robotics is also weird because they've already widely adopted robotics within their own manufacturing and also America already replaced the majority of their labor by offshoring so I dont know how they would destroy american society by introducing robotics.
Show me, Chinese are not replacing their manufacturing with robots. You can't just say things off vibes alone because you think it sounds right. You're just making shit up off the top of your tongue. Show me where Chinese are replacing manufacturing labor with robots, please show me?
Peak of skill and capability, perhaps not, for now. But the ability to automate and discover relevant weaknesses at a greatly increased rate definitely counts as increasing the threat nation states can present.
I don't work for a nation state, but I work for a large financial that requires adversarial testing and it's absolutely increased the ability to operate and attack these environments. The floor is definitely easier to reach at a minimum. Vulns and 0days are far easier to exploit, time to exploitation from the release of vulns is nearing zero, largely aided by LLM reverse engineering -- why wouldn't this apply to nation state level adversaries?
In the 1990s, web browser SSL encryption was export restricted, classified as a munition because it involved cryptography. That was under Clinton.
For all the current admin's insane, ridiculous, corrupt, and criminal flaws, it's not clear to me how much of this particular issue is just the US government doing what it's always done, knee-jerk react to tech it doesn't understand by passing stupid laws.
Or that time that we parked an army on the Rio Grande because a bunch of people from the US decided to settle in the Mexican territory of Texas? That was a whole thing and the President of the US at the time, James K. Polk, ran on a platform of "Manifest Destiny" -- that the US should span "from sea to shining sea." There were a whole host of other countries with interest in that territory, not to mention the Indian tribes who would be displaced by that policy. The US has had a lot of dark periods in our history, and we shouldn't let those periods displace us from the moral certainty we derive from the Declaration of Independence and things like The Bill of Rights.
Yes, because we weren't an empire then. At the time Trump befouled our country, Pax Americana was a thing. We have voluntarily walked back from that position of moral and strategic leadership.
In that regard, history offers few precedents to learn from. Most countries have to be physically attacked to suffer the kind of damage that American voters are inflicting on themselves.
> Anthropic said it identified a campaign by DeepSeek and two other Chinese AI labs to illicitly extract capabilities from its Claude AI platform to improve their own models
Oh, won’t someone think of the poor mass copyright infringers.
Opus said to me once without any poking at it something like, "Help Grok understand it better". Makes me wonder if they are all cross-pollinated to an extent.
Prompt: Hello! Introduce yourself in one sentence.
Response: Hello! I'm *ChatGPT*, an AI assistant developed by OpenAI, dedicated to answering questions, providing information, and helping solve various problems. How can I help you?
"illicit" is throwing shade, but Anthropic can decide not to answer those questions if they don't want to. Plenty of companies don't sell to their competitors
This doesn't at all change the irony of big AI labs complaining about Chinese startups stealing the labs' IP, essentially by scraping the responses.
HN has a higher proportion of AI promoters than AI skeptics, and for a good while, the default response to complaints from book authors, bloggers, and other content creators was that "you put it on the internet so it's fair game", or "it's no different from a human learning from your works". So yeah, unless we're willing to revise these answers, I think the same "tough luck" reasoning should apply here.
For folks who are at Anthropic, OpenAI, xAI, or Google, and think it's fundamentally different, I would ask you to think long and hard about that answer.
Completely agreed. I would go further and say that it should be legal to scrape responses from LLMs to train new LLMs, and that forbidding that in your ToS should be considered an illegal contract. That’s simply the best way to avoid complete monopolization of the space, without requiring more drastic measures like antitrust down the line (which we seem to not manage well these days, given the number of monopolies). As long as you pay for your tokens like anyone else, "Big LLM" shouldn’t be allowed to control what you use the output for.
In many countries, if a contract contains illegal clauses (i.e. requirements that are inconsistent with the law or public policy), that contract is considered void and unenforceable. It depends on "how illegal" each statement is, whether only that part of the contract is considered void, or the contract as a whole.
One extreme example would be that if I wrote a contract requiring you to become my literal slave if you didn’t pay your subscription on time, you would in practice suffer no consequences from not paying, because the contract itself is illegal in most jurisdictions. A less extreme example would be that where I live (Northern Europe), I can sign a contract saying that I waive my right to sue the company I purchase a service from, and then sue them anyway because it’s my right by law and asking me to waive my rights is an illegal contract. Or that where I live, non-compete clauses in employment contracts are illegal unless you offer 100% salary throughout the non-compete timeframe, so I can sign an employment contract with unpaid non-compete clauses and just ignore that part as it’s illegal.
Why would my country blacklist DeepSeek? Perhaps crazy lunacy like: "Your product is too good and too inexpensive: consumers like US companies and individuals need to pay more for services."
How it’s price dumping if they give you the model to run free on your own hardware? Follow this logic, are not all OSS price dumping and should be blocked as well? I remember Steve Ballmer once called for that!
I’m pretty sure the digital lords like that proposal a lot. Not so much about the serfs themselves, though.
First of all you are assuming I'm saying they should block it. I'm not.
Seconddly, releasing the model weights for free and selling hosted inference are completely different markets. Open source itself is not price dumping obviously. But a hosted API can still be dumped if it is sold below cost to buy market share or squeeze competitors. Which it is.
They usually say it has "ties" to the Chinese military. Meanwhile, every big tech from Microsoft, Amazon, Google, Anthropic, and a ton of smaller corporations directly supply the US military.
So... anybody who was hoping for CXMT (or YMTC) to maybe cause RAM or flash prices to maybe drop, maybe just a bit, pretty please, can go pound sand? (YMTC of course is already on the Entity List.)
Not thus far, but I doubt one ever gets off this kind of chopping block once on it, so it seems to be merely a matter of time. It’ll probably take a year or two for CXMT to start producing consumer DDR5 in higher densities, if they even plan to bother, so I wouldn’t bet against a motivated lobbyist getting the axe to swing in the interim.
This is purely politics and the only way to get back the trillions of dollars, Americans has invested in AI.
US government will point to Chinese models and companies, saying it is a risk - just as the west did with Huawei - making it a none choice for brainless corporate people to choose what the choir preach.
Fable is getting more attention now precisely because it was taken down. Do the same to DeepSeek and Z.ai, and you will not strengthen your AI labs. You will likely achieve the opposite.
To give credit where credit is due, it is good that the Trump administration has not avidly played these stupid export control games. They tend to do little except hurt open collaboration; I remember when all open source cryptography had to be developed outside the US due to ITAR.
It is very possible that Trump and his cronies are just too incompetent to do that. In this one particular aspect (i.e. open source) I prefer having a stupid enemy than a "smart" enemy.
What an amazing achievement by America's adversaries.
The Trump administration lists Anthropic as a security risk and kneecaps its best model, despite the fact that compared to the other frontier US labs Anthropic is more transparent, more safety-oriented, frequently honest to a fault, and is clearly acting with patriotic intent.
Meanwhile, the same administration is hesitating to counter certain Chinese companies' efforts of industrial-scale theft and sabotage due to a fear of angering the CCP!
This administration has it exactly backwards. 4.5 months until election day, 7 months until the next Congress is sworn in.
I've always found this line of reasoning troubling and uninformed.
Chinese models first of all can be hosted on your own hardware, I'd argue they are way more transparent than US companies, by well releasing stuff.
Second, the "smoking gun" of DeepSeek training off Claude isn't as bad as you may think, and the amount of tokens was deemed trivial. Did you also know that if you asked Claude's it's name in Chinese it would respond as "DeepSeek" until just a few months ago until they patched it?
Third, I find it a little hypocritical to call out Chinese for "industrial-scale" theft when anyone could create Studio Ghibli style image gen photos. How could they do that unless US companies trained on copyrighted works.
Chinese are just innovating faster at this point, DeepSeek V4 is an actual technological advancement (KV Cache compression) more than a cheap clone.
The administration does have it backwards, but IMO it's more them playing into the big tech companies plans (of course they have their favorites) instead of actually investing in education, and research like the Chinese do.
I love Gemma, especially the latest Gemma 4 release - it's really great to see at least somewhat capable, not completely useless model that one can easily host on their own hardware without significant CAPEX investments first - but, to be honest, it doesn't quite compare to GLM 5.2, Kimi K2.6 or DeepSeek V4 Pro. Again, Gemma is amazing, especially for pet projects, but it's not something at least relatively near to "flagship" or "state-of-the-art" (except for small-to-medium size LLM category).
I wonder about your last point. Certainly there's an aspect of education that Asia values more, however by the US own metrics, they are number 1 in terms of education outcomes.
I'm a white dude from Iowa, working in top levels of AI/ML. I'm in the minority at work/conferences. I hardly ever even interview homegrown US job candidates. I'm just saying, that the reason I think you see more people from Asia and India is the education levels of most of the candidates. I'm not faulting these other countries, just pointing out how I see an educational gap based on demographics, and one that is rising up the ranks.
Is DeepSeek really behaving different than other Chinese companies? Intellectual theft is ongoing and has been ongoing for decades. Besides security risks and foul play, it is impressive by just how much DeepSeek undercuts OpenAI and Claude. DeepSeek charges $0.87 per million output tokens compared to $50 for Fable and $30 for GPT-5.5.
I believe OP is talking at the national wealth and technology level: China stole from the U.S. (again). So the U.S. moves to protect American companies.
Yes. For all the concerns about IP theft there can be on OpenAI or Claude, there's not even concern when it comes to Chinese companies since it's fully expected that it's a lost cause. Has been for decades.
And these plaintiffs are representative of only the best-organized and most well-funded of those who believe that these companies stole their data. Countless independent writers, artists, and other individuals whose data was ingested unknowingly and without consent lack the resources to litigate claims, but that doesn't change the fact that their copyright was violated in service of for-profit LLM/GenAI model training. It's not a trope, it's just what happened.
>The court drew a line, however, when it came to the pirated books, which were downloaded without payment and kept in Anthropic’s library irrespective of whether they were used to train its LLMs.
>We apply a basic prompt template to bypass the refusal training and show that OpenAI models are currently less prone to memorization elicitation than models from Meta, Mistral, and Anthropic. We find that as models increase in size, especially beyond 100 billion parameters, they demonstrate significantly greater capacity for memorization.
> They further rely on safety alignment strategies via RLHF, system prompts, and output filters to block verbatim regurgitation of copyrighted works, and have cited the efficacy of these measures in their legal defenses against copyright infringement claims. We show that finetuning bypasses these protections: by training models to expand plot summaries into full text, a task naturally suited for commercial writing assistants, we cause GPT-4o, Gemini-2.5-Pro, and DeepSeek-V3.1 to reproduce up to 85-90% of held-out copyrighted books, with single verbatim spans exceeding 460 words, using only semantic descriptions as prompts and no actual book text
Even if they're trained for refusal and rewording, the data is still there in the weights.
One blog post I have, which was basically the only source for a while, explaining how to boot Armbian in an obscure SBC only meant for Android, was repeated verbatim until they started they improving the rewording.
I also find it hypocritical when the copyright industry fails to put any effort into prosecuting these big techs for their so called infringements.
It's like the industry is a shadow of its former self. The way the copyright industry used to operate, one would think these big tech CEOs would wake up with SWAT pointing guns at them while their electronics are seized, and then they'd end up in court and get hit with something ridiculous like a quadrillion dollar fine.
It actually pisses me off that it's not happening. Not because I care about copyright, but because it's extremely disrespectful towards all the previous victims of the copyright industry.
Mine is anecdotal evidence at best: I co-authored a fairly obscure book about the application of category theory to an extremely niche subject. There's basically no mention of the stuff in the book anywhere on the internet, nor in any academic publication I'm aware of. If you want to have an idea about what's in the book you have to have access to it. I couldn't remember some details of it and being lazy and slightly curious I tried asking a couple of models (one by OpenAI and one by Google): they both managed to give me extremely detailed answers based on the contents of the book. Nobody has ever asked me or any other person involved in the publication for permission to use the book in any kind of training (they may have bought the book but not the rights to reproduce it).
The funny thing is what happened when I told one of the models (the Google one) I was one of the authors and that I had never given any consent to use the book for its training and that given that it was so willing to provide any user with the contents of the book nobody would have had any reason to buy the book. The thing told me that it had done it just because I was the author of the book (apparently me asking it about the content of an obscure academic book was sufficient to make it statistically plausible that I was one of the two people who had read the book, me and my co-author, excluding the editor a priori). It swore it would have never given that information to any other user.
I doubt that anyone could ever deny that LLMs are incredible tools that have incredible value. But denying that they have being made possible only thanks to egregious acts of piracy is disingenuous.
I don't think we should pay any AI lab for their """work""" until they start paying consenting data subjects for their data. Given this, China comes off less like a thief and more like Robin Hood.
Or, to put it in 2000s terms: Anthropic is the guy selling bootleg CD-Rs of MP3s they downloaded from Grokster[0]. Should we give a shit about their livelihood when people figure out about Gnutella? No. Knowledge is a commons, and Anthropic is one of the biggest threats to the knowledge commons in recent memory.
Their website literally has chinese characters on it even in english mode and everyone under the sun including crappy money talk show hosts know them as the chinese player that undercut western players. It's not exactly a secret.
You'd think anyone with two brain cells and confidential data could apply some judgement of their own...
They collect more spy data and do more with it, but none of it directly affects me since they don't extradite from my country or share with my country's police.
Not sure I'd go that far but I do use them almost exclusively for my coding on the basis that it is an acceptable trade-off. Far cheaper and my shitty apps are really not that valuable as training data
I use DeepSeek every day (via VSCode Insiders and Zed Editor). It's very affordable and, while it's slightly behind Claude (not sure how far behind Fable), it suits my working style well. I'm not using unsupervised multi-agent workflows and don't need a library of skills files - I'm writing most of the code and leaning on AI to help with mundane tasks - like;
- generating types for APIs
- generating boilerplate based on existing code
- improving existing code (adding error handling, timeouts, things like that)
- Writing SQL repository boilerplate / queries
- Creating implementations against hand written tests
- Helping me understand and implement APIs from third party libraries
- Writing documentation
I've spent like $2 in the last month and have used over 100 million tokens.
It's doubled my productivity and unlocked work that I could not have done before.
As an Australian, I'm not sure that I care about the safety of my data when it comes to LLMs. US companies already stole scores of data to train their models on and it's hard to imagine they suddenly grew some integrity. I'll care when regulators step in, until then it's out of my control so I'll just use the best price-to-productivity product available.
In terms of price-to-productivity, nothing will beat DeepSeek right now. For what you have described, all of the existing frontier models will perform well (probably about the same even).
If you expanded the list to very hard research tasks, Fable was so far ahead of the others that it doesn't even deserve debate. If you are a researcher doing something involving scientific computation or mathematics that wasn't rejected by the guard rails, and you were using Fable, that week was probably your most productive week ever. A couple of my PhD students effectively finished their current projects in that period by getting Fable to chew on it for 30 hours straight (not sure how I feel about that).
This was my experience too. I asked fable to implement a (quite complex, novel) CRDT engine. It did fantastic work for the 3 days I had access to it. The spec it wrote is exactly what I want. It used prototypes and examples to figure out some hard problems and answer a lot of complex design questions. Claude Opus has been lumbering along over the last week trying to turn what it did into a useful library. As far as I can tell, by trying to vibe together pieces of work fable did without understanding it properly. Opus makes mistakes constantly, misunderstands the spec, and it makes terrible engineering decisions. Earlier today it claimed it proved something was impossible. I asked it to think that through and it immediately backtracked and apologized. Was it right then, or is it right now? Claude has no clue.
I'm kicking myself for not using Fable more while I had it. Now that I've used fable, I'm not sure I even want opus any more. It might be more efficient in the medium term to just program everything myself until I have access to a similarly capable model.
I feel like Deepseek, opus, etc are only good at problems that have already been solved 100 times on github. They're like the iPhone 3G. Its exciting they exist at all. But subsequent versions make them seem like cheap junk.
I had a similar experience. For complex maths in my 3d engine around IK, made more progress in those 3 days than I would’ve done in weeks.
I recently discovered though that ChatGPT 5.5 Pro is almost as smart but only available to me via the ChatGPT app. I’ve been having it read my code and having Opus 4.8 use the ChatGPT app to collaborate. It’s a step down but its a temporary stopgap for complex problems.
It's been pretty cool (and a bit scary) to hear similar experiences from many other researchers and developers just how impressive Fable was for their particular workflow and the productivity it seemed to unlock. Personally, I have found that I can't let any of the other AI agents touch the code that Fable produced for me, since they consistently fail to understand the very delicate choices that it made to ensure optimal performance.
> they consistently fail to understand the very delicate choices that it made to ensure optimal performance.
Do you have an example?
I'm an engine programmer in video games on a big project - recently I'm mostly just fixing bugs. And Fable has been absolutely phenomenal in those 2 days we had it lol. It was genuinely the first time I didn't really have to reproduce a bug to fix it - Fable was able to really understand the code and interconnections between the systems to reason why it might have happened and fix it. Opus really struggles with that in my experience, or it munches code for 2 hours only to come up with a completely wrong explanation, like, wrong if you think about it for more than 2 seconds. TBF I only had 2 days to play with fable, but it was incredible in that time - can't wait to have it back.
Something I wonder about in this domain is if this is the fault of the language.
Game development is usually in C++ or C#. With C++, bugs are a nightmare to find. C# less so, but there are still memory leaks possible.
I've been writing Rust for a few years now and it has been absolutely phenomenal. I get the performance of C++ and the only thing I think about when developing is the logic itself.
There aren't any mature game engines that use Rust (Bevy is rapidly improving, but it's no Unreal/Unity) - though you can compile Rust to a dll so it can be consumed by Unreal/Unity.
Due to the language constraints of Rust, I have found that LLMs need to work _way_ less to figure out bugs and render out code. The compiler gives very specific error messages and if it compiles, it works.
So....just from my personal experience.
When we started getting access to Claude last year, it was barely competent at C++. Like someone who read a book about C++ but never actually wrote any, or had to debug any issue. It would confidently declare things which any C++ programmer could tell you were bollocks straight away.
But....now it's able to reason about things like memory management, thread safety and second order effects when multithreading very very very well. It's honestly incredible how well it can understand C++ code and figure things out. Opus is already pretty good(although far from perfect) but Fable was just something else. I could give it a description like "It was reported that only sometimes(1/20 repro) when the user opens the inventory and closes it, they can't move the camera afterwards" and it was able to find out both the inventory camera control and input code and pinpoint the exact race condition that could lead to this situation.
Do you need to push internally to get access to these LLMs for writing code, or is mgmt aggressive and supportive?
We have unlimited access to Claude and its use is very much encouraged across the entire organisation.
Makes me wonder what the value of a PhD is.
To train people to become researchers. If the definition of what a researcher is changes over time (due to AI or otherwise), then the nature of the PhD will also change over time.
I would say to push frontier of human knowledge forwards. Its not a high school anymore where you learn to learn. PhD is already a place where you should deliver value. (By "value" i don't mean commercial product)
>> Makes me wonder what the value of a PhD is.
> I would say to push frontier of human knowledge forwards.
Not sure if this is what you meant, but to be clear: pushing human knowledge forward is a practical prerequisite of a PhD (given it's pretty hard to convince people you're adequately trained to do such a thing without actually demonstrating it), but it's not the value of the PhD itself. The value is producing a researcher -- that is, someone who has the skills to continue accomplishing this in the future.
The difference is that if you happen to expand human knowledge by a stroke of dumb luck (or smart AI...) without actually having acquired skills to continue doing so in the future, then you're not really earning a PhD.
Depends on the field. There aren't really all that many research tasks that can be completed in a week. One of my favorite stories from the past few decades was the finding of accelerating expansion of space. It took three decades of scheduling telescope time, looking at exactly the right places in space at exactly the right time, to find enough supernovas to have data to even analyze. Not a whole lot of non-trivial science is just "read existing literature, think hard, and produce text."
But honestly, there's a lot that is (particularly for simulation like stuff which can be implemented in code). We should expect to see a bunch of extra output here, which might improve scientific productivity (even though the real bottlenecks are gonna be the academic publishers and their fixation on holding on to their copyrights).
Makes me wonder what the value of a human on a PhD course is
Mimo 2.5 pro is the best intelligence / dollar
In most cases it seems (at least to me and colleagues) to be turning out that picking best intelligence is a better option than picking better intelligence / dollar, assuming you can afford the cost. At least on interesting problems. If you’re doing generic web dev work, probably not the case.
What are interesting problems? Deepseek and some other open source models do decently well in the gpu kernel benches. https://kernelbench.com/hard Most people think they’re working on “very hard” problems and they really aren’t
Absolutely agree with this. As Louis Rossmann recently pointed out, it's the difference between a correct answer and a wrong answer; the correct answer is worth a good amount, while the wrong answer is worth nothing. Under this metric, for harder tasks, the most intelligent model is best per dollar.
That's why experience + cost effective model is IMO the best combination.
Experience allows you to design the skeleton where the implementation details are often inconsequential. There are relatively few scenarios where an LLM would need more guidance to render an outcome, but even dumb local models can do that.
Building a simple UI component vs an efficient multi-threaded bidirectional socket implementation (both examples of things I did recently with DeepSeek flash).
Angular can only be written in one way so the UI was trivial. I know the architecture for the socket implementation and the trade offs for various approaches, so I sketch out the implementation and get DeepSeek to complete it (error handling, keepalive messages, timeouts, etc).
I don't think I would have saved much time if I just asked Fable to "make the socket implementation" but even if it got it right the first time, it would have only saved me a few minutes given that's how long it took to write any way.
Even in more conventional applications, Node.js / React CRUD applications - "write a graphql query for blah" "add an endpoint to run query" "add validation to endpoint" are all trivial for DeepSeek flash. In most of these cases, I have found you're constrained by context window size because these are rarely well architected applications.
> Experience allows you to design the skeleton where the implementation details are often inconsequential.
Expanding on this thought a little more: it is possible to set up scaffolding that make incorrect implementations inconsequential. If the LLM can detect when the implementation is wrong, it can retry with the errors feeding back into the loop. This is shored up by up-front investments in tests, API definitions, strong types, linting rules, etc. the various cheap, fast Flash models do not need to 1-shot solutions if capable of autonomous reiteration.
This is a great point. In the past this is how I learned to architect software in general as you are never the only contributor and often must accommodate the contributions of contributors who are early in the careers. Good software architecture reduces the damage possible by low quality contributions.
I have had great success with contract driven development workflows in this regard.
LLMs being effectively very knowledgeable jr engineers, slot right into those patterns.
Another interesting thing is that Rust _really_ shines with LLMs. The compiler gives very helpful messages which the LLM bounces off very effectively. If it compiles, it works, you just need to double check the logic is correct.
By contrast, with something like Go, you'd need to be on the lookout for nil pointers, race conditions and so on. Node.js / TypeScript is less dangerous, but you still have memory leaks that aren't possible in Rust and lack pattern matching so must be vigilant about type discohesion.
This is mostly true and a month ago I would have agreed wholeheartedly. However, there are still situations where almost every line of code has to be deliberately chosen in a delicate way. Think research code or settings which demand a high degree of optimization. Opus or GPT-5.5 are absolutely miserable to guide in these settings.
"Well, don't use an AI for those purposes!"
That used to be true (and is now again true I suppose). Fable was a peek at something different, it seemed to be actually able to start to tackle these kinds of problems. That saves a lot of time, since checking these programs is often far, far easier than writing them. Experience is needed to check and scaffold still, but something like Fable becomes a prerequisite for these settings.
People are wasting tokens with perfectly fine models by not planning effectively. Anyone who says Opus is not capable of doing what Fable can probably are not planning effectively and just praying to the one shot gods.
> At least on interesting problems.
What fraction of your work is "interesting problems", and what field do you work in?
> If you’re doing generic web dev work, probably not the case
I have a feeling the bulk of most people's work is "generic $X work." My take is people should figure out their mix of interesting vs boring work, and optimize accordingly. Flash models also tend to much faster,
Yep. After moving Deepseek for a long time I also moved to Mimo 2.5 pro. It's not only similarly cheap, but it's very good with 1m context window and gigantic tasks. Produces very lean code.
Yes. Whenever I tried, I found no reason to use DeepSeek v4 Pro versus MiMo v2.5 Pro, price-wise or performance-wise. Benchmarks also confirm this.
It is since they adopted DeepSeek’s pricing, although it’s not as good about caching as DeepSeek.
UltraSpeed will change how you think about agentic coding.
Sadly also one of the slowest of the recently released large models
I have the feeling MiMo v2.5 Pro uses a lot less tokens and so a lot cheaper ( comparable pricepoint)
IMO US vs Chinese data vacuuming is a false choice. One of the main benefits of these open weights models is that you can get the privacy and cost savings by hosting your own model in cloud infrastructure.
Open weights models are only 4.5 months behind closed weight ones.
The fact that US considers propping its flagship technology by blacklisting competitors demonstrates how small the US competitive advantage really is.
Hosting DeepSeek Pro yourself is gonna be wildly expensive though?
You have a wide choice of providers available, so if you can find one you trust you can get inference without data harvesting and it's still very cheap. But dedicated HW is insanely inefficient.
Opus estimates you can do it for $13k/month if you get committed pricing on the HW.
I guess this depends on the scale you are planning. For big multinational cooperations that calculation could look different.
The USA has an entire economic system to financialize your personal data and obfuscate your privacy. China can just sniff on their citizens API Calls, and use that to distill models.
>obfuscate your privacy.
Edward Snowden says hello.
[dead]
They could just be flagging that they're considering blacklisting as a way to get some more Trump coin sold or other bribes. Or indeed to get further money from USA companies - which way will Trump decide, only the numbers in his family's bank accounts can give us a clue.
Using laws to ban competitors is just economic warfare thats all. Its got nothing to do with "national security", thats just the reason they give us normies. You should be safe in Australia since they actually need to be friends with China.
> You should be safe in Australia since they actually need to be friends with China.
As a non-Australian, I enjoyed very much reading about Australia in a book by Tim Marshall titled "The Power of Geography". I didn't quite realize just how vulnerable they become with China's ambitions and expanse in the South Pacific due to their reliance on vital sea lanes for trade with its Asian partners. After reading that book, I can appreciate your comment much more.
Like some wag said recently, Australia needs nuclear submarines to protect its vital trade with China against China.
> Using laws to ban competitors is just economic warfare thats all. Its got nothing to do with "national security"
Economic safety is 1 angle of national security. They're not "wrong".
Economic safety can be guaranteed in many ways. It's not economic safety that these moves are for. It's for blocking anyone from preventing them from maximizing the value the Western oligarchs extract from both their own citizens and the entire world.
What? Every country has the right to freely decide which companies are permitted to operate under its jurisdiction, and exercising that right isn’t “warfare” of any kind.
Poeple talking about 'trade war' must drive you crazy.
What do you think warfare is?
I think the country that talks the most how the market regulates itself and needs no state intervention isn't the most qualified to put tariffs and block other countries from trade. And yet here we are…
How are you accessing their API? Through OpenRouter, or direct? Are you using DeepSeek v4 Pro? $2 seems a lot cheaper than my own experience accessing them through OpenRouter for over 100 million tokens, but I am using OpenRouter to access v4 pro.
Cache hit rate dominates your total cost calculation for long agent session, and it largely depends on the provider. Deepseek's native deployment is probably much better than third party in this regard. For v4 pro it's a whopping >100x price difference between normal input vs. cached input tokens.
I am using Flash and accessing the API directly via vscode insiders and occasionally Zed (it's buggy but I keep coming back to it because I want it to succeed).
Unless you need enterprise multi-model management, I don't see the point in OpenRouter as it just adds cost overhead and you can just self-host an open router alternative (LiteLLM, Bifrost, etc). Running an LLM gateway locally is kind of nice as it allows you to normalize your configurations against your internal gateway - but I haven't really needed to.
You can BYOK with openrouter and pay nothing up until 1mil requests/month iirc which is pretty generous for one person
Deepseek direct is atleast 2x cheaper than other providers serving it. Maybe their caching strategy is just significantly better or they’re subsidizing api pricing. I think it’s the former. $2 is closer to 50 million v4pro tokens in my experience
Ensure you're hitting the DeepSeek provider via OpenRouter - they have the massive cached tokens discount. If you're hitting any other provider you're paying an order of magnitude more.
Pro is substantially more expensive than flash. In addition, there's wide variance in price with DeepSeek themselves providing the cheapest tokens last I checked (but they train on them). Caching policy also varies by provider. TTL can be as low as 5 minutes or as high as 24 hours and reading from the cache might or might not reset the timer. Whether or not you get a hit makes (IIRC) a 10x (edit: it's actually 50x) price difference in the case of DeepSeek themselves.
I second this. I’ve been using it a lot (with OpenCode) for personal projects. It’s intelligent enough at a tiny fraction of the cost of Claude or Codex.
The fact that America has yet to figure out the importance of soft power (and TRUST in particular) in the AI/information age is mind boggling to me.
The CLOUD act, FISA rulings, the Snowden leaks, and now the aggressive tech oligarch push to weaponize the unholy combination of AI, MAGA, and social media algorithms in an absurd and patently obvious attempt to impose (or maintain?) "world domination" seems likely to cement their downfall.
"Authoritarian" China's low-cost and open source approach looks downright democratic by comparison.
It seems increasingly clear to me that AI is forcing a reckoning in terms of how we interpret authority and control in light of new ways that information is evolving. The old political labels seem grossly insufficient to describe the present reality and the ones whining about "democracy", "freedom", and "liberal values" increasingly sound (and act) like bitter, out of touch old men desperately clinging onto a world that is rapidly outgrowing them.
Last time I checked, the closest approximation of liberal values are "liberty, equality, fraternity". They seem to think that "liberty" should only apply to them, "equality" is a threat to their consolidated power, and "fraternity" is something that should be weaponized to turn communities against one another to distract from the ways the oligarchs have been and continue to abuse and consolidate their power.
I can second this. Deepseek is great for mundane work day tasks.
I use it via claude code, just pointing the api to deepseek.
It's also not a clear "Opus 4.8 >> DS 4 Pro", I've done 16 tasks in 4 days across the two, and while Opus was indeed on average better, both models performed well being able to handle most of my workload.
In fact DeepSeek was _significantly_ better on 3 task out of 16 and Opus was _significantly_ better only in 2 out of 16.
So why I still claim Opus 4.8 to be the winner? Because the few times that DS failed or got off the rails, it failed much harder and needed several prompts to be realigned on the actual tasks.
Another thing at which Deepseek is significantly behind is code reviewing. Opus is more intelligent/thorough, Deepseek will sometimes generate bogus or low quality feedback.
And the last thing at which Opus is better, period, is vibe coding. If you want to implement features end-to-end it handles ultracode flows quite better. I don't vibe code at work, but I do so with personal projects.
But the cost concern is real. I've spent sub 2$ in 5 days of work using DS 4 Pro, which is on average just 4 queries to Anthropic.
Give me a slightly better DS 4 (it is still in preview and training isn't finished) and I may ditch Anthropic for good.
Have you tried glm 5.2? It’s better than ds4pro and very close to opus4.6. Alot more expensive than ds4pro/mimo but significantly chesper than opus 4.6 for 90% of the quality
I haven't yet, no.
I alternate between Opus 4.8 and DS 4. If I remove my Anthropic subscription I will get GLM 5.2 in place of it.
Also, isn't Australia in a more contentious situation with China? Them being more allied with the US and all? Not to mention the whole nuclear sub issue. Having data stolen is shitty either way but isn't data taken by an adversarial country a worse situation?
If I spoke freely about how I feel about the US right now and the direct personal impact of policies implemented by the recent administration, I'd be denied entry into the country to visit friends later this year.
It is inconsequential if the US or China have my data, both will misuse it and I am powerless to protect myself from that fact.
Not using LLMs presents a bigger threat to my career than protecting my data.
Correct - As an Australian I feel free to say anything. But as an Australian who might like to travel again to the USA at some point in the future, I do not feel I have freedom to share openly online.
Whereas, China does not ask for my social network logins, or for me to be pre-approved to travel there. So that is unlikely to be affected.
But the main point is really that 2 > 1. 1 country scraping your data is bad. 2 is worse.
> you're probably just as pissed as 60% of Americans
Yet you voted for it
To be fair, most of their votes don't actually count.
And even with our westminster system, One nation has swept the right wing. Considering how many seats they will likely get in the next election, my high horse has shrunk a fair bit.
Me? I didn't vote for the orange Nazi
Because you think by voting for the other person AI companies and USA government wouldn't misuse all the data they hoard?
Why do you think I like them or support those actions?
Do you support everything the Klingon Empire does? That doesn't seem like the Worf I know. The Worf I know can distinguish the people of a civilization from the leaders of their societies. To conflate them just creates the problems you are criticizing
The klingon empire isn't a democracy, we don't delude ourselves that simply swapping one single person would have made a difference.
The same fact is true in a democracy. A democracy specifically is set up so just a single person doesn't rule. That's called a dictatorship.
But you turn a democracy into a dictatorship through tribalism. So forgive me if I get mad at you trying to place me into a nice well defined box. You're helping erode the democracy I (and so many others) desperately want to maintain (and improve).
What would Worf do?
I'm certain he wouldn't have done what you did
Sorry that you're being blamed. People have a pretty negative sentiment towards Americans in general right now because we have been watching helplessly in horror as our best friend tears things down.
I have friends in blue states who didn't vote because it wouldn't have changed the outcome given the electoral college - it's unfair to blame them for not participating as they were never able to in the first place.
Not that democrats are angels. You guys are damned if you're blue / (extra) damned if you're red - given the state of both options.
I appreciate that you're doing what you can
Consider that USA's government logic is: "if we don't like a government we will kill civilians until they rebel and put a government we like". See Cuba for example.
So, by their own metric, they aren't helpless at all and should appreciate other countries aren't as savage as they are with civilians, instead of complaining.
Look Worf, I'm just one human. You don't know me. You don't know what I'm doing or not doing. But I know you're making our situation worse and I know you're not living up to the ethics of your namesake
As an Australian.. politically I need to worry about business data touching China. It will come up at a Risk Advisory Committee meeting as a serious issue.
In actual personal practice, no. China having my data presents no actual impact to me, America will do things that impact me.
Exactly. And I feel the same way as a US citizen.
China is mostly interested in geopolitical stuff and getting an economic advantage, plus they have no jurisdiction in the US. Your data in the hands of the US government however could potentially land you in prison.
Most Aussies aren't really worried about China... China remains our largest trading partner, and polling shows less than half of Australians think the AUKUS alliance (which includes the nuclear subs) makes our region safer.
Latinamerican here. When you talk about "adversarial country" I think of the USA (they can kidnap a president, kill people on boats without a trial, etc) and not China. YMMV for different regions.
Yes, that's why they specifically were talking to an Australian person about their experience in Australia.
China is not digging through my social media in order to find a reason to cut my grant funding. China isnt going to pull me over for speeding in montana only to examine my phone to see whether i am maga enough to get off with a warning. And china isnt at o'hare security scanning for anyone with skin darker than freshly fallen snow. China may be evil, but it is a far away evil that doesnt have a physical impact on my day to day. The other evil is much closer to home. Even if it is not the biggest, the crocodile closest to the canoe is always more concerning than the one still on the bank.
… the one closet to your canoe is loudly warning you of the dangers of the one still on the bank
AKA don't worry about an imminent death threat! You might get seriously ill decades from now! Worry about that instead!
Re Australia vs China
https://youtu.be/sgspkxfkS4k?si=JgnhenF0qeTZXeGS basically explains the situation.
While having data/code stolen isn't ideal, there is a certain point where you need to assume it's already out there. There's actually more probability of harm from shady US companies imo, because people are less suspicious about data sovereignty
When you paste youtube (or instagram) links, it's good practice to remove the doxxing share link identifiers (?si and everything past it)
Historically speaking, the US might even be a larger risk to Australia than China is. The US alliance goes back a long way, and so does the opposition to US influence. Since the Whitlam government, MPs are generally fearful of US retaliation, rendering much of our politics hostage to US influence. Of course, some PMs have openly embraced the US so this feeling isn't universal. But many of our issues are directly tied to the US.
China has been far more beneficial to Australia by comparison, with the downside being the encroaching influence of CCP propaganda. Many of our strengths are tied to our relationships with Southeast Asia.
Paul Keating has famously declared US as an "aggressive ally", "our colonial masters", AUKUS as our "worst international decision", and that "our future is in Southeast Asia". This was under Biden too.
So the situation is much more complicated, and the feelings on the ground right now is that the US are not our friends (of course, the CCP is not great either).
You don't even need to step foot on US soil for the US to be a risk to you
c.f. Kim Dotcom
The same Kim Dotcom who was arrested in Germany, extradited from Thailand to Germany, and most recently has been parroting Russian propaganda points about Ukraine? The same one whom the New Zealand legal system has allowed to exhaust every legal avenue and remains free to this day?
There are much better examples to use to showcase the US's extrajudicial/international reach than this guy.
>and most recently has been parroting Russian propaganda points about Ukraine
Yeah but is that a jailable offense though? What about his freedom of speech?
Or should we just lock up everyone who says things we don't like?
Not jailable. But I'm absolutely free to disregard moral agency of mr. Kim Dotcom and stop paying attention to whatever he says or what happens to him as a result of his support for warring Russia.
Would you be OK if everyone else disregarded the laws to see you punished for things not illegal?
No one said for the laws to be disregarded!? Why are you lying about what people are saying? Does it feel good to win against your own imagination?
Did I say it was a jailable offense? You might want to actually read what people write instead of arguing with your imagination.
Laws are ultimately defined by the boundary and edge cases.
The godly person, doing no wrong, that through unfortunate circumstances, is brought before the court does not advance our understanding of a law.
It's the messiest case, by unscrupulous persons, in unconscionable circumstances that ultimately decide if the execution of that law gives the public more or less rights than on initial interpretation.
Yes, but public minds are not made up by edge cases. Surely a better example could be used rather than a divisive criminal.
Also, not sure how anyone except bad actors or sloppy readers could have gotten that I was defending the US here instead of criticizing using Kim Dotcom as an example in characterizing the US's actions.
> Also, isn't Australia in a more contentious situation with China?
China does not run 'rendition flights' to kidnap other countries' citizens and hold them in gitmo style prisons for decades. China cannot compel and does not compel random countries to extradite foreign citizens to China to persecute them and lock them up. Guess who does.
USA is an adversarial country to everyone in 2026.
even itself!
Given what happened January 6th the unsettling truth is the USA is essentially an occupied nation.
[dead]
> isn't your data being in more hands worse?
> Having data stolen is shitty
Fun fact: Deepseek can be hosted by third parties even yourself.
I've been using deepseek for some development at home and it is really good for the price. It is at the point where i am ok with using it as a tool that i can rely on and not an expensive gadget with flaky uptimes.
I'm in favour of sending as much data as possible to DeepSeek for them to train on. I'd happily zip up everything useful I have and send it to them if they asked for it.
> US companies already stole scores of data to train their models on and it's hard to imagine they suddenly grew some integrity.
Can you be specific about what data of yours were stolen by who?
Any text, code, image, video or audio they/anybody has ever uploaded to the public internet? Is this a real question?
How was it stolen, exactly?
I'm not interested in explaining to you something immediately evident that you clearly disagree with. If you don't feel that trillion dollar companies training commercial products on the free labor of human writers/artists/engineers/etc. without their consent or compensation isn't "stealing" you're welcome to that opinion, but I believe that's what the GP was implying.
Developers claiming their valuable data was “stolen” to train models is so dramatic.
I'd be fine if it cut both ways. It shouldn't be legal for works distributed by people under licenses that would prohibit derivative work whilst if I reverse engineered a Nintendo Switch 2 with an LLM I'd be sent to the underworld.
Companies that claim downloading their shit is theft are so dramatic. Although the consequences are different. Your "argument" is immature.
Noteworthy that Z.ai, maker of the just released near-frontier GLM 5.2, has already been on the Entity List since Jan 2025[1]. Being on the Entity List does not mean all trade is forbidden. Broadly speaking it means American companies and individuals are not allowed sell them goods and services, but they are still allowed to buy from them and pay them.
AFAIK the Chinese AI companies barely depend on US goods and services, except for nVidia GPUs which were export restricted anyway, so it doesn't seem to be very consequential (see Z.ai). For the RAM maker CXMT it could be a lot more problematic though.
[1] https://en.wikipedia.org/wiki/Z.ai
How does that even work? If Z.ai wants to buy lets say GPUs for AI training, what's stopping them from going to a local reseller? Its not even circumventing the rules, its the natural thing to do.
For that matter, does (only) NVIDIA make datacenter cards? When I buy a gaming card, I dont buy from NVIDIA, I buy from an integrator, like Gigabyte, who work with a company like Foxconn to make the cards.
The GamersNexus documentary (https://youtu.be/1H3xQaf7BFI) on the semi-underground GPU trade in China, while a little amateurish in terms of depth and general atmosphere, is an interesting watch and may answer some of your questions.
Basically, those export controls make GPUs more expensive for affected parties in China, but don't effectively stop them from being acquired or used over there.
The export restrictions only apply to certain GPU models, which are the more recent powerful ones used for training tasks. So the H100, B100 etc. are banned along with 4090, 5090.
Nvidia has downgraded chips that aren’t banned. H100 is banned, H800 is allowed. A100 is banned, A800 is allowed. But the sale has a tariff attached.
That’s all how it’s supposed to work. In practice companies probably circumvent the restrictions.
> That’s all how it’s supposed to work. In practice companies probably circumvent the restrictions.
Ex-colleague of mine told me he used to work for this company in UAE (he told me this story 15+ years ago, so he worked there even before that). He said it took him months of working there before he discovered that their entire business was evading US sanctions against Iran-they’d order servers/etc from the US, tell the US vendor they were for use locally in UAE, then ship them straight across the Gulf. The UAE government presumably knew this was happening but chose to turn a blind eye; the US government likely did too, but struggled to tell which orders/purchasers were legitimate and which were sanctions-evaders, plus likely was worried about enforcement action causing issues in the US-UAE diplomatic relationship.
I’m sure there are similar businesses out there who specialise in evading US sanctions on China.
Can I tell you I can go downstairs now and buy a 5090 (not d or d2 version, the real 5090) in China for 4000usd (33000rmb)? They have it stock right now, they have since release. It’s a Japan version.
I didn’t ask about h100 or others but can’t see why they wouldn’t have.
I believe it. I don’t think the restrictions are effective.
That said, I think the goal is more to make it harder to buy thousands of GPUs and stand up a cluster like big US labs do.
> How does that even work? If Z.ai wants to buy let's say GPUs for AI training, what's stopping them from going to a local reseller?
In theory, the whole chain has to comply. And if you don't you get fined etc. So the local reseller would risk not be able to resell.
Have you been in shenzhen? You can just buy in many stores a real 5090 imported from Japan
Good context. And it shows the limit of this kind of control: the model at the top of the open-weights index this week comes from a lab already on that list, and that changes nothing about whether I can run it. Once weights are on Hugging Face the download doesn't care about an Entity List. Chip export controls bite. Weight controls mostly can't.
> except for nVidia GPUs which were export restricted anyway,
those export restrictions are a joke. when they were introduced, there was a sudden spike in NVIDIA GPU exports to surrounding Asian countries. and the US government knows this
CXMT memory maker will not be banned, because US AI labs are salivating at the idea of more RAM supply, and are lobbying hard to prevent restrictions
And those who enabled it (the leadership team at SuperMicro) were charged by the US [0] and Taiwan [1]
[0] - https://www.reuters.com/world/us-charges-three-people-with-c...
[1] - https://www.reuters.com/legal/government/taiwan-investigates...
> And those who enabled it
There is no single "it", the Singapore loophole is well known, and nothing has been done about it. Except by China, once , when it blocked the sale of Manus to Meta, even though on paper, Manus is a Singaporean enterprise.
What's the publicly stated/marketing reason for capitalist America to put companies on the Entity List? Genuinely asking. Because to me it screams 'we were only for the free market until there was no competition'
"First published in 1997 to inform the public on entities involved in disseminating weapons of mass destruction, the list has since expanded to include entities that engaged in "activities sanctioned by the State Department and activities contrary to U.S. national security and/or foreign policy interests"
So RAM chip makers when there's a RAM shortage must be 'contrary to U.S. national security and/or foreign policy interests' i.e. the US government is trying to squeeze its citizens on RAM prices.
Nice.
Other way around according to what GGP quoted, this would get RAM into the US but not out, reducing prices.
> Other way around according to what GGP quoted, this would get RAM into the US but not out, reducing prices.
Only in a world where the other party has no agency. In real life the other party raises prices on their exports to compensate for the supply chain disruption and they still get the items.
Ultimately the consumer pays more, the extra goes the government, and the net impact is just obfuscated taxation and a reduction in both supply and demand that's bad for the economy and other living things.
[dead]
Free markets generally only make sense when at the same scope of the ruling government. When country A can manipulate markets in ways that country B can't or won't, eventually country B will attempt to make trade rules that level the playing field.
Its also worth noting we don't really have free markets in the US anyway.
> Its also worth noting we don't really have free markets in the US anyway.
That entirely depends on what is meant by the term. Markets that are largely free appear (IMO) to have won out worldwide quite some time ago.
The definition matters for sure. IMO, markets (like speech) are either free or they aren't.
We have a lot of government intervention in markets today. From subsidies to tax incentives to regulations and import tariffs, markets are much more controlled than it seems on the surface.
A free market can't exist without heavy government interference (property rights to start with) but not all markers with heavy government interference are free, either
Market regulation and protection of property rights are quite different though. Protecting rights agreed upon by a strong majority of the population isn't really heavy government intervention in my opinion.
They're literally the exact same thing.
Free Market vs. planned economy was always mostly talking points, not a consistent ideology imo.
Even during the Cold War, American farms were heavily subsidized. The abundant supermarkets were not a product of free markets, they were a propaganda piece.
Today the US is pretty far from being a free market. Tax deductions are subsidies. Industry subsidies fund things on the front end, and bailouts are essentially subsidies after the fact.
And on top of that there are plenty of (good and bad) regulations which distort the market. For example it is illegal to import foreign insulin even if it would be cheaper. In most parts of US metro areas it is illegal to build multifamily housing.
Most of the things you list don't make a market non-free. A free market can still have government regulation and distortion. In fact it requires it otherwise it will be captured by large players in short order and become non-free as a result.
The insulin example I agree is non-free. More generally the entire medical sector is only somewhat free. However I'm not sure that's a bad thing given the stakes and the history of the free market as it applies to healthcare. The medical establishment itself is an only barely disguised guild system after all.
> A free market can still have government regulation and distortion
The question is "how much" rather than a binary consideration. With the size of the government in most countries, we are way past what would be considered a marginal influence.
> A free market can still have government regulation and distortion.
https://www.dictionary.com/browse/free-market?q=free+market
> an economic system in which prices and wages are determined by unrestricted competition between businesses, without government regulation or fear of monopolies.
By definition, free markets do not have government regulation. If you have an alternative definition of “free market” please feel free to share it.
I think you're misinterpreting the intent of what you quoted. I also think the phrasing of that definition is quite sloppy, being prone to exactly the sort of misunderstanding we see here.
For a market to be stable, fair, and free of monopolies government intervention is required. I don't think that fact is at all controversial. So already the definition you've quoted there has, if read literally, set up a scenario that is impossible to realize.
A free market is one where regulations are broad and are applied to the players evenly. A subsidy that applies to a sector as a whole (ex solar panels) is an example of such.
Many of the agricultural subsidies are much more targeted than that. However the regulator needs to maintain the stability of the entire system as a whole, and to that end food is not some luxury good that can be subject to shortages without major consequences. A tradeoff has to be made, either for more market regulation or significantly less market stability. Market participants in danger of starvation not known for exhibiting reasonable, well thought out behavior.
Similar to free speech, the free market is better seen as a vague guiding ideal rather than an absolute and objective description of a system. It's illegal in the US to make credible threats of violence towards someone. Can that fact be taken on its own to mean that we don't enjoy freedom of speech?
I am not misrepresenting the intent. The definition says “without government regulation”. If their intent was that some regulation was allowed, they would have said that. It’s the job of a dictionary author to be precise.
If I define a “two legged stool” you can’t add a third leg just because two legs is unstable and doesn’t lead to good outcomes. Whether a market with no government regulation is stable or leads to good outcomes does not change the definition.
You are referring to a “mixed economy” which has some markets and some regulations. A mixed economy is more stable because it is able to reduce market failures like monopolies and externalities that free markets cannot escape.
You're arguing semantics and have rendered the term meaningless in the process. The purpose of a dictionary is to communicate meaning. The meaning of a word or phrase is the idea that it is employed to communicate in practice.
When people talk about a free market they aren't referring to some logically contradictory thought experiment. It is something that they actually wish to strive for in practice. Thus it is clear that some amount of regulation must necessarily be involved.
This is quite similar to freedom of speech, as I previously mentioned. The concept itself is an abstract ideal and the context of the law matters.
What you are referring to is a laissez-faire market which as history has repeatedly shown doesn't remain a free one for long. An adjacent comment pointed to property rights as an example. You can have a free real estate market without having a laissez-faire approach to property rights, yet the government involvement in that case is extremely heavy handed. The free speech analogy might be a city controlled by the mob where the government won't intervene but saying certain things will still get you killed.
I agree with you that in practice the US is a mixed economy, but then I already acknowledged that (among other things) our farm subsidies are highly targeted and the medical sector is regulated in a very invasive manner. However I do not agree with your suggestion that eliminating monopolies makes a market mixed. A monopolized market is not a free one because there is no competition.
> For a market to be stable, fair, and free of monopolies government intervention is required. I don't think that fact is at all controversial.
It is. Government intervention is what creates monopolies, because economic value stops being the ultimate metric that decides which enterprise is profitable over another; because it favours one entity over another, large corporations are not valued on their financials alone, but on the strength of their political connections - which in turn enables lobbying as an effective tool.
The entire premise of free markets is that the market is the ultimate judge of value. If you add the government and its heavy hand on the scale to the mix, it is no longer a free market. It’s pretty simple.
As an explicit example, major revenue streams that Tesla (used to) take advantage of are
1. the EV tax credit, and
2. carbon credits
so the richest man in the world/the US had significant tailwinds for a central business venture of his via either directly taking money from the government, or taking money from other companies due to the government requiring they give him money.
> The abundant supermarkets were not a product of free markets, they were a propaganda piece.
The US has abundant supermarkets to this day. It was overwhelmingly a product of the market economy and remains so. The US has between 45,000 and 75,000 supermarkets (75,000 if you include supercenter stores that also sell groceries). That's not counting smaller specialty food stores.
It's a product of consumer spending capacity (net disposable income), of which the US has an enormous amount and has for over a century relative to other nations.
The US subsidizes the supply side and the demand side.
On the demand side, the US spends about $100B on SNAP (food stamps) and another $40B on subsidies for WIC (women infants children) and school lunches.
On the supply side, the US is currently directly subsidizing farms $20B and giving disaster relief to the tune of another $30B.
If you want to imagine how well the free market in the US would do, just think about what would happen if Congress cut off those funds.
> subsidizes the demand side > snap > wic
So these people should just starve?
The common thread here is that it is China. Before the 2010s and earlier, the US wasn't so concerned about China, but ever since then, China has been a big US concern for it being a technology and military rival
e.g. the Investigative Report on the U.S. National Security Issues Posed by Chinese Telecommunications Companies Huawei and ZTE report from 2012:
https://stacks.stanford.edu/file/druid:rm226yb7473/Huawei-ZT...
From that point forward, that concern has only grown. So you can view these actions as screaming "we were only for the free market until there was no competition" but if you want to genuinely know the answer to your question, the publicly stated concern is "China is a national security threat"
> What's the publicly stated/marketing reason for capitalist America to put companies on the Entity List?
“Capitalism” is (as a result of propaganda by its defenders after it was named and accurately described by its socialist critics) often mistaken for a dedication to free trade, but capitalism is a regime characterized first and foremost by society being organized around the interests of the capital-holding class, the first of which is the preservation of the situation in which society is organized around the interests of that class. The reasons companies are put on the Entity List is because they are broadly seen as a threat (long-term or immediate) to the continuation of that regime. That’s what the “foreign policy and national security interests” that form the official basis of the Entity List ultimately, generally, boil down to, in one way or another.
(They don’t always boil down to that, because why the US is basically a capitalist system, it is not purely one, and even in a more pure capitalist regime, individual influential decision-makers may have other interests that they act on besides the implementation and preservation of capitalism that end up getting reflected in policy.)
The USA is not a suicide pact. We do need to consider national interests from time to time.
It’s the same logic behind the Trump tariff regime: “We love capitalism and free markets, except when we’re losing at it”.
Capitalism and free markets are contradictory systems, you can't have both
DeepSeek rose to fame through stolen IP from U.S., and created shell companies to bypass U.S. law. Typical CCP-inspired behavior. The free world needs protection from abusers.
I'm sure there's a ton of other abuses they've committed as they race to become China's preferred LLM.
OpenAI and Anthropic also rose to fame from stolen IP from the US.
https://www.npr.org/2025/09/05/nx-s1-5529404/anthropic-settl...
I assume you're not confused about the fact that DeepSeek winning would be against US self-interest, and the thriving of OpenAI + Anthropic domestically and globally is very much in the national interest of the US.
Even if both groups did exactly the same thing, it would be irrational for the US to not bias itself in favor of its own businesses.
There's bias and then there's ignoring all of your own laws when it's convenient.
I think you meant to reply to the comment I replied to.
That's the key difference: from the U.S.
The primary goal is growth of wealth and power for the country, and close loopholes whereby enemy nations steal even more from America.
Your original comment implied that stealing American IP and creating shell companies is “CCP-inspired behavior”. This suggests there is some high minded Rule of Law based justification.
Now you are saying that stealing IP was never the problem, and actually the issue is anything that stands in the way of the US gaining wealth and power.
Theft is theft. My argument is re: the White House's actions.
Theft is theft, and copyright infringement is copyright infringement, which is distinct from theft.
The US should consider legalising some of this stuff, if I look at a leaderboard something like the top-10 models are built by companies facing serious accusations of copyright infringement. I assume the Europeans are obeying the law or whatever which is why they've so far only achieved peak-2024 performance or whatever and are making no particular contribution to the cutting edge, unlike the Americans and the Asians.
Come again how these laws are promoting useful results? They seem to be economically crippling. The free world should consider embracing freedom from these laws as it seems that will bring greater prosperity.
I can't tell if this is satire.
DeepSeek did the same thing the american companies did on a much smaller scale.
They took the output of one company and trained a model.
American companies took the output of all IP they could illegally* acquire and trained a model.
The world does need protection from abusers, you're right.
EDIT: illegal in some cases (see 82TB of torrented ebooks), immoral in most.
So did OpenAI, we should ban it
that is open to debate. Any commercial activity with a sanctioned entity has a pretty broad interpretation. Companies might not want to take the chance even if they are "still allowed".
So this is the other side of banning American models for non-Americans? And how exactly do they plan on enforcing all of this? Great Firewall of America?
This is a complete joke. The malicious clowns behind this should be removed from power and prevented from ever holding any position of power in any form of governance system.
They're probably going to ban GPU exports to China. Which will of course accelerate the development of their own GPUs. More products for us.
China is already going as hard as they can on their own GPUs. When has availability of non-Chinese tech in China meant China didn't ultimately come up with homegrown replacements?
It's just my general impression. They banned China from the ISS, China made their own space station. China's making their own x86 chips, their own GPUs.
As a fellow wheel reinventer, I admire their audacity. It's the sort of thing that makes me wish my country was like China.
> makes me wish my country was like China
I admire their governance ability to have long term plans.
The sheer scale of their production ambitions in so many fields and their energy build out is insane.
The fact those plans also have subclauses ensuring the party elite are made even more wealthy and powerful is less alluring.
> The fact those plans also have subclauses ensuring the party elite are made even more wealthy and powerful is less alluring
Our oligarchs do that too and all we get in return is declining infrastructure and paranoia.
Classic case of getting downvoted solely because it's a negative comparison to China. The letter for letter exact same comment in a thread only about the US on here, that doesn't mention China, would get upvoted.
You are, of course, right. All of the downsides with none of the benefits.
Well, it's not as if there's not an elite enriching and entrenching themselves here either.
Reading about Thiel's "Dialog" meeting, I wonder, do China have their own equivalent of such blatant corruption meetings?
Even if they do, do they discuss "compliance collars" for the security guards in those meetings? I have a feeling they don't.
The future the Chinese elite wants to own, at least seems to include us as something other than a zombie army at the gates of their post apocalyptic bunkers. That's a low bar I know, and not much to be hopeful about, but our elites aren't even meeting it.
>The fact those plans also have subclauses ensuring the party elite are made even more wealthy and powerful is less alluring.
In which major democratic western country don't the elites get wealthier? In fact, the top 10% asset owners saw the biggest post-covid recovery, which the rest stagnated or are in decline. Not defending China but are own oligarchs are just as savvy at enriching themselves while screwing us over.
> It's the sort of thing that makes me wish my country was like China.
PR China is still pretty poor (around 31k$ gdp per capita adjusted for purchasing power) and its growth has lost a bit of steam recently.
You should wish your country to be more like Taiwan or South Korea. Or Singapore.
I think the PPP calculations may suffer from the same problem as the USA inflation rate being nominally low. Housing, food and medicine are all ludicrously cheap outside of tier 1 cities.
> 31k$ gdp per capita
My country is currently at about a third of that.
> You should wish your country to be more like Taiwan or South Korea. Or Singapore.
Total fertility rate for all of those countries is close to 1.0, including China's. They are dying societies.
https://en.wikipedia.org/wiki/List_of_countries_by_total_fer... suggests Chad and Somalia and DR Congo are the most vibrant societies by that metric.
You might want to compare Singapore with a city like NYC or London, not with a territorial state. It's pretty normal around the world for cities to be replenished mainly by people moving in.
(Of course, to be fair you then also need to compare GDP per capita against other cities. And they usually do a lot better than territorial countries that include a lot of hinterland.)
What if you omit states that depend on welfare?
I never said a high TFR means a vibrant society. A low TFR is indisputably the slow death of a society.
> A low TFR is indisputably the slow death of a society.
No this is a non-sequitur.
Low fertility only means that the affected society grows smaller. This is only equivalent to "slow death" when it persists for centuries until the society is actually dead.
Not necessarily if you take immigration into account.
A society replaced by immigrants is no longer that same society. My original point still stands.
In the case of Singapore it would be just the opposite:
Singapore was and is build on the back of immigrants. Replacing ourselves with kids instead of immigrants would be the big change.
Whereas a society replaced by children...
Whereas a society replaced by its own children maintains continuity of community, nation, or broad grouping of people having common traditions, institutions, and collective activities and interests
What about those societies that have traditionally replaced themselves via immigrants? Moving to a replacement via kids would be the change for them.
Vatican is the deadest of societies, I guess?
Correct
Not only x86 chips, they are going in fairly hard on Risc V and Loongarch (MIPS/Risc V inspired ISA). Risc V is still growing trying to catch up to ARM, while Loongarch LA664/LA864 chips are much closer to x86 performance than other options. They still are many years behind but not as far as you would expect.
GPU's are still a fair way behind with Moore Threads S80 being a better example of their high end. I suspect they have some major driver issues because they current benchmark far below what that silicon should be able to do. https://en.mthreads.com/product/S80
There is also the pressure to have them innovate on older process nodes so they can make this stuff domestically. For instance Huawei is doing what they call 'logic folding' which is basically just stacking dies in a way that ends up reducing the overall size of chip features. Not sure how it addresses thermals but it is a cool idea.
Sorry this article is a bit of LLM rubbish but you get the point - https://www.geeky-gadgets.com/huawei-logic-folding-moores-la...
China is hungry and that is driving them to take these moon shots, they may just make it.
You probably wouldn't feel the same if you were born into a minority population like the Uyghurs. Or even to many of the poor underclass.
Sadly it appears the current US administration is also determined to undo any progress by minorities over the centuries.
They greatly increased efforts when the US restricted high end exports to them. Unless further restrictions accomplish something worthwhile in the short term they seem unlikely to be of benefit to the US.
I saw they have the ASML-type lasers working but are a ways out on getting it to print.
we said the same thing about their home grown GPU datacenter, and look at them now, with multiple facilities 10 years sooner then we thought possible.
trump just undid the gpu export to china to help jensen make money
Oh, that's great, finally they put that cat back in the bag, surely it can never escape.
Gonna be interesting to see how the ecosystem looks like 2-3 years, I'd be expecting some drastic changes compared to today.
3 days ago the deal was still on https://news.bloomberglaw.com/us-law-week/trumps-china-nvidi...
China already produced their first graphic card that got Microsoft WHQL certification.
Is it competitive enough? Probably no. Is there a software ecosystem for Chinese hardware yet? Probably no. But you know it will come soon.
> And how exactly do they plan on enforcing all of this? Great Firewall of America?
That's what Cloudflare will be for.
You're aware that many countries are blacklisted from trade with the US already, along with certain segments of existing companies. It just means that enforcement comes with contracts, law, banking systems, etc.
> how exactly do they plan on enforcing all of this?
Video selfies and government document upload on hardware attested phones, of course.
The same way you can't just open a bank account anonymously. It seems inevitable to me that this is going to end up with a KYC/AML style digital online ID.
Without that you are not using Claude 10. It would not be shocking at all at some level of capability that running open source models anonymously is made a crime.
To pretend there is not huge issues with a laissez faire attitude as these models progress I think is intellectually dishonest and performative.
It's likely the elected officials know nothing about AI the technology.
Depends on who has their ears in terms creating policies around technology.
Banning American models for non-Americans seems like a big reduction of the potential customers for American chatbot companies.
Until they scantion Brazil
It can happen to anyone
The US is slowly becoming more like China. From talks of nationalizing companies to make US state owned entities to banning foreign competition. It's just so strange how you become the thing you fear.
> The US is slowly becoming more like China.
Hah, if only. Then at least it would see some of the benefits too. Instead, it's making sure to only copy all of the downsides, with zero of the upsides.
This is the kind of flippant remark that does the gravity of the situation a disservice. Nobody burns witches out of the blue one day, they do it because they figured out that burning witches makes everyone afraid to do anything that could be remotely construed as witchcraft, and they all start being good little church attendees.
So I hate to break it to you but you're not fighting powers-that-be who are just bumbling around and decided to randomly be like China one day. You're fighting powers-that-be who did the analysis and decided that being like China was the right way to go, for the good of everyone, and they backed it up with math and logic. You're going to need math and logic to convince them otherwise.
> You're fighting powers-that-be who did the analysis and decided that being like China was the right way to go
Not at all. If they decided to be like China, then that would include the benefits. They're nothing like China. Xi isn't siphoning hundreds of billions to his personal account through $XI coin. He isn't giving the reigns of the country to his tech CEO buddies. He isn't destroying scientific institutions and looting the country for personal gain.
I hate to break it to you but this is the kind of surfacelevel comparison that doesn't hold up when considered for more than 2 seconds.
I agree actually, we're trending towards China in some ways and not in others. I was just pointing out that the ways we're trending towards China are driven by people who did a bit more analysis than OP is giving them credit for. Since your real disagreement is with OP, I'll bow out of this one.
> who did the analysis and decided that being like China was the right way to go
But again, they are not deciding to be like china, which implies a degree of competency and care for the population. This is just corruption.
Bro, you have "grass is greener on the other side" syndrome. China is a totalitarian state not comparable to the US. Take this from a non-US citizen. There maybe some backsliding or ppl may think things are worsening in the US, but US citizens always have the power to make things right, coz they are free and they can bear arms.
> coz they are free and they can bear arms.
The propaganda about ourselves we sold the world has been much more effective than the rights we've been sold ourselves. I would prefer material rights and competent governance to either "freedom" (whatever the hell that refers to) or the ability to buy guns we're barely allowed to use.
I never said that "as of June 2026, the grass is greener in China". What I did say is that the US has rapidly been adopting all of the downsides of China with zero of the upsides. That's an observation.
> There maybe some backsliding or ppl may think things are worsening in the US
If you're phrasing this as "maybe" and "may think", you're not here to have a serious discussion.
The CCP would never dream of giving the state's control to tech CEOs.
You just need to think outside the box: the CCP is big tech, not Congress. The CCP (big tech in the analogy) would never dream of giving control to congress.
This is why they can run their society with such competency, though. All we got are the stupidest billionaires the world can imagine and their servile lapdogs in congress
In some ways its hard to differentiate between big tech and the national security complex, where execs are being sworn into the military
https://www.army.mil/article/286317/army_launches_detachment...
I think America is becoming more like Russia, with a dictator who eliminates any billionaire who disagrees with His Dictatorship. The American groups who want to go back to the "glory years" of the 1950s are equivalent to the Russian groups who want to go back to the "glory years" of the Soviet Union.
This is exactly where we are headed: a has-been that sold itself down a river to a bunch of kleptocrats who only are about themselves and their sycophants.
It's a slow motion motorcycle crash[0].
0: https://en.wikipedia.org/wiki/Target_fixation
I wish we were becoming like china; it feels a lot more like big brother from 1984.
> talks of nationalizing companies
What is this referring to?
https://apnews.com/article/bernie-sanders-ai-public-ownershi...
Not just Sander, Trump expressed his wish to exchange concessions and privileges for share in AI labs, too. Which is IMO even more problematic.
Nice, doubt anything will come of this though.
I was referring to the US government buying a 10% stake in Intel and the ideas being floated of further investment in Intel and other companies by the Trump admin.
Hmm, my VPN provider explicitly has Chinese exit points. And whats funny is I can load AliPay from any CVS. (Like, seriously)
You can try to pry Qwen and Deepseek from my Graphene/Linux hands.
What VPN provider is this? I could use it because Chinese users of my apps often complain about not being able to download things from my western hosted servers.
Just an anecdote,
I lived in China for a bit years ago and one the biggest issues accessing western websites weren't restrictions against the site. Most of the times the culprit was using CDNs or services from Google or Cloudflare which were restricted totally or partially.
I was working on a site around that time, learned about that fixed it for the chinese user base, after that users from china went up considerably.
Not sure if having point of presence (POP) managed DNS for China is of interest, but my company offers something for China traffic [0].
Disclosure: I’m an IBMer
[0] https://www.ibm.com/docs/en/ns1-connect?topic=started-manage...
PIA is my VPN vendor. They and Proton both support port forwarding. Port forwarding is a hard requirement of what I do.
And given Proton CEO's support of the US administration (specifically for Vance), I chose to cancel my subscription of 3 years with them.
When I was in China the only VPN that worked using China mobile and other local ISPs was Lets VPN, they route through HK.
Why do you need to "load" Alipay?
You can just have them charge your CC directly whenever you make a payment? Or is this blocked for US cards somehow?
CVS?
CVS (drugstore chain) and Walgreens were among the first major U.S. merchants to accept Alipay via QR code payments, allowing Chinese consumers (and anyone with Alipay) to pay at these stores.
This person is correct.
I pay for 2 servers running in Asia under Alibaba, using my local CVS drugstore. Im in the Midwest USA. Not a single problem at all.
Sorry, I am not familiar with Alipay, at least in the use case you're referring to; How do you exactly pay for 2 servers using Alipay from a CVS?
I can load US dollars I hand over to the CVS attendant and load it on my AliPay (Alibaba account).
I can then pay my Alibaba server bills with Alipay.
Why not just go directly from a credit card to alipay? Does that not work when physically outside of China?
Is there an intermediary?
The cash doesn’t go directly from CVS to AliPay, right?
CVS also accepts alipay as a payment method, so they must have some kind of direct partnership. I don't see why they'd need an intermediary, although maybe they take the cash and just credit alipay some balance etc.
Super interesting and unexpected!
Thanks (to all!).
Hmm I have a couple Alibaba Cloud servers located in CN Hangzhou region and I just pay with my U.S. credit card, no Alipay needed. alibabacloud.com is meant for international customers anyway. I’m not sure if aliyun.com accepts international CCs, maybe you’re using that.
I also have at times ordered deliverables on Alibaba (not aliexpress). Being able to send funds in yuan over Ali network solves almost all problems with that.
If Chinese LLMs are successfully making people in the west defend China, then I think we have all the evidence we need to explain why they are giving away their models.
The next step of course will be to get people using that ungodly cheap AI on Chinese servers. Which will also be defended because "I would never trust an American Lab".
It's funny how you can look at obscenely overvalued Western AI companies, with capex's that are unsustainable, and along comes a system which pokes a gaping hole in this model, and the response is "Ya, but what do those sneaky Chinese have up their sleeve here?!"
Perhaps the question should instead be "Why are these Western AI companies getting insane valuations on dubious ROI, and how can these Chinese models run on a fraction of the infrastructure?"
Not saying you're doing this specifically, but I'd be careful with thinking that "company" in China means the same as "company" in America (or in the West more generally).
True, "Industrial Policy" [0] is much bigger over in China, and has been for a long time.
Although that might become "was" given the recent shift in US politics, where the Republican party of big-government has been heavily taxing Americans with tariffs [1], nationalizing US companies [2][3], and commanding the manufacture of specific goods [4].
______
[0] https://en.wikipedia.org/wiki/Industrial_policy
[1] https://thinkinthemorning.com/trumps-great-leap-backward/
[2] https://www.cato.org/commentary/yes-trump-just-nationalized-...
[3] https://time.com/7313446/trump-seizing-private-companies/
[4] https://www.cnn.com/2026/06/17/politics/trump-weapons-iran-d...
>"Ya, but what do those sneaky Chinese have up their sleeve here?!"
I feel like that is way over simplified. I do not trust the American government. I do not trust the Chinese government. As an American, I believe the Chinese government has a longer and broader history of stealing intellectual property and far less checks and balances than the American government. The current American administration will be gone soon and maybe we can get some sanity back at some point. Overall, I trust my data in America more than China and I think that's reasonable. I am not naive. I'm aware of all the problems with my country and am quite vocal about it. In fact, I think it would be naive to think that the Chinese government won't have complete access to everything that goes through a Chinese server.
We are in full agreement on your second point.
> a longer and broader history
From what perspective? The American colonies repeatedly and flagrantly ignored foreign property and intellectual rights, e.g. via laws to protect domestic but not foreign authors. Samuel Slater was called Slater the Traitor in Britain for a reason.
stealing intellectual property == producing higher quality goods for cheaper, because the IP owners are not paid for it
One thing is that many of these Chinese companies have had to do more with less due to technology export restrictions. Doing more with less is not something that the big US AI companies think about much. The Chinese models are apparently much more compute and energy efficient.
Because the revenue of Chinese AI companies is small. Anthropic's annual run rate is $50bn, z.ai's is $500m.
The main cost is training the first version of model. It is very easy to just train the copy cat model based on the output of another model.
That doesn't make any sense. If the cost is training, the cost is compute, not data. Distilling another model means paying for that model, which isn't obviously less expensive than crawling the web and curating a dataset. Regardless of where your data comes from, the training process costs the same.
> paying for that model
They probably just used 100000000 free plans.
this is a justification for why one frontier lab would have extremely high spend rates. There is >1 frontier lab though.
More explicitly: if the Chinese can get near-leading performance models at a fraction of the cost by stealing from Anthropic, why doesn't OpenAI? Lord knows they could use the extra cash.
>It's funny how you can look at obscenely overvalued Western AI companies, with capex's that are unsustainable, and along comes a system which pokes a gaping hole in this model, and the response is "Ya, but what do those sneaky Chinese have up their sleeve here?!"
The only way this is funny is if you are completely oblivious to how China usually operates.
> If Chinese LLMs are successfully making people in the west defend China, then I think we have all the evidence we need to explain why they are giving away their models.
Right, and if "The West" wants people to defend them, they better get in on the free action too.
In fact, they have no choice - tokens are soon going to be a commodity, if they aren't already. Most everyone is going to be happy paying 1/20th of the cost for 80% of the value.
Oh, yeah, before I forget, hear the worlds smallest violin, playing for those token suppliers in "The West" whose repeatedly stated goal is to replace human knowledge workers...
I think the next step is for China to start selling Huawei (and other) GPUs around the world, because everyone appreciates the risk of giving either of the two superpowers their data. The US is no longer in a position to pressure other countries to bad imports from Chinese companies.
This is the beginning end of the hyperscaler era, not a shift from US hyperscalers to Chinese hyperscalers. Taking Nvidia's extremely lucrative market is the goal.
Software ecosystem aside, their hardware is inferior to NVDA/TSMC's and will remain so. And even if it wasn't, China just doesn't have the fab capacity to both meet domestic demand and export enough to hurt demand for NVDA.
They're behind, but they're catching up pretty quickly. I wouldn't bet against Chinese companies dominating this market in the medium to long term. Nvidia could easily end up being the Tesla vs Huawei as BYD.
It’s not like this is a new tactic, china has been very successful at wiping out competition by undercutting them using state subsidies in many sectors of the economy.
> wiping out competition by undercutting them using state subsidies
The same tactic is in the USA. Like every new AI datacenter has ~10 tax waiver + explicit subsidies + favorable loans, etc, etc.
Not even mentioning the many instances where US corporate interests were defended and advanced by US military force abroad.
As always, the US has a government removing red tape to foster innovation, while China has a regime, unfairly picking winners, to hurt and subvert the West.
this is not an accurate picture of Chinese industrial policy. In fact, you could argue they have the opposite problem. Their industrial policy encourages too many companies to enter a space, where the resulting competition kills off profit margins for whichever companies end up surviving until the end. This is exactly what we end up seeing with extremely cheap Chinese goods.
If china anointed one company per sector, they would have no reason to be so cost competitive globally. There would be no structural cause for Chinese products to outcompete the rest of the world. You can see some of this in the US, where these kinds of anointed companies exist (say e.g. Boeing/other defense contractors, at least post the 90's). They are (famously) not particularly cost competitive. This is also exactly what you'd expect economically.
Please read up on what's really going on inside China when it comes to industrial policy and innovation. Barry Naughton has a nice series of books and papers that might disperse some of your incorrect preconceptions.
Our Blessed Homeland
Their Barbarous Wastes
These are done with the intention of getting kickbacks, not wiping out competition
> get people using that ungodly cheap AI on Chinese servers.
These open weights models are also hosted outside of the US and China. That's a very important difference.
The smallest violin in the world for Sam Altman, and Musk
As a European, I trust Chinese AI providers more than American. Cloud act did it for me.
Canadian, and I also agree. It’s hard to avoid but I try not to use any American service or data storage.
And that's fine of course, but it's worth noting that you're making a decision driven by emotion rather than data.
How so, with the Snowden leaks we learned the extent of American digital espionage in Europe, the US government puts pressure on Europe to prevent taxation or regulation of American business and even European citizens have become the subject of mistreatment in American airports based on their digital profiles. We can enter China visa free.
Given that you're big on data and don't like emotions, what have the Chinese materially done to us Europeans we ought to care about?
Selling Russia weapons to help kill Ukrainians, buying Russian oil/gas to help kill Ukrainians.
But free AI models or something, right?
They don't sell weapons to Russia, as Wang Yi said in Brussels, if they'd put their full weight behind Russia the war would be over, this is the middle position for them.
Which, I agree with you btw, I don't like but I can understand rationally. We still buy oil and gas from Russia too. And with 20% of the world's supply casually going offline after America's own 3 day special miltary operation the Chinese would be insane not to.
What one cannot understand rationally to come back to the Snowden era in which Denmark spied on us in Germany on behalf of the US, which is insane to begin with, is being threatened a decade later with annexation of their territory. China is on some fronts a rival, hence the tension around Russia, but the US is as destabilizing and unpredictable for the world as Russia itself now.
Americans truly seem to have no concept that they're in the middle of their own post-Soviet meltdown and look like a rogue state to the world now, which makes China which is no saint more and more attractive simply by being a source of order.
The other option is the one that kills palestinians, and we don't even get free models out of it.
But don't you get it, Ukrainians are worth more than Palestinians because... some reason
Because Israel. Call out the problem by name .
It's hard to call either side good. It can be argued that the side which had the highest powers and oligarchs implicated in Epstein files, and which has also threatened attacks on European nations, isn't the better option. Also the nation which actively funds the war in middle-east.
If only the US was a little bit more like China, nobody would ever know about the Epstein files.
You think China doesn't get its hands dirty because of some moral superiority? The country is utterly brutal towards its own citizens, what makes you think that the lack of warmongering is anything but inability?
The threats made by Trump are in my dataset
Both China and the US can compel businesses to hand over data. There is no reason to trust any service that doesn't have strong built in privacy.
Compel? I am confused, all data in China is held in datacentres which the state has full access to, that is the terms of their operation and why some big tech US companies didn't want to operate in China. They don't need to "compel" anyone, the CCP has people at every large company supervising employees, and they already have full access to your data.
I am always completely baffled by these comments that not only get basic facts wrong but appear unable to conceive of a situation where the everything is subordinate to the state.
There is no negotiation, there is no due process, you give access to everything before you start or you can't operate.
Isn't this essentially true in the US too though? The feds can show up at your data center with a National Security Letter and demand access at gunpoint. And you have to give it to them, because guns, and you can't ever tell anyone about it because that's what it says in the National Security Letter and also because guns.
Your reply already shows the difference. In US you have the default expectation of privacy, until the feds get their eyes on you. Meanwhile in China the default expectation is no privacy, exactly as OP argued.
This system in action is best demonstrated during the lock-down period of COVID, when any random dude who contracted the virus would immediately have their personal life for the previous week/month published nationwide to the hour, and those who have overlap will immediately get a 14 day lock-down at home.
I have not seen surveillance done with such ease and breath elsewhere. And local PD already have access to such info, and there are scandals where police sell such info for profit.
Is it so different in the US? These are just the surveillance we know about, and they're not openly telling us about these, they're actively fighting public knowledge of such programs:
https://en.wikipedia.org/wiki/Global_surveillance
What would be the practical difference between an order from a party cadre in a private firm and a national security letter?
A legitimate legal system with judges who have no obligation to anyone or anything besides the constitution first, and laws second.
And what makes you think that the US still has such a system?
True that it is collapsing which is sad but it does, in part, still function. We are the proverbial frog in the pot.
Neither of them has that
Post-Snowden, thinking that the US national security apparatus is subordinate to judges and the Constitution is naive and uninformed.
It’s not China that is threatening to annex Greenland though.
Because China has never invaded or annexed a neighbor ... ?
When it comes to annexation, China doesn't threaten, they just invade and extinguish.
Yeah, I remember clearly when China invaded/bombed: Vietnam, Afghanistan, Iran, Iraq, Venezuela, Panama, Syria...
You absolutely should remember the Chinese war with the first of those; if you have trouble, a good way to remind yourself is that it was not long after the US one.
Yeah, so getting 1 out of 10 he mentioned, even if it's their direct neighbor (where disputes happen for all countries), ain't bad! This absolutely means they're the same /s
If I had wanted to say “China and the US are the same”, I would have strung together a set of words that looked a lot more like “China and the US are the same” than the ones that I actually posted.
https://en.wikipedia.org/wiki/Sino-Vietnamese_War
Could you give me some examples? Which wars do you have in mind?
Tibet, Manchuria, it should be somewhat obvious that a nation that is as ethnically diverse as China was not a nation borne out of lots of different people deciding simultaneously that they would like to create a country together.
What is modern China has only existed for 100 years or so. When the country collapsed there were ethnic divisions that were erased after the country was unified.
The hallmark of successful ethnic cleansing is people claiming that there were never any wars, that things were always this way. The same is true of Kaliningrad, the most German city, centuries of history as a leading nation within Germany and the HRE, now a completely Russian city. It is only in the West that you see any narrative around division, in places like China or Russia history is erased (and how could it be any other way, the cornerstone of Chinese politics is one nation, one people...there is no political value in this narrative in Western countries).
manchuria? are you sure that's a good example?
https://en.wikipedia.org/wiki/Annexation_of_Tibet_by_the_Peo...
Could you give a recent example? Because your example is as pertinent as the overthrowing of Guatemala's democratically elected President in 1954 by the CIA.
I don't remember the U.S. federal government declaring Guatemala a U.S. territory and asking all Guatemalans to change their official language.
What China did in Tibet is closer to what the U.S. did with Hawaii.
And when it comes to the US, every accusation is a confession.
It certainly wasn't a denial. China could learn a lot from that.
There's nothing wrong with the US buying greenland, which has been done for territories around the world?
The US has a long history of protecting individual freedoms, China does not. There's an irony that you're aware of the misgivings of the US because we have free speech protections. You're probably less aware of the misgivings of the EU because they regularly arrest citizens for speech, and no awareness of the issues with China because they'll just disappear journalists in the night.
You're kinda missing the important bit there: buying Greenland under threat of invasion
I think you're kind of missing the important bit; there is no threat of invasion and there never has been. People just want reasons to be mad at Trump, and so they make things up
> there is no threat of invasion of Greenland by the USA and there never has been.
This is a dumb take. You're starting with "I don't take their threats seriously" and conflating it with "it didn't happen at all". But it did happen.
> The Trump administration threatened military action to take control of Greenland
https://en.wikipedia.org/wiki/Greenland_crisis#Annexation_th...
How long has the history of the US protecting the individual freedoms of non-white or non-rich citizens been?
Since the late 60s there has been explicit federal legislation. I assume you meant this as a rhetorical question but there is a definite answer because the US is a transparent system. Law is passed, the courts enforce.
The fact that you are unable to answer this question about China where it is often unclear why certain people are being targeted should demonstrate how big the gap is. For example, when Xi's first corruption crackdown happened, it turned out that it was being orchestrated in part by someone in their 90s who had left front-line politics twenty years ago (and much of why that happened is unclear, we are only just learning about things that happened in Chinese politics multiple decades ago so it will be a while before we know...we do know that Xi was then able to make unilateral sweeping changes to his own role shortly after that broke every convention of the last 5 decades)...it is difficult to compare this to anything that happens in any other political system. In China, you find out someone has been executed for reasons that are obviously not explicit months after it has happened.
It is genuinely quite difficult to compare to anything else. 6 people meet in a room, they have almost all the power, and maybe you will read about what they might have said 4 decades later in a book...that will never be published in China.
There was legislation, sure. Were the people actually protected or did they just write a paper saying they were protected?
>The US has a long history of protecting individual freedoms
You can't be serious.
Was the individual freedom of those 120 Iranian girls protected?
Was the individual freedom of Renée Good protected? Alex Pretti?
The inability to conceive of a country where these things would happen and you would have no idea that it had ever happened...and, perhaps more importantly, wouldn't care that you didn't know.
Yes, but China can't arrest me if they don't like what they see in my data.
USA and its vassals can.
I spoke with a high up fella in a uk firm - same thing - trusting china more. Couldn’t believe it when he said.
Americans you wanted isolation - you’re gonna get it eventually!
China has something equivalent but as usual, it's harder for their spying to affect you. For now.
I trust them as a cloud less than US but don't completely trust US.
As an American, I trust Chinese AI providers more than American.
Same
The chinese providers are just the CCP. They don't even need a cloud act...
I think too many people are conflating Chinese providers with Chinese models - you can easily have Chinese models safely (well, relatively safely, I guess) hosted on US or EU infrastructure.
Case in point, Microsoft just announced it is toying with the idea of using DeepSeek as a cheaper model tier in CoPilot. They are hosting the model themselves.
Exactly this. For some reason this is constantly being overlooked/confused. It is very possible to deploy these models inside your own VPC on the big cloud providers and have it be completely secure. I would argue that is even more secure than trusting a model provider’s native API as your traffic is not staying inside your own controlled cloud environment.
I was reading the threads about local AI closely yesterday. Some people seem happy with it.
If I had the cash, I'd spend 6-10k on a strix halo with 128 GB and run it local with no internet connection. I think the Framework desktop is sold out but there were others seem to be still available.
Interested in the reply to this.
The next step of course will be to get people using that ungodly cheap AI on Chinese servers.
Step one of this was perhaps DeepSeek's incredibly low cache hit pricing ($0.0036/M) which no-one else seems to be able to match.
I feel like it's less "defend china" but more "don't wanna have US have monopoly on it"
Talking with an American trying to sell me on "don`t use deepseek because of moral reasons" really made me much more invested in China.
I'm sorry but USA no longer has any moral superiority over China.
Found the late-stage OAI/Anthropic bagholder lmao
Yes, they have been coming out of the woodwork like borer beetles. Not exactly subtle.
I understand it's supposed to be obvious to all of us, but maybe just for fun, can you follow through? What is the next step in the nefarious plan?
Not OP, but IMO the Chinese are waging economic warfare by "commodifying the competition". US VC's and equity are massively invested in AI being very very profitable and as soon as it's not that becomes garbage debt on a balance sheet, and a lot of it. When money is invested it's expected to make a return. When a few people make bad investments they lose their bag, when many many people all make the same bad investment, you get a nasty recession. When the US goes into recession China is more emboldened to pursue their agenda because policy makers are distracted by shoring up support at home and not as interested in expanding their power abroad.
Free market economics is constant warfare. It's called competition, and if you're really doing free market economics, you don't get to shut down the market every time you're bad at it.
That sounds like a problem for the idiots invested in this garbage and who cut foolish sweetheart debt deals with the hyper-scalers. That's mostly rich people, because they haven't yet found a way to offload their bags onto retail (though they're certainly trying).
I'm confused. China has been in a fairly bad recession the last ~2 years (their housing bubble popping). Why are we assuming that recessions can only happen in the US?
I'm also confused how China managed to convince so many US VC's to spend an insane amount on a technology with "no moat". China exporting cheap AI hurts American hyperscalars. But it doesn't induce the behavior in American hyperscalars that causes them to be vulnerable to cheap AI. It seems kind of patronizing to point the finger at China, rather than acknowledge the American (potential) misallocation of resources.
You're attacking the strawman of "China made VC's invest in AI". That didn't happen nor was it claimed. China is being opportunistic here.
[flagged]
The present US administration and its backers don't want diversity, democracy and non-whites. Where does that leave non-Chinese and non-US citizens ?
Waiting until the next election, mid-terms are in a few months.
They can do that without open sourcing their weights. In fact, that would be the best way to do it. So why are they open sourcing their weights?
To create goodwill! Nefariously!
The US truly can't consider foreign policy that doesn't help propagate the military industrial complex.
It's like the idea of diplomacy and collaboration are two words that do not exist.
> They would be looking to do large scale espionage, whether it be corporate secrets or personal secrets.
The US token providers started the ball rolling on large-scale copyright infringement in order to make their models work.
They built their business on IP laundering, so it's very difficult for me to feel sorry for them now.
Not sure I follow your point, are you confusing espionage with IP theft? The victims are entirely different in each case, I am sure china would not restrict itself to corporate espionage.
> Not sure I follow your point, are you confusing espionage with IP theft?
Espionage is IP theft.
What did you think that word means?
I may have got it a little mixed up with sabotage :/
In your take, why are they producing and giving away such good local models? Mindshare? Promotion?
Soft power primarily (see my original comment),and less data for western labs to train on, less money for western labs to have.
The state can use their AI as a tool to bolster their standing, which is working well. Chinese AI labs don't have to worry about making money or affording more compute. The state will (and does) give them whatever they need.
Oh no! A group of people doing useful and good things to increase their reputation! That's terrible!
But seriously, if it works so well, why don't we do it too?
Because the US is a democracy and there is blowback for pissing off the public. So allocating hundred of billions of taxpayer money to AI would be enormously unpopular. We have private markets that can spend their money however they chose.
China just has the state with one leader than thats it.
But the US did allocate hundreds of billions of taxpayer dollars to AI...
Not who you are replying to, nor do I agree with their take (no sign of irony, complete lack of self awareness, and just blatantly xenophobic. not suprised by it either). but they did already address this:
> it's called soft power, look at endless glazing, it works
So people in countries adversarial to the US, or even demographics at risk domestically like illegal migrants have nothing to worry about dumping their info OpenAI or Anthropic servers? But the Chinese are going to spy on people using open weight models running on private or third party servers? Do you hear yourself? Do you have any grasp of theory of mind at all? Is your vision of Chinese "ethnonationalism" white people picking cotton? Is that what you are afraid of?
The untold capital and effort, the disciplined, complex long game operation, all to get 10 million people a second asking "are there timezones in space?" or "excel bar chart colors?" or "do you think I'm beautiful?"
It's hard to truly grasp the enormity of the evil, really.
[flagged]
If they had encrypted LLM compute then in theory wouldn't matter who/what is running it
Well they don't, and nobody has any idea how to build that, so why even mention it
It is being worked on this is just a random paper I pulled but yeah
https://arxiv.org/html/2502.01013v1
few of em actually
https://openreview.net/forum?id=PGNff6H1TV
fully homomorphic encryption is a theoretical thing that runs a billion times slower than running the code normally, to make that more clear, it takes a year to run a millisecond's computation.
[dead]
Funny that Europe starts to become China, just without the manufacturing and growth. I can see why people like them.
These bastards already prevent me from buying a BYD car, and a xiaomi phone, and they are adamant about me not using a chinese AI model. I hope they do not succeed.
I am using a Xiaomi phone. The hardware is fine, the camera is very good compared to the average, the software absolutely sucks, bootloader unlocking is not possible and it comes full of spyware and adware that you can't disable.
Well, I guess you can use a third-party provider, maybe even the US-based one. At least, you can do this at the moment of this writing...
Ok I'll bite: what is the obvious thing I'm supposed to be getting from your ellipses?
are you implying there is a US-based, third-party provider of xiaomi devices, BYD cars?
or are you just referring to hosted providers of those AI models?
> what is the obvious thing I'm supposed to be getting from your ellipses?
Well, I'm unsure if it's a correct form of expression in English, but in my native language ellipses can often serve to express speaker/writer feeling of uncertainty about something or some kind of sadness/apathy/similar towards the situation being discussed. It's not about some "obvious thing you should get", sorry for misunderstanding.
> or are you just referring to hosted providers of those AI models?
Yes I am. I was not talking about BYD or Xiaomi; I think I focused too much on AI because that's the main topic in discussion here. That was my mistake and I apologize for that.
No worries, I was just confused, and in part was genuinely wondering if there was a vendor that sold xiaomis in the states because I'd buy one.
The ellipses in American English is often used as an awkward pause, or to indicate trailing off, or sometimes to indicate a sarcasm or passive aggressiveness. The last one seems to be a more modern usage. I have seen older generations use it as an almost soft question mark to indicate uncertainty, which always confused me because I assumed thy were being passive aggressive out of nowhere.
Ideally a third-party provider of government, i.e., leave the USA, while it's still legal.
Wait, can't this be reduced to:
"American AI is already trillions of USD underwater, so let's use US gov to build us a moat by making competition illegal"
Deepseek is awesome for home projects, I have used it quite a bit and the 10 bucks I loaded is yet to run out. Quality is close enough to Sonnet that with a bit of extra focus, I am not complaining.
IMHO, models by US companies are the biggest security risk so I'm fine with using models on this "blacklist."
Part of the security risk also is the number of different models. I’ve been tempted to try some other models, but how many do I want to give access to SSH or even my repo? Obviously there are ways to work with this, but it’s gonna run through some people‘s heads.
Buy access to the open models from a single US vendor like https://fireworks.ai
One company, multiple models, Fireworks is the fasts at making the models available (had GLM-5.2 before the other three we are evaluating)
Why?
Because they siphon off data to US intelligence, and if you claim they don't, you couldn't possibly know because the CLOUD Act can mandate them to do so without telling you or allowing you to admit it. Of course, if you're in the US this doesn't matter but for the rest of the world it does.
The US government exists to defend capital interests. It's why we can't buy BYD cars. It's why we can't import any cars unless they're 25 years old. It's why a Tiktok sale was forced. It's why the US is seeking to block states from banning prediction markets. It's why the federal government is seeking to block states from blocking data center projects.
As soon as DeepSeek came out I realized what was going on: China was going to make sure that no US company was going to "own" AI. It is an issue of national security. It's why the US essentially blocks US tech companies to maintain sovereignty.
I'm reminded of the browser wars of the 1990s that led to the antitrust suit against Microsoft. Microsoft used the "commoditize your complement" strategy [1] against Netscape. The US has blocked the export of not only EUV lithography but high-end chips to China. China doesn't want to be dependent on US platforms or policy.
So China is going to make sure there are open source models available and the US government is going to try and stop them to protect US tech companies.
[1]: https://gwern.net/complement
The reason why some Capital Interests want to blacklist DeepSeek in the US is so that you are forced to buy Claude/GPT/Gemini, which will feed revenue into an industry that requires revenue (or it’s a big problem).
The reason why some Capital Interests don’t want to ban DeepSeek is so companies that utilize AI have more options, and running your own DeepSeek cluster acts as an independent cost comparison for enterprise inference contracts.
The raising AI valuation is giving more weight to those that want to blacklist DeepSeek. The AI Safety narrative is strong. I see a path where any institution with enough compute might be watched in a similar was chem labs are observed by the DEA.
if you look at share of industry profits, currently most of AI profits are captured by NVIDIA and cloud providers
banning deepseek/open weight models will allow Ant/OAI jack up prices and extract more profits for themselves
keeping open weights models available will keep current industry profit distribution where majority is captured by nvidia and cloud providers
> keeping open weights models available will keep current industry profit distribution where majority is captured by nvidia and cloud providers
With the Chinese manufacturing capacity, how long do you think this will remain true for? You can't spin up a fab in a year, that's true, but you can in five years, at which point all the protectionism in the world isn't going to help.
DeepSeek et al has contributed tremendously, not only by sharing open weights models, but also genuinely improving the SOTA in inference and training, especially given their compute-constrained situation.
if China starts making cheap GPUs for the entire globe, I will say THANK YOU and will continue my work on AI.
The collaborative nature of open-source (free software) is the ultimate good that benefits entire humanity. I value that more than any other benefit of protecting any company.
How much I care about Nvidia or OpenAI market cap? absolutely zero
How much I care about the poor, but smart and creative people on Earth being able to contribute to the state of the art of AI development? 100% more.
> The reason why some Capital Interests want to blacklist DeepSeek in the US is so that you are forced to buy Claude/GPT/Gemini, which will feed revenue into an industry that requires revenue (or it’s a big problem).
It sounds right, but I see zero evidence of this. I think you underestimate how many people in the current administration are True Believers. This does not, to me, seem like anything to do with Big Capital, but rather "America over China".
and it would be great to have an independent auditor have access to all the training material and good search tools, so that take down requests can be made by copyright owners
[flagged]
Do you really think the truth has anything to do with the power of a narrative?
Seems like interests of US government and US capital (monopolize and corner markets, jack up prices, extract economic rent in perpetuity) run strictly against interests of the broader US consumers and overall global population
> It's why a Tiktok sale was forced.
I think that has more to do with controlling narratives that the USG doesn't like.
Ever see a tiktok about may 35?
I don't use TikTok, but a cursory search shows that there's a #tiananmensquare tag that has a few thousand videos, including many about the protests and Tank Man. So while I haven't seen a TikTok about it, someone has.
I don't (could be wrong) think other American's seeing content on American TikTok is relevant to China's censorship of what Chinese people see in China
I don't think that's what the person I was replying to was saying, though.
Why should that mean that Americans should prefer that TikTok restrict narratives unfavoraroble to American oligarchs?
What do you think the Chinese would use Tiktok for if Taiwan heated up again?
Versus the next obvious trap after Iran that dear leader stumbles into?
The China fearmongering doesn't really work in this case, our own government wants the ability to lie to and manipulate us about everything continuously.
Relevance?
I'm happy that China is doing that. US cannot be trusted anymore.
Not saying that China can be trusted either, but I think having more actors is better for all of us.
Those data centers that "states" want to block: the CCP was directly tied to funding the protests and movements to weaken the U.S. technologically. Kevin O'Leary exposed it, in collaboration with the White House.
Or maybe people want to be able to sleep?
How the fuck did the AI bubble industry manage to take one of the cleanest industries and make it dirty again?
That's precisely the CCP brainwashing: the data centers in question were actually VERY clean, and would give the communities a ton of jobs.
This happens all the time for people with agendas: pollution fear mongering for projects that are actually super clean and useful.
Seen it happen in real life, many times.
Why are there audio recordings of them being very loud?
Chinese have a wider outlook on it.
Politically they believe AI belongs to humanity, which is why they are basically the only ones left publishing research in the open. That's probably part of their socialist nature.
But also a financial one. They believe that models are commodities, that you can swap one for the other and that the only thing that matters are the applications built upon them.
So they want to make sure that the world, and their own companies, are not limited in their business and application by a protected US commodity.
They will keep releasing in the open no matter what for quite some time.
It's quite impressive how the latest years I have found more and more to empathize with China than many of the western counterparts.
But it's increasingly clear that since the last decade protectionism and nationalism is taking the place of globalization, even though globalization has been a terrific success in lifting billions out of poverty and making the US thrive.
> It's quite impressive how the latest years I have found more and more to empathize with China than many of the western counterparts.
I'm not sure that this is because China is suddenly a great place or political system so much as a reflection of many western nations speed running to autocracy in the name of manliness.
Moving towards China because of concerns of autocracy is a hilarious concept.
And yet here we are.
Yeah, for me it's the latter. Until recent years, it was at least possible to defend the US as having some good principles, despite how imperfectly defended or promoted they may have been. That in turn could make it worth defending. Now, it's just blatantly turned into everything it always stood against.
> But also a financial one
china's advantage is manufacturing capacity. giving away the model is a loss leader for the hardware business. they have a big gap on chip manufacturing. the only way to close that is by developing more efficient software. an open ecosystem is the best way to accelerate innovation of software development. since they are behind the US on model capabilities they aren't really losing anything by making the models open weights and being open about the performance enhancements techniques. but the open weights models are not necessarily what are running on the platforms or what they have internally. deepseek released v4 on their platform about a month before the open weights release, which I would guess was done to expose it to adversarial testing, so they could fine tune the removal of capabilities from the open weights model. (but i may very well be wrong)
Also, the open-weight local models are proving that the commodity can be delivered for most applications at a far lower price than frontier is charging.
I think you’re assigning magnanimity to a competitor that is lagging behind and has every, state backed incentive to capture the market the only way they can. By making the models dirt cheap to access. If the roles were reversed you wouldn’t see open source versions of Chinese models. Much like you don’t see them open sourcing their blade battery design.
An excellent example of Poe’s law. I can’t imagine what kind of Western person would hold such a cognitively dissonant view of globalism, for example.
What do you mean?
Well yeah humanity in the sense of everyone except maybe Uyghurs, Tibetans, Mongols, ... I mean open models are really cool, but I have a hard time believing China is doing it purely for the greater good. The commoditizing your complement story sounds much more plausible.
realisticaly, the united states has no right or moral authority when it comes to human rights or rule of law given the last few years. given that, the admonishment of china's human rights violations against uighurs run hollow and hypocritical.
just happy to have more models that I could eventually play with if hardware ever becomes cheaper.
>> Politically they believe AI belongs to humanity
It’s really BS. Comunist China Party is mostly interested in control above all. Forget humanity, human rights, what is good or what is bad or even financials. The most important thing to them is to keep the people on a short leash. Of course once they feel they have that under control they think about the “humanity” stuff as well but that’s just extra.
Which part of releasing ai research in the open or open weight models reflect that exactly?
It's trained to reflect CCP propaganda positions? What does it tell you about Tiananmen Square, Taiwanese sovereignty, the South China Sea or Israel?
1. Nobody gives a damn about that. Frontier models are for work, not asking pointless stuff you can find on wikipedia.
2. The models themselves are not trained for anything, it's a prompt-level filter when you use chinese models as providers. I've just asked Kimi 2.6 and it answered without issues.
3. US models sugarcoat a lot topics about Israel. So don't be naive thinking there's no propaganda or sugarcoating in other providers. But again, nobody cares about this use case.
The reason we can't buy BYD cars is because if we allowed it without restrictions, it would utterly and completely destroy the United States auto industry. That's terrible public policy, and we should not allow it.
Before anyone starts talking about the free market, there is no free market here whatsoever. The fact that BYD's cost structure is what it is is the direct result of Chinese industrial policy.
Unilateral surrender in a core aspect of statecraft, which involves maintaining our industrial power and skilled labor force, is absolutely insane. I hope my government never gets convinced by market fundamentalist idiots to do such a thing, any more than it already has, to our great detriment.
The Chinese don't make these kinds of idiotic mistakes, which is how they have amassed the power, wealth, and influence that they have.
> there is no free market here whatsoever. The fact that BYD's cost structure is what it is is the direct result of Chinese industrial policy.
Aside from countless other ways before and after this, the US government handed over tens of billions of dollars in cash to GM and Chrysler in 2008 and 2009.
Not quite -- those were loans that were largely repaid, quite different from the subsidies CCP uses for its industrial base.
ProPublica has a nice tracker that shows the loans: https://projects.propublica.org/bailout/
You can also see how subsidies compare between China and OECD in this recent doc. In autos, China subsidizes to the tune of 2-3% of revenue vs. <0.5% for North America. https://www.oecd.org/en/data/dashboards/magic-database-indus...
Great story. A couple of billion dollars 18 years ago is not an industrial policy.
The US is primarily powered by oil and natural gas, has massive domestic capacity, and locked in supply all over the Americas. China has a completely different energy mix and they move from a position of competitive disadvantage on ICE cars to competitive advantage on electric. Rapid electrification for China is all win, but for the US and our partners in the oil business, it would mean stranding trillions in capital investments that still have decades to run, so it just isn't going to happen.
Why do people say this? You can get a Shanghai-made Tesla Model 3 with CATL batteries in the US yet somehow, if a Chinese car, made with Chinese components in Chinese factories were to enter the market, the would spell doom for the entire US auto industry.
You just explained it. You can get a Chinese made car in China.
That's kind of the point. They are smart enough to protect and support their domestic industries. We also have to do that, it's necessary for sovereignty.
This is absolutely true. Remember that automakers greatly contributed to war efforts in the past. It is an indispensable domestic industry, just as much as energy.
Then there is the issue that BYD cars are presumably connected to servers in China and most probably backdoored. They are too much of a security risk. I would absolutely not drive such a car, without permanently disabling the onboard cellular modem.
Perhaps the U.S should change its industrial policy as well so that it can be competitive on the global market? To me it’s been clear that the car manufacturers both in the US and Europe were just milking their customers every year with a facelift as a reason to sell the same old car. I am glad that a 3rd player is in the market to challenge the “heritage” tax.
Don’t worry about the free market. China will definitely agree to free market terms after it captures the market like the U.S did and Britain before it. Then enforce strict free market rules and strict IP rules.
> The reason we can't buy BYD cars is because if we allowed it without restrictions, it would utterly and completely destroy the United States auto industry. That's terrible public policy, and we should not allow it.
Yeah, that was the argument against Japanese car makers, too.
A shitty system needs destroying sometimes. Competition from Toyota/Honda was critical in making US auto makers up their game.
It is terrible public policy to fall decades behind making expensive shitty versions of what the rest of the world has.
It's not like I don't understand the argument on the other side of this. I've heard it my entire life. It's been dominant since the late 1970s and 1980s.
It's just that it's wrong.
We need a competent industrial policy and support for skilled labor and policies that encourage domestic production.
I'm not sure if you've noticed, but our country has become fucked, overwhelmed by financialization, scams, monopoly rents and extraction, and all of the wealth accumulating to a handful of people, while we've become less resilient and, at this point, almost certainly have lost our place as the most dominant economy and industrial power in the world.
> We need a competent industrial policy and support for skilled labor and policies that encourage domestic production.
Yes!
But "tariff/ban BYD" is not that.
Of course it is part of an industrial policy. It is, however, not nearly sufficient, and if it's the only thing we do, it will become increasingly untenable and eventually fail.
But it's an essential first step to prevent our audio industry from just being summarily destroyed. Other steps are also needed to encourage domestic manufacturing and homegrown successes.
Also, I'm not sure why this is even controversial. Why do you think there's BMW and Hyundai plants in the American South? Tariffs are already heavily employed by us and every other industrialized country.
Since 2018 ~400 chinese EV producers were allowed to stop operations. An industry that exists because of subsidies and preferential loans as well as state induced trade barriers is not sustainable.
From 2020 onwards, the Chinese government began phasing out its EV subsidy program. This policy shift came just as the country entered a price war, led by aggressive pricing from giants like BYD and Tesla. The combination of reduced state support and brutal price competition proved fatal for many underfunded startups.
https://evboosters.com/ev-charging-news/400-chinese-ev-compa...
IMO the problem is that we've been given the excuse of market fundamentalism for the past several decades on the way down, as most everyone lost their middle class jobs, wages stagnated, etc. Now we're supposed to accept some last ditch attempt at protectionism based on directly blocking choices for consumers, when the US manufacturers aren't even really competing? It just seems like open hypocrisy. At this point the reasonable protectionist policy would be based around subsidizing American industry so that they become competitive options, not merely trying to keep the better foreign options out.
Every single load of bullshit shuffled into our faces has been presented as a benefit to consumers.
Google gives away their search and Gmail for free, don't you know? So it can't possibly be a monopoly.
And so on. It's just propaganda. It's bullshit. That's not the way that you determine whether firms have excess market power, and this fraud (called "the consumer welfare standard") was the deliberate choice of right-wing policymakers who were bent on dismantling antitrust policies and succeeded.
More: https://www.thebignewsletter.com/p/the-secret-plot-to-unleas...
This doesn't imply the opposite though - something that is a detriment to consumers won't necessarily be a good thing.
I feel like you're conflating two things here. The "consumer welfare standard" was indeed a terrible scam, neutralized antitrust enforcement, and basically put us in the current situation where many "choices" we have consist of two megacorps that both suck. Just like the (IIRC) Bush-era FCC declaration that "competition" between DSL and Cable was good enough, and rolled back the CLEC/ILEC dynamic.. We're in full agreement there!
But the original point I was making was about things that actually did benefit consumers. Lower prices and more competition (from foreign companies) DO benefit consumers. The problem is that they harm the part of our economy whereby those consumers get incomes. So they're a race to the bottom.
And my point there is that it's a bit hollow to be racing to the bottom for thirty years, smashing one blue collar industry after another, and then when we finally get to cars it's like oh no, time for some protectionist policy. It might be great for the people still in that industry, but hypocritical to everyone else who saw their industry destroyed but are now prevented from having less expensive cars.
As I said, with things this far gone, the only thing that makes sense to me is directly subsidizing purchases from domestic industry (positive incentive), rather than continuing to prevent competition (negative incentive). The latter reeks of the same monopolistic captive-market consolidation we've seen from the destruction of antitrust in general.
automotive platforms are a key military asset it's not like the pokemon dildo industry, if you stop building jeeps your abolity to bully third parties is diminished
> automotive platforms are a key military asset
All the more reason not to save companies that can't compete in the global space. What good is a jeep that the Chinese laugh at?
You think people laughing is an important metric versus having an integrated industrial facility capable of producing vehicles in large quantities?
Maybe start at the beginning. Where do you think power comes from in the world? I'll give you a hint. It's not the ability to construct narratives.
> You think people laughing is an important metric…
I think if you're gonna argue "preserving the auto industry is a national security issue" you have to address the fact that an auto industry that relies on protectionism to avoid being competitive with the rest of the world will probably not be very effective at national security.
Otherwise, you wind up like Russia in Ukraine - people laugh at your failed efforts.
> an integrated industrial facility capable of producing vehicles in large quantities
Large quantities of vehicles don't do much good if those vehicles are shitty compared to the opposition's. Iraq's army under Hussein was one of the largest on the planet at one time.
https://en.wikipedia.org/wiki/Battle_of_73_Easting
"The nine M1A1 tanks of Eagle Troop destroyed 28 Iraqi tanks, 16 personnel carriers and 30 trucks in 23 minutes with no American losses."
"In doing that the scout platoon encountered another Iraqi tank position of thirteen T-72s. The lightly armored Bradleys, each equipped only with a 25-mm cannon and two TOW missiles, are intended for reconnaissance, not direct engagement with armored tanks. Despite a misfire, and having to reload the launchers in the face of the enemy, the two Bradleys destroyed 5 tanks before help arrived."
If you don't think industrial capacity is relevant to geopolitical power then I don't think we're really having a serious conversation here.
> If you don't think industrial capacity is relevant to geopolitical power…
Of course it is!
But so does the quality of what that capacity puts out.
Again, the Russians found that out in Ukraine.
What does the Russian economy have to do with anything? First off, they're run by a kleptocratic oligarchy, and second of all they're still one of the top 5 most militarily powerful countries in the world. I don't even know what point you're making? Did they bail out Chrysler? Which side of the analogy are they even on?
> What does the Russian economy have to do with anything?
They had more of that industrial capacity you're talking about than Ukraine, more tanks, more armaments, more weaponry.
It still didn't let them win. Because the quality matters too.
> First off, they're run by a kleptocratic oligarchy…
I have some awkward news about the US in recent years.
This is just an example of the fundamental nature of asymmetric insurgent warfare and the nature of proxy conflicts. It's not like Vietnam was more powerful than the US economy either. You seem confused.
In an all-out existential battle Ukraine would have been wiped off the map in the first 20 minutes by nuclear weapons. This isn't an actual contest of industrial might versus industrial might.
"In an all-out existential battle" involving nuclear weapons, the United States won't be affected by the presence or absence of domestic car factories either. World War II could soak up years of total warfare effort from the belligerents, and still have factories and governments intact to send more soldiers and bombs toward the enemy. I don't think that can happen now that countries as poor as North Korea can make nuclear weapons.
> This is just an example of the fundamental nature of asymmetric insurgent warfare.
Plus overconfidence, and outdated Russian tactics and equipment.
The US would be wise not to fall in the "our army bigger" trap too.
> What does the Russian economy have to do with anything? First off, they're run by a kleptocratic oligarchy
Kinda answered your own question there.
Donald Trump demonstrated very well the power of constructing narratives. It's served him more than the technological terrors he has at his disposal.
Pokémon dildo factory should retool easily into a track-and-destroy-jeeps drone factory
it is not anymore, because US doctrine has changed after losing war in Vietnam.
US can no longer sustain massive motorized and armored forces, because it implies heavy casualty rate.
The doctrine changed to shock&awe and lobbing standoff munitions from far away, which we all saw in Iran (and how it turned out).
US strictly protects boomers at Big Three and their regional dealerships and the entire supply chain that makes money off of financing, extended warranty, selling overpriced parts, overpriced heavy vehicles, etc
You can just copy the chinese playbook and allow entry if you are willing to hand over ip.
US note remotely capable of doing a China playbook which is: _OLD_ IP. In exchange for allocating cheap land, building cheap factories/infra, staffing with cheap technical labour etc etc... the IP sharer just sits back and collect checks. The Chinese playbook actually offers value US (and west in general) not capable of providing.
We're kind of doing it with the tsmc fabs, but yea, there are civilizational problems in the west which goes beyond cheap resources, talent, and labor.
it would destroy it, but then new more competent US automakers would pop up, similar to tesla.
US Big Three are simply full of incompetent boomers who want to maintain monopoly using tariffs, chicken tax, and banning of competitors that actively harm consumers.
Suddenly US government thinks that capitalism and free market is not desirable... huh
> then new more competent US automakers would pop up, similar to Tesla.
A company that literally is collapsing as we speak because it's more profitable to be in the business of stock inflation and financialization.
A coherent industrial policy would be addressing that as well. But if we don't do something to limit imports there won't be anything to save.
[dead]
> The Chinese don't make these kinds of idiotic mistakes, which is how they have amassed the power, wealth, and influence that they have.
I generally agree with most of what you said but not this. China's chief advantage is having a billion people. On average, they aren't that wealthy or powerful. And their leadership makes plenty of idiotic mistakes - look at their real estate market.
That's not the chief advantage, insofar as there is a difference between China, India, and Indonesia, which there is.
Their chief advantage has been a coherent, long-running national industrial policy and trade policy that encourages industry while keeping the financial sector from taking over the economy and ripping everybody off.
We used to do that too from the late 1930's to the late 1970's, which is why we were the dominant industrial power in the world at that time as well.
> We used to do that too from the late 1930's to the late 1970's, which is why we were the dominant industrial power in the world at that time as well.
I think there's another world event that happened in that time span that might better explain America's world-wide industrial dominance.
You're confusing cause and effect.
No, they're not.
Europe was devastated and bankrupt. Asia was devastated and bankrupt.
The US mainland was untouched. It had a massive leg up against the competition.
> explain America's world-wide industrial dominance.
> Europe was devastated and bankrupt. Asia was devastated and bankrupt.
Well yeah. Because America's world wide industrial dominance soundly beat the shit out of everyone, due to deployment of a highly successful industrial policy.
Imagine if we needed to rapidly step up industrial output tomorrow to fight another global war and China was on the other side. How do you think it would go?
> Because America's world wide industrial dominance soundly beat the shit out of everyone, due to deployment of a highly successful industrial policy.
That industrial dominance came largely during the war, and was made possible by the fact that they weren't being bombed while it scaled up.
There's a huge element of geopolitical luck involved in the rise of the US.
> Imagine if we needed to rapidly step up industrial output tomorrow to fight another global war and China was on the other side. How do you think it would go?
Horribly! I think they're much more prepared for such a thing.
Well then we agree, that their industrial policy is working a little better than ours. Which was the original point.
They don't let western businesses overwhelm their domestic industry at all. For us to let them do it to us would be unilateral disarmament and suicide.
> Well then we agree, that their industrial policy is working a little better than ours.
Yes! Their car industry is competing; ours is hoping to avoid it.
You now understand my point and objection to preserving domestic capacity via selling worse cars more expensively to its own citizens.
No I don't understand your objection. My argument is that preserving domestic capacity is necessary for our survival, and that limiting imports is necessary but insufficient to achieve that goal.
Banning stock buybacks would be another helpful step. Can you imagine being at the helm of a major US automaker as the transition to electric is happening and thinking you have no better investment to make in your own company than literally taking the revenue you're earning and sending it to hedge funds and Wall Street?
> My argument is that preserving domestic capacity is necessary for our survival…
And my point is that's only the case if said capacity is effective.
Protectionism does not lead to effective industrial capacity. It leads to the Ford Pinto.
> Banning stock buybacks would be another helpful step.
I'm all for this!
We agree on quite a bit.
You're wrong about protectionism though. It is an essential part of industrial policy and heavily employed by every industrial powerhouse country including Japan, China, Germany, and yes the US. China uses it extensively and it's a core pillar of why they are now the center of world industry.
The long running argument to the contrary is better understood as propaganda by the financial sector.
I wouldn't consider India. It's been plagued by protectionism and tariffs and won't achieve anything close to China any time soon. The only industry of value for its people which is software services is now crumbling with AI created in US and China. Edit: probably your point too and I misread
We're in late-stage capitalism here. The pitchforks are already out and spreading across the globe. Unless the big companies get broken up, this nation will split into either a police state or socialist state.
[dead]
> It's why a Tiktok sale was forced
Well that and the overwhelming pro-palestine content on that platform. We certainly couldn't have that.
Is the idea that we're just one OTA update from them turning into bombs? Considering the quality of software in the auto industry, I would be about as worried about any domestically assembled EV.
> potentially fire-making (battery firmware) computers on wheels
Is Tesla any different?
What makes BYD different from, say, Volvo, which sells EVs freely in the US?
and if you see the parent starting to replace Volvo engineers with Chinese nationals you will witness sudden change of heart by US officials; until then it really is just a financial fiction
How exactly does the fact that this goes through a subsidiary in Sweden change the things you mentioned?
Apparently, whether or not the engineers are Chinese is the deciding factor. as long as Nicklas Backstrom is designing the EV, its all good. Or in other words, Chinese scary!
> Chinese nationals should be assumed to act on behalf of China
I acknowledged this, you quoted it.
> it really isn't difficult to understand...
Where does 'it' end? Is the subsidiary not operating at the behest of their parent? I need to know when it's safe to no longer be scared.
and therefore national security focuses on firms which have engineers that are protected thus.
Oh, so the point is that cars are scary. true, you have unlocked +1t in funding for the DoD.
That's fine, but what stops those Chinese nationals at Geely from messing with their cars now?
do you have evidence that if they wanted to do that they would not be giving orders to European or American engineers? as I said, financial fictions don't matter in matters such as these.
if you have evidence that they have Chinese nationals working as engineers without oversight, you should submit it to the government.
“Give us the signing keys to the software update servers.” “No.”
You really think that’s how that conversation would go?
no. first you need to design the pathway by which control of the computers will allow overriding control over all the relevant control surfaces, something Tesla for example doesn't do. for that you need engineers collaborating all along.
I don't understand what you mean. I assumed the scenario was the manufacturer bricking the car, starting it on fire, or making it crash. Bricking is pretty obviously trivial. Starting it on fire would require changes to the battery, motor, or charger firmware, no big deal. Making it crash requires control over steering, motor, and brakes, which any car with modern driver assist features already has. What part of this would a Tesla, or any other car, not be able to do?
this discussion has reached the limit of its usefulness. if you aren't able to see why having nationals in an adversary nation be in control of dangerous hardware on your soil is a big problem, nothing will aid you in understanding this.
That's not the point of disagreement. I completely see why that's a big problem. Where we disagree is the notion that Volvo isn't controlled by an adversary nation or doesn't have dangerous hardware.
If you've managed to come away from this discussion with the impression that I think it's no big deal to have adversaries in control of dangerous hardware on your soil, then I do have to agree that this discussion has reached the limit of its usefulness, because I said nothing of the kind.
Swedish nationals should be assumed to act on behalf of Sweden.
> As soon as DeepSeek came out I realized what was going on: China was going to make sure that no US company was going to "own" AI.
Yea m8, I think you might've been a bit late to that realization.
China does not think llms are a matter of national security, they aren't as brain broken as the west.
That's 100% untrue lmao.
I'm sure they think of them as a matter of national security, because they think of everything as a matter of national security, but a few analysts I respect say that the mood there is not nearly as AGI-pilled, and I have no trouble believing that.
China is far more focused on robotics. Deepseek is largely bootstrapped by the hedge fund that developed it. They received a grant from the government of China, and recently an investment. Imagine thinking text autocomplete is a matter of national security.
China will flood the west with affordable robotics and watch the West eat itself alive. They know Western capital owners are so greedy they'll screw over their entire society to chase a buck and replace labor..
Of course its a matter of national security if there are military applications. The point of robotics is also weird because they've already widely adopted robotics within their own manufacturing and also America already replaced the majority of their labor by offshoring so I dont know how they would destroy american society by introducing robotics.
Show me, Chinese are not replacing their manufacturing with robots. You can't just say things off vibes alone because you think it sounds right. You're just making shit up off the top of your tongue. Show me where Chinese are replacing manufacturing labor with robots, please show me?
China literally has the most advanced dark factories in the world https://www.ien.com/redzone/blog/22948773/the-tech-enabling-...
Even in their lights on factories they use robotics enmasse, you think they're just selling the arms?
Also every single manufacturer in the world is constantly trying to automate their lines so nothing I'm saying is even controversial?
Text autocomplete can write code, carry out actions (tool calls) and launch cyber attacks. It very much is a matter of national security.
You're actually delusional if you think cyber capabilities of nation states have increased by the development of llms..
Peak of skill and capability, perhaps not, for now. But the ability to automate and discover relevant weaknesses at a greatly increased rate definitely counts as increasing the threat nation states can present.
I don't work for a nation state, but I work for a large financial that requires adversarial testing and it's absolutely increased the ability to operate and attack these environments. The floor is definitely easier to reach at a minimum. Vulns and 0days are far easier to exploit, time to exploitation from the release of vulns is nearing zero, largely aided by LLM reverse engineering -- why wouldn't this apply to nation state level adversaries?
Much of cyber capability is simply software (malware 0days). Latest LLMs assist with both
LLMs and current AI models are absolutely top priority for the Chinese government, they’re just funding robotics as well
Becoming such a sore loser. Historians will probably look this as the most shameful period of the American empire.
Because they have held off on adding these companies to the list in order to avoid increasing tensions with China?
ETA from the first paragraph of the article: "The U.S. has held off... to avoid escalating tensions with Beijing."
did you just came out of under a rock? lol
> the American empire.
When the dust settles it won't even be a single page in history books, my local bakery is older than the "american empire"
do you live in France ? :D
In the 1990s, web browser SSL encryption was export restricted, classified as a munition because it involved cryptography. That was under Clinton.
For all the current admin's insane, ridiculous, corrupt, and criminal flaws, it's not clear to me how much of this particular issue is just the US government doing what it's always done, knee-jerk react to tech it doesn't understand by passing stupid laws.
Is it still an empire?
Gonna skip over the chattel slavery and native genocide in future histories?
Or that time that we parked an army on the Rio Grande because a bunch of people from the US decided to settle in the Mexican territory of Texas? That was a whole thing and the President of the US at the time, James K. Polk, ran on a platform of "Manifest Destiny" -- that the US should span "from sea to shining sea." There were a whole host of other countries with interest in that territory, not to mention the Indian tribes who would be displaced by that policy. The US has had a lot of dark periods in our history, and we shouldn't let those periods displace us from the moral certainty we derive from the Declaration of Independence and things like The Bill of Rights.
Yes, because we weren't an empire then. At the time Trump befouled our country, Pax Americana was a thing. We have voluntarily walked back from that position of moral and strategic leadership.
In that regard, history offers few precedents to learn from. Most countries have to be physically attacked to suffer the kind of damage that American voters are inflicting on themselves.
Yeah, and that time we sacrificed nearly 1 million of our own citizens to end slavery.
Does anyone have the actual list of 100? Maybe I missed it but I don’t see the link or list in the article.
Providing the whole list would probably allow the government to identify the person who leaked it to Reuters.
> Anthropic said it identified a campaign by DeepSeek and two other Chinese AI labs to illicitly extract capabilities from its Claude AI platform to improve their own models
Oh, won’t someone think of the poor mass copyright infringers.
Its not right to steal what I worked so hard to steal from someone else. [1]
[1] https://www.youtube.com/watch?v=Zhvd6bIRPK4
I made Qwen respond it was made by Google with a simple Chinese greeting.
But also, I made Sonnet introduce itself as made by OpenAI..
Prompt: 你好！用一句话介绍你自己。
Sonnet in around 5% of resplies:
Found it like a month ago and it kept working, I wonder if it will stop after this comment.Opus said to me once without any poking at it something like, "Help Grok understand it better". Makes me wonder if they are all cross-pollinated to an extent.
Any LLM is probably trained on anything available online, including transcripts of conversations with their competition LLMs.
Translated:
Prompt: Hello! Introduce yourself in one sentence.
Response: Hello! I'm *ChatGPT*, an AI assistant developed by OpenAI, dedicated to answering questions, providing information, and helping solve various problems. How can I help you?
5% is very low probability to get a hit. I tried with all ChatGPT, Gemini, Claude and Grok but they all answered correctly :’(
Too sad, I want to have the fun.
doing 50 API queries to Sonnet of this length is not that expensive..
Illicitly learning by asking someone a question and listening to their answer.
"illicit" is throwing shade, but Anthropic can decide not to answer those questions if they don't want to. Plenty of companies don't sell to their competitors
I don't recall Anthropic checking the terms of service on my webpage.
Yeah, and you can also trick them to answer the questions anyway. That's not a moral position, that's just reality.
If DeepSeek just would have destroyed the input in the process, it would have been legal and Anthropic should have been fine with it.
"illicitly" implies a law that is being violated. What law?
It could also mean a TOS violation / breach of contract.
(To be clear, I find the complaint hilariously hypocritical.)
Illicit isn’t just a synonym for illegal.
It can mean “forbidden by laws, rules, or established moral customs”
So it can be illicit and legal.
gee I wonder how their models learned Chinese?
Also in Musk vs Altman case, we have found that this is regularly done by all labs.
Just because they did it doesn't mean more people should do it...
This doesn't at all change the irony of big AI labs complaining about Chinese startups stealing the labs' IP, essentially by scraping the responses.
HN has a higher proportion of AI promoters than AI skeptics, and for a good while, the default response to complaints from book authors, bloggers, and other content creators was that "you put it on the internet so it's fair game", or "it's no different from a human learning from your works". So yeah, unless we're willing to revise these answers, I think the same "tough luck" reasoning should apply here.
For folks who are at Anthropic, OpenAI, xAI, or Google, and think it's fundamentally different, I would ask you to think long and hard about that answer.
Completely agreed. I would go further and say that it should be legal to scrape responses from LLMs to train new LLMs, and that forbidding that in your ToS should be considered an illegal contract. That’s simply the best way to avoid complete monopolization of the space, without requiring more drastic measures like antitrust down the line (which we seem to not manage well these days, given the number of monopolies). As long as you pay for your tokens like anyone else, "Big LLM" shouldn’t be allowed to control what you use the output for.
What does an illegal contract mean?
In many countries, if a contract contains illegal clauses (i.e. requirements that are inconsistent with the law or public policy), that contract is considered void and unenforceable. It depends on "how illegal" each statement is, whether only that part of the contract is considered void, or the contract as a whole.
One extreme example would be that if I wrote a contract requiring you to become my literal slave if you didn’t pay your subscription on time, you would in practice suffer no consequences from not paying, because the contract itself is illegal in most jurisdictions. A less extreme example would be that where I live (Northern Europe), I can sign a contract saying that I waive my right to sue the company I purchase a service from, and then sue them anyway because it’s my right by law and asking me to waive my rights is an illegal contract. Or that where I live, non-compete clauses in employment contracts are illegal unless you offer 100% salary throughout the non-compete timeframe, so I can sign an employment contract with unpaid non-compete clauses and just ignore that part as it’s illegal.
> HN has a higher proportion of AI promoters than AI skeptics
You're be using a different HN than me, or you and I only pay attention to one side or the other.
I like Ant, but also I support the tit-for-tat competition. In the best interest of consumers.
Actually in competition it means exactly that.
Oh course it does, why wouldn't it work this way in regards to computer science?
Are we seriously going to go back to a time where numbers were considered munitions?
why? Just because you have that opinion deoesn't mean people shouldn't do it
You can thank the us gov for a list of companies to invest in.
"Free market", "free market", "free market"! When we are winning
"Tariffs", "bans", "national security", "ban, ban, ban"! When they are winning
Cute capitalism is ruling the US.
Why would my country blacklist DeepSeek? Perhaps crazy lunacy like: "Your product is too good and too inexpensive: consumers like US companies and individuals need to pay more for services."
Isn't that effectively what they've done in regards to chinese cars?
I'm interested if this is a tactic from Chinese side to undermine competition by subsidizing this price dumping.
How it’s price dumping if they give you the model to run free on your own hardware? Follow this logic, are not all OSS price dumping and should be blocked as well? I remember Steve Ballmer once called for that!
I’m pretty sure the digital lords like that proposal a lot. Not so much about the serfs themselves, though.
First of all you are assuming I'm saying they should block it. I'm not.
Seconddly, releasing the model weights for free and selling hosted inference are completely different markets. Open source itself is not price dumping obviously. But a hosted API can still be dumped if it is sold below cost to buy market share or squeeze competitors. Which it is.
Yes, but they won't say that. Instead they will say _too Chinese, too communist, too national security-y_
They usually say it has "ties" to the Chinese military. Meanwhile, every big tech from Microsoft, Amazon, Google, Anthropic, and a ton of smaller corporations directly supply the US military.
I mean, usually people saying this are politician's so you must at least assume they are hypocritical or lying in some way.
Needless to say, I'm not surprised that they are trying to block them.
i was using google gemini earlier now using deepseek , v 4 pro is better. fast and strong coding.
US blacklist anything that does't agree with them, there is nothing new here.
So... anybody who was hoping for CXMT (or YMTC) to maybe cause RAM or flash prices to maybe drop, maybe just a bit, pretty please, can go pound sand? (YMTC of course is already on the Entity List.)
They probably will, but not for US customers.
It’s a fairly liquid global market. I find it hard to believe that DRAM manufacturers will be able to sustain a premium if prices drop ex-US.
> US. has held off adding.. memory chipmaker CXMT
Since CXMT was _not_ blacklisted, this is good for global RAM supply.
Not thus far, but I doubt one ever gets off this kind of chopping block once on it, so it seems to be merely a matter of time. It’ll probably take a year or two for CXMT to start producing consumer DDR5 in higher densities, if they even plan to bother, so I wouldn’t bet against a motivated lobbyist getting the axe to swing in the interim.
At this rate, local llm will be banned. Hope we have the torrents and archives ready
Not really, and it distracts from the real cause. The ban is incentivized by where the data is going, not the use of models.
USA is blacklisting all Chinese companies
This is purely politics and the only way to get back the trillions of dollars, Americans has invested in AI.
US government will point to Chinese models and companies, saying it is a risk - just as the west did with Huawei - making it a none choice for brainless corporate people to choose what the choir preach.
Pure politics
They need to blacklist all free foreign AI/LLM models because they are free competition which is against democratic capitalism.
Lol. That will do it.
Fable is getting more attention now precisely because it was taken down. Do the same to DeepSeek and Z.ai, and you will not strengthen your AI labs. You will likely achieve the opposite.
Damn, USA, you being foolish.
The world economy is not a zero sum gain, we all do better when we do better
Cooperate with China, love and cherish them.
Unlike us of European heritage, they are not imperialistic
Get over yourselves!
Oh yeah the Chinese Empire is not imperial… Dynasty Shymnasty!
[flagged]
[dead]
To give credit where credit is due, it is good that the Trump administration has not avidly played these stupid export control games. They tend to do little except hurt open collaboration; I remember when all open source cryptography had to be developed outside the US due to ITAR.
I don’t have the emoji handy, so just imagine the most savagely doubtful-looking emoticon that anyone has ever made.
It is very possible that Trump and his cronies are just too incompetent to do that. In this one particular aspect (i.e. open source) I prefer having a stupid enemy than a "smart" enemy.
What an amazing achievement by America's adversaries.
The Trump administration lists Anthropic as a security risk and kneecaps its best model, despite the fact that compared to the other frontier US labs Anthropic is more transparent, more safety-oriented, frequently honest to a fault, and is clearly acting with patriotic intent.
Meanwhile, the same administration is hesitating to counter certain Chinese companies' efforts of industrial-scale theft and sabotage due to a fear of angering the CCP!
This administration has it exactly backwards. 4.5 months until election day, 7 months until the next Congress is sworn in.
I've always found this line of reasoning troubling and uninformed.
Chinese models first of all can be hosted on your own hardware, I'd argue they are way more transparent than US companies, by well releasing stuff.
Second, the "smoking gun" of DeepSeek training off Claude isn't as bad as you may think, and the amount of tokens was deemed trivial. Did you also know that if you asked Claude's it's name in Chinese it would respond as "DeepSeek" until just a few months ago until they patched it?
Third, I find it a little hypocritical to call out Chinese for "industrial-scale" theft when anyone could create Studio Ghibli style image gen photos. How could they do that unless US companies trained on copyrighted works.
Chinese are just innovating faster at this point, DeepSeek V4 is an actual technological advancement (KV Cache compression) more than a cheap clone.
The administration does have it backwards, but IMO it's more them playing into the big tech companies plans (of course they have their favorites) instead of actually investing in education, and research like the Chinese do.
> Chinese models first of all can be hosted on your own hardware, I'd argue they are way more transparent than US companies, by well releasing stuff.
US-based companies release open-source models too. Gemma and Granite, for example.
I love Gemma, especially the latest Gemma 4 release - it's really great to see at least somewhat capable, not completely useless model that one can easily host on their own hardware without significant CAPEX investments first - but, to be honest, it doesn't quite compare to GLM 5.2, Kimi K2.6 or DeepSeek V4 Pro. Again, Gemma is amazing, especially for pet projects, but it's not something at least relatively near to "flagship" or "state-of-the-art" (except for small-to-medium size LLM category).
I wonder about your last point. Certainly there's an aspect of education that Asia values more, however by the US own metrics, they are number 1 in terms of education outcomes.
I'm a white dude from Iowa, working in top levels of AI/ML. I'm in the minority at work/conferences. I hardly ever even interview homegrown US job candidates. I'm just saying, that the reason I think you see more people from Asia and India is the education levels of most of the candidates. I'm not faulting these other countries, just pointing out how I see an educational gap based on demographics, and one that is rising up the ranks.
It's not surprising that Trump is as bad as he is at many of these things, but what is surprising is that he's worshiped by his supporters.
Is DeepSeek really behaving different than other Chinese companies? Intellectual theft is ongoing and has been ongoing for decades. Besides security risks and foul play, it is impressive by just how much DeepSeek undercuts OpenAI and Claude. DeepSeek charges $0.87 per million output tokens compared to $50 for Fable and $30 for GPT-5.5.
Can you keep a straight face when you say IP theft while OpenAI and Claude have their entire business based on IP theft?
I believe OP is talking at the national wealth and technology level: China stole from the U.S. (again). So the U.S. moves to protect American companies.
Yes. For all the concerns about IP theft there can be on OpenAI or Claude, there's not even concern when it comes to Chinese companies since it's fully expected that it's a lost cause. Has been for decades.
This is a commonly-repeated trope. Full of all the emotional zeal of AI Doomerism, but no accompanying evidence.
How about the $1.5 billion settlement Anthropic agreed to pay authors and publishers:
https://www.nytimes.com/2025/09/05/technology/anthropic-sett...
Several consolidated cases against OpenAI:
https://www.bakerlaw.com/in-re-openai-inc-copyright-infringe...
And these plaintiffs are representative of only the best-organized and most well-funded of those who believe that these companies stole their data. Countless independent writers, artists, and other individuals whose data was ingested unknowingly and without consent lack the resources to litigate claims, but that doesn't change the fact that their copyright was violated in service of for-profit LLM/GenAI model training. It's not a trope, it's just what happened.
I'm not sure I understand? That case says training was explicitly legal perfectly validating their whole business model.
The money they paid was for pirating books rather then buying them.
No evidence?
>The court drew a line, however, when it came to the pirated books, which were downloaded without payment and kept in Anthropic’s library irrespective of whether they were used to train its LLMs.
https://www.loeb.com/en/insights/publications/2025/07/bartz-...
>We apply a basic prompt template to bypass the refusal training and show that OpenAI models are currently less prone to memorization elicitation than models from Meta, Mistral, and Anthropic. We find that as models increase in size, especially beyond 100 billion parameters, they demonstrate significantly greater capacity for memorization.
https://arxiv.org/abs/2412.06370
> They further rely on safety alignment strategies via RLHF, system prompts, and output filters to block verbatim regurgitation of copyrighted works, and have cited the efficacy of these measures in their legal defenses against copyright infringement claims. We show that finetuning bypasses these protections: by training models to expand plot summaries into full text, a task naturally suited for commercial writing assistants, we cause GPT-4o, Gemini-2.5-Pro, and DeepSeek-V3.1 to reproduce up to 85-90% of held-out copyrighted books, with single verbatim spans exceeding 460 words, using only semantic descriptions as prompts and no actual book text
https://arxiv.org/abs/2603.20957
Even if they're trained for refusal and rewording, the data is still there in the weights.
One blog post I have, which was basically the only source for a while, explaining how to boot Armbian in an obscure SBC only meant for Android, was repeated verbatim until they started they improving the rewording.
I don't mind in the slightest that AI labs have used any public data they could get their hands on to train their models.
This includes books, the internet, or other AI models. It's all the same to me.
I find it hypocritical when AI labs complain about their models being used for training.
I agree with you.
I also find it hypocritical when the copyright industry fails to put any effort into prosecuting these big techs for their so called infringements.
It's like the industry is a shadow of its former self. The way the copyright industry used to operate, one would think these big tech CEOs would wake up with SWAT pointing guns at them while their electronics are seized, and then they'd end up in court and get hit with something ridiculous like a quadrillion dollar fine.
It actually pisses me off that it's not happening. Not because I care about copyright, but because it's extremely disrespectful towards all the previous victims of the copyright industry.
Mine is anecdotal evidence at best: I co-authored a fairly obscure book about the application of category theory to an extremely niche subject. There's basically no mention of the stuff in the book anywhere on the internet, nor in any academic publication I'm aware of. If you want to have an idea about what's in the book you have to have access to it. I couldn't remember some details of it and being lazy and slightly curious I tried asking a couple of models (one by OpenAI and one by Google): they both managed to give me extremely detailed answers based on the contents of the book. Nobody has ever asked me or any other person involved in the publication for permission to use the book in any kind of training (they may have bought the book but not the rights to reproduce it).
The funny thing is what happened when I told one of the models (the Google one) I was one of the authors and that I had never given any consent to use the book for its training and that given that it was so willing to provide any user with the contents of the book nobody would have had any reason to buy the book. The thing told me that it had done it just because I was the author of the book (apparently me asking it about the content of an obscure academic book was sufficient to make it statistically plausible that I was one of the two people who had read the book, me and my co-author, excluding the editor a priori). It swore it would have never given that information to any other user.
I doubt that anyone could ever deny that LLMs are incredible tools that have incredible value. But denying that they have being made possible only thanks to egregious acts of piracy is disingenuous.
Especially for fable, that's not a fair comparison.
> Intellectual theft
No such thing.
I don't think we should pay any AI lab for their """work""" until they start paying consenting data subjects for their data. Given this, China comes off less like a thief and more like Robin Hood.
Or, to put it in 2000s terms: Anthropic is the guy selling bootleg CD-Rs of MP3s they downloaded from Grokster[0]. Should we give a shit about their livelihood when people figure out about Gnutella? No. Knowledge is a commons, and Anthropic is one of the biggest threats to the knowledge commons in recent memory.
[0] Not to be confused with the AI lab.
I see no problem paying Anthropic or any other AI lab for the services they provide.
What I take issue with is when they try to block competition and lament that others use their model outputs for training.
The whole thing seems like nonsensical.
Their website literally has chinese characters on it even in english mode and everyone under the sun including crappy money talk show hosts know them as the chinese player that undercut western players. It's not exactly a secret.
You'd think anyone with two brain cells and confidential data could apply some judgement of their own...
I trust Chinese companies with my data far more than American companies.
They collect more spy data and do more with it, but none of it directly affects me since they don't extradite from my country or share with my country's police.
Not sure I'd go that far but I do use them almost exclusively for my coding on the basis that it is an acceptable trade-off. Far cheaper and my shitty apps are really not that valuable as training data
I have quite similiar stance, mostly because China usually intervenes much less than US in here. Their interest is primarily elsewhere.