The AI company, which is planning to go public this year, is buying Caterpillar generators that equipment supplier OnePWR had originally ordered, according to a Financial Times report. The generators are capable of producing up to 1.7 gigawatts of electricity.
Historically, Anthropic has preferred to rent rather than buy. The company disclosed in its S-1 filing in May that it is paying SpaceX subsidiary xAI $1.25 billion per month through May 2029 to use xAI’s computing power.
>build chips that require a tiny amount of electricity ?
That has been one approach of many possibilities.
Fortunately people around the world had been working on this for decades to great success, without even needing to think about superconductivity. Otherwise today's AI wouldn't even be possible yet.
That's pretty much what today's AI consists of, super-efficient chips combined with a willingness to use as much quantity and energy as it requires to get some brains out of it, regardless of whether it is a waste or not. With no other consideration than the scale of other peoples' money at stake, they can afford all the waste or anything else that costs money, as long as it can accelerate the venture. Like no companies ever before.
I can only imagine that there's somebody out there that could figure out how to do the same level or better AI than there is now but only using a fraction of the energy. Give them $3.4 billion and they could probably come out under budget.
Semiconductor engineers are still working on more efficient chips, but are there really appropriately massive numbers of experts working on way more efficient code, even if it requires a completely different AI paradigm?
There's got to be more money now, and lower-hanging, in getting the code to use less energy than getting the chips to do so.
The AI company, which is planning to go public this year, is buying Caterpillar generators that equipment supplier OnePWR had originally ordered, according to a Financial Times report. The generators are capable of producing up to 1.7 gigawatts of electricity.
Historically, Anthropic has preferred to rent rather than buy. The company disclosed in its S-1 filing in May that it is paying SpaceX subsidiary xAI $1.25 billion per month through May 2029 to use xAI’s computing power.
Why dont they use all that intelligence to find super conductive materials and build chips that require a tiny amount of electricity ?
>build chips that require a tiny amount of electricity ?
That has been one approach of many possibilities.
Fortunately people around the world had been working on this for decades to great success, without even needing to think about superconductivity. Otherwise today's AI wouldn't even be possible yet.
That's pretty much what today's AI consists of, super-efficient chips combined with a willingness to use as much quantity and energy as it requires to get some brains out of it, regardless of whether it is a waste or not. With no other consideration than the scale of other peoples' money at stake, they can afford all the waste or anything else that costs money, as long as it can accelerate the venture. Like no companies ever before.
I can only imagine that there's somebody out there that could figure out how to do the same level or better AI than there is now but only using a fraction of the energy. Give them $3.4 billion and they could probably come out under budget.
Semiconductor engineers are still working on more efficient chips, but are there really appropriately massive numbers of experts working on way more efficient code, even if it requires a completely different AI paradigm?
There's got to be more money now, and lower-hanging, in getting the code to use less energy than getting the chips to do so.