News Speculation mounts that Musk will raise tens of billions for AI supercomputer with 1 million GPUs: Report

His investors will lose all their money, and then in 2 years he'll have to find new people to take money from for his next bigger-er computer.

At some point, the number of people willing to lose everything in AI is going to run out - and then the next big money-burning fad will pop up out of nowhere.

Then, after everybody has already realized AI is a bottomless money pit, Facebook will finally get around to renaming their company from Meta to AIVerse, or some junk.
 
Elon Musk: Artificial Intelligence is our biggest existential threat. ... AI is summoning the demon. Holy water will not save you.

DWave Founder Gordie Rose (A Tip of the AI Spear): When you do this, beware. Because you think - just like the guys in the stories - that when you do this: you're going to put that little guy in a pentagram and you're going to wave your holy water at it, and by God it's going to do everything you say and not one thing more. But it never works out that way. ... The word demon doesn't capture the essence of what is happening here: ... Through AI, Lovecraftian Old Ones are being summoned in the background and no one is paying attention and if we’re not careful it’s going to wipe us all out.
 
The smell of 200,000 cooked power connectors in the morning...

Seriously, won't all the GPUs be obsolete before the project even goes online? Won't it need continual re-investment with that $50-$60 billion being just (!) a down payment?
 
Last edited:
Roughly estimated, we are talking about 1 GW of energy demand - requiring a massive nuclear power plant. It would take at least a decade to plan, build, fuel and commission such a huge power plant.
So, is that idea just a dream or is it realistic? After all, those sweet B200 GPUs will be hopelessly outdated and outperformed in 10 years from now.
The question is, whether such a brute force attack really is the best option we have. Another possibility would be to optimze workflows and spend a few of those billions in hiring the world's best engineers and top-level experts and scientists who could tweak the code and re-think the entire compute-, training- and inference-infrasctructure so that it runs massively more effiicient.
 
Last edited:
Buying that much Nvidia hardware for this purpose would be idiotic. There's already a few different purpose-built pieces of hardware showing impressive numbers - on cheaper silicon, with lower power draw, while outputting less heat.

I. Di. O. Tic.
 
  • Like
Reactions: ottonis