Correction: Can you yet?

Anthropic and OpenAI: Dear Uncle Sam, it seems laptops became too powerful, can you please do something

Uncle Sam: For national security reasons starting from now on every purchase of GPU model with higher than X Petaflops will need written permission by the US president

Anthropic and OpenAI: Look poor citizens, we are willing to share our capacity with you in limited form, by using our LLM you can avoid spending 35 years in the waiting list to buy a GPU, by the way, to simplify pricing, here is our new pricing with 5700% increase. Enjoy

It's the age old "Help me, politicians. You're my only hope".

Anthropic, Open AI & Co. realized at some point that if they can't make money with barely any competition they sure as hell won't if the market is flooded. So here they are slamming the door behind them.

> Anthropic, Open AI & Co. realized at some point that if they can't make money with barely any competition they sure as hell won't if the market is flooded. So here they are slamming the door behind them.

Jokes on them, there won't be new entrants not because the door is shut but because it doesn't actually make money. The whole scheme is propped up by illusions to grift the investors, fewer competition only breaks the illusion. But I guess those folks aren't the types to understand "rising tide lifts all boats", or in this case, rising sewage buoys all rats.

Don't forget: "Questioning our results on the awesomeness of the current administration is treason."

China: look at these GPUs we reverse engineered, do you want an RTX 5090 for half the price Nvidia sells them for?

finally, our GPU 'Come and take it' flags will be understood

Extrapolating the progress in both hardware and model efficiency, that will take decades

Depends on the point of view, I suppose.

Powerful enough to shock someone in 2010 with a wikipedia chat bot? Possibly.

Powerful enough to shock jaded HN commentators right now? Possibly not.

its not about whether you can shock anyone, if anyone is driving cars outside, you can't say you have SOTA-horse to go from A to B.

When models are good, expectations are adjusted accordingly to deliver things on par with the whole industry, you can't just say, I have built my own Intel Pentium II, now I will try to use it to compile Electron App and run 3DS Max there

You don't need SOTA-level LLMs to create value with AI. Hell, you can build good solutions with a simple small finetuned models.

> When models are good, expectations are adjusted accordingly to deliver things on par with the whole industry, you can't just say, I have built my own Intel Pentium II, now I will try to use it to compile Electron App and run 3DS Max there.

I know you are taking your analogy to its breaking point but it really depends on what you are doing. I know people that use 10+-year old thinkpads and they do just fine.

> You don't need SOTA-level LLMs to create value with AI.

person said "insanely powerful". expectations of "insanely powerful" is way beyond the graph of an individual using consumer-grade hardware.

I agree with you but to be fair I also think that even 10 year old hardware is insanely powerful by any reasonable metric. Modern computers are truly ridiculous when considered in historical context.

Maybe not in absolute terms. But if there is no access to the actual SOTA models, it's as if they don't exist.