[swiftcoder]

There's a core problem this analysis overlooks: OpenAI and Anthropic don't have a moat. The Chinese labs are consistently able to replicate their LLM capabilities a few months after the fact, and then release open-weight models a few months later...

The only way for "Big AI" to become a thing is for them to establish a moat, and right now the only path to that appears to be achieving regulatory capture in the US, which is a fickle and unstable state of affairs.

[aniceperson]

You are not allowed to use any model from Chinese labs at American companies. Even from other companies like Mistral it is hard to get access, since enterprises require their own instances on vertex/bedrock/azure, and to run alternatives you have to go through multiple layers of management. In practice, no employee will proactively advocate for alternative models and the oligopoly is written in the wall, specially since the investments are circular anyway.

[overflow897]

For now. When they cost 5x less and are of comparable quality that may change very quickly. Cost cutting applies everywhere.

[lenkite]

Well, some Chinese AI companies like Zhipu AI are actually increasing price. Let us see the trend in 2027 when the money runs out there too.

[zombot]

And how, pray tell, will that make for trillions of dollars return on investments?

[ShinyLeftPad]

Big AI is a thing. Insane money and construction projects makes the government invested in it one way or the other. It can't fail or it'll cause economic shock. It's a play to normalize it and make everyone depend on it before people question how they trained it all on their data and now drive them out of their jobs.

If they ask for a moat they will get it, the government can simply require licensing for hardware required to run any model comparable to the cloud ones (for "national security"). good local models will only be used by .01% who are crazy neckbeards running servers in their basement and even then they will suck compared to cloud ones.

[swiftcoder]

> If they ask for a moat they will get it, the government can simply require licensing for hardware required to run any model comparable to the cloud ones (for "national security")

This is the "moat by US regulatory capture" I mentioned above, but it only extends to the US, and pretty much cedes the rest of the world to the Chinese labs. It only makes sense if the US is going to become an isolationist power

[ShinyLeftPad]

Why do you assume the rest of the world is going to be under PRC labs? First it assumes the rest if the world will go as nuts on LLMs as US which if they see the same future as the author they might not. But even if they do, if PRC could do it, Europe could do it.

[swiftcoder]

> But even if they do, if PRC could do it, Europe could do it.

Sure, my point is that in this scenario, the US labs would lose out on all the overseas business, regardless of whoever else is training models for the rest of the world

[ShinyLeftPad]

True. But they already lose out, in many places they cannot sell for national security. I think it will only tighten. Government contracts probably will make up difference

[dmurvihill]

That's why they're so keen to build out all these data centers; the massive capex is their only conceivable moat. Whether all those additional parameters actually make a competitive difference is still an open question.

[warumdarum]

Moat by having the investors in a bubble in to deep. Thanks, i never looked at it that way. It can not fail, because IT CAN NOT FAIL or we are done for

[smrtinsert]

We're in the Pets.com era of AI for sure. There is 0 guarantee the current providers will be around in 5-10 years. I keep thinking they will have to lean in to government protection to do so.

[toasty228]

If your government weaponize AI against you it doesn't matter if the AI is American, Chinese or Uzbek

[trashface]

Memory might be a moat. A cynical view is this is why openAI is buying up all the capacity for HBM. If they have all the memory then we plebs won't be able to afford hardware to run local models. So we instead we have to rent their models basically just to get access to the memory.

[AnimalMuppet]

I doubt it. Memory manufacturers are going to ramp up, given the demand. Can AI make enough money to buy up the increased capacity forever? Whereas each user just needs enough money to buy up the memory that they need for their use. AI needs to invest billions, small users have to invest thousands. And the payback is going to be there for the small guys, so that even if they have to borrow to get the memory, they will quickly be able to repay it.

This is not a game that big AI is going to be able to win.

[bilegeek]

> Memory manufacturers are going to ramp up

Given the prior boom-bust cycles, they are essentially slow-walking expansion to whatever plans they had prior to the AI boom. Don't expect things to let up for awhile, even if AI crashes tomorrow they'll keep supply controlled as long as they can.

[ThrowawayR2]

> "release open-weight models a few months later"

They are going to want to profit from their resources expended eventually and when they do, that will end.