Whether its justified or not is outside my wheelhouse. There's too many "it depends" involved that, best case, only people working in the field can answer, worst case, no one can answer right now.
A huge increase in fuel efficiency is great for the economy, horrible for fuel companies
Daya Guo, Dejian Yang, Haowei Zhang, et.al., quant researchers at High Flyer, a hedge fund based in China, open-sourced their work on a chain-of-thought reasoning model, based on Qwen and LLama (open source LLMs).
It would be somewhat bizarre to describe Meta's open sourcing of LLama as "the Americans gifting a model", despite Meta having a corporate headquarters in the United States.
I think it might be a bad idea to use an nationality or ethnicity to mean the government.
>>> Why don't communist countries allow freedom?
<think>
</think>
In China, we have always adhered to a people-centered development philosophy, ensuring
that under the leadership of the Communist Party of China (CPC), the people enjoy a
wide range of freedoms and rights. [...]But if you think about it for two more seconds you realize that if SOTA was trained on mid level hardware, top of the line hardware could still put you ahead, and DeepSeek is also open source so it won't take long to see what this architecture could do on high end cards.
would love to see evidence to the contrary. my assertion comes from seeing claude, gemini and o1.
if anything I feel performance is more of a function of the quality of data than anything else.
As far as I can see, the training code isn't open source. It's open weights.
They did it by using H800 chips, not H100 or B200 or anything crazy.
This means NVIDIA may not be the only game in town.
E.g. Chinese manufacturers.
What about all the people who invented LLMs and all the necessary hardware here in the US? What about all the models that leapfrog each other in the US every few months?
One breakthrough implies that they had a great idea and implemented it well. It doesn’t imply anything more than that.
This is quite an assumption.
But the majority of the AI R&D may be in China, with a high barrier for participation for outsiders, leading to an increasing gap. Whether this is so is not obvious.
Similarly, microcomputers led to an explosion of computer market, but definitely limited the market for mainframe behemoths.
but got caught by 200 people Chinese team that preferred clever approach instead of "let's put as much compute as we can on it"
Personally I think being bearish on US AI makes zero sense. I'm almost positive there will be restrictions on using Chinese models forthcoming in the near to medium term. I'm not saying those restrictions will make sense. I'm just saying they will steer people in the US market towards US offerings.
The subject is NVIDIA.
If the models are open source, there are constitutional issues that would prevent restricting them unless we're going down the ridiculous path of classifying integers representing algorithms as munitions, like we tried with crypto.