@thedudeabides made a post about how, during the Biden years, it was sometimes said that China would never catch up with the West in AI, that if the West agreed to pause AI development then China would too, and so on. The post called for members of the rationality community who had believed and said these things, to check their premises and otherwise engage in some reflection.
That wasn't me so I don't much care about that, but I do want to know about Chinese companies that might be contenders to create superintelligence. A month ago I made a list of the entities worldwide that I regard as contenders. Most of them are American and I spent some time on their relationships with each other and with the new American government. DeepSeek was the only Chinese contender and I had nothing to say about the political and other conditions under which it operates. But if China in general is going to become a breeding ground for multiple contenders in the race for superintelligence, I feel that I need to improve my model of what's happening over there.
As a lifelong resident of countries in the American bloc, and a user of American Internet sites and services, it's a lot easier for me to put together a model of the American situation. Even before one considers tech, China has a language, history, culture, and politics that are all different from anything in the USA or its allies. And as far as tech and the Internet are concerned, it's a parallel world where whole cyber-dynasties have risen and fallen without the West even noticing, really.
About ten years ago, I did have accounts on Weibo and WeChat, and I may want to revive them just so I have a direct window on Chinese social media. I also know that Bilibili is their Youtube, Zhihu is their Quora, Baidu Baike is their Wikipedia...
But to grasp the big picture, one really needs guidance. For that, I have a tweet and a Substack post. The tweet is by a China tech watcher called T.P. Huang and it's a brag about the breadth of the Chinese AI sector. It's nothing systematic, just a list of companies using AI all through the economy. But if I ever want to understand how China is using pre-LLM AI, it's a list of names to investigate.
The Substack post is more about AI as we understand it here. It does list the companies that are China's answer to OpenAI et al. But it also puts them in the larger context of China's Internet sector. Part of understanding the American AI contenders, is knowing their relationships to companies whose central business pertains to earlier stages of the information society, i.e. hardware, operating systems (Microsoft, Apple), and Internet services (Facebook, Amazon, Google). AI is like a final layer in this evolution, in which software becomes capable of replacing the human users as well.
In any case, this post, "A Not-So-Quick Overview of the Chinese AI Scene" by Aksel Johannesen, gives us the following model of AI in China as of mid-2024. There are currently four big Internet companies in China, Tencent, ByteDance, Alibaba, and Baidu. Four "old dragons" of "AI 1.0" are mentioned, but they are involved with pre-LLM AI, such as facial recognition, and they don't seem very relevant to this discussion. And finally, we have four "new tigers" of "AI 2.0" - and these are the LLM companies: MiniMax, Zhipu, Baichuan, and Moonshot. Oh, and then the author adds a fifth company, 01.AI.
The LLMs are coming from the big, rich, old Internet companies and from the "new tigers". Some of their relationships are described. Tencent and Alibaba invest in the tigers; Baidu directly produces the most popular LLM in China, "Ernie".
We're also warned that the Chinese tech scene is hyper-competitive and fast-evolving, as one would expect, and indeed, DeepSeek, which revolutionized many people's perception of Chinese AI at the start of 2025, is not even mentioned in this essay from mid-2024. So anyone who wants to keep up will need a news source. Jordan Schneider's ChinaTalk comes highly recommended; I welcome any other recommendations, especially forums and sources from within China.