Sinovation Ventures chairman, 01.AI founder and CEO
Kai-Fu Lee
Profile
Kai-Fu Lee is one of the few people who can speak with equal authority about AI in both Silicon Valley and China — and that binocular view is what makes him worth listening to. His academic credentials are deep: a CMU PhD in 1988 where his dissertation system Sphinx was the first speaker-independent, continuous speech recognition system, built on the statistical hidden-Markov-model approach that would dominate speech for the next two decades. From there he ran research and engineering at Apple, Silicon Graphics, Microsoft — where he founded Microsoft Research Asia in Beijing in 1998, a lab that trained an extraordinary fraction of China’s current AI leadership — and then Google, where he opened and ran Google China.
In 2009 he left to found Sinovation Ventures, a VC firm that has become one of the most active AI investors in China, backing companies across autonomous driving, robotics, enterprise AI, and foundation models. Then in 2023 he did something most VCs in their sixties don’t: he stopped investing from the sidelines and started building. 01.AI ships the open-source Yi series of large language models, and within months of launch the Yi-34B was competitive with Western open-weights models and the company was valued north of a billion dollars. It’s a rare case of a senior industry figure betting his own reputation on shipping product.
His 2018 book AI Superpowers is the reason a lot of Western readers first took Chinese AI seriously. The thesis — that China’s data abundance, execution culture, and government support would let it catch up fast — looked contrarian at the time and looks prescient now, especially after DeepSeek and the rise of Liang Wenfeng. Lee doesn’t frame the US-China AI dynamic as a zero-sum war the way many policy hawks do; he treats it as parallel ecosystems with different strengths. Whether you agree with that framing or not, it’s the clearest articulation of the Chinese side of the story written for a general Western audience.
For developers learning AI, Lee is useful for a specific reason: he has watched, funded, and now built through every wave of this technology, from 1980s speech systems to 2020s transformers. When he writes about what’s next, it’s grounded in having been wrong and right many times already.
Books
Videos
YouTube
Spotify Podcasts
