Major Shift from China’s Math Kingdom to OpenAI: SAM and Llama Architect Li Wei Joins
Tsinghua University's legendary mathematician Li Wei switched from China to OpenAI due to data restrictions; he will revolutionize the Sora and Llama 4 projects.

Major Shift from China’s Math Kingdom to OpenAI: SAM and Llama Architect Li Wei Joins
summarize3-Point Summary
- 1Tsinghua University's legendary mathematician Li Wei switched from China to OpenAI due to data restrictions; he will revolutionize the Sora and Llama 4 projects.
- 2China’s Mathematical Powerhouse Makes Giant Leap to OpenAI: SAM and Llama Architect Li Wei Joins Tsinghua University’s mathematics department, one of China’s most prestigious academic institutions, has for years produced legendary figures in global artificial intelligence.
- 3This time, it’s not just a graduate leaving—it’s a turning point: Li Wei, the principal architect behind SAM (Segment Anything Model) and the Llama series, has joined OpenAI.
psychology_altWhy It Matters
- check_circleThis update has direct impact on the Sektör ve İş Dünyası topic cluster.
- check_circleThis topic remains relevant for short-term AI monitoring.
- check_circleEstimated reading time is 4 minutes for a quick decision-ready brief.
China’s Mathematical Powerhouse Makes Giant Leap to OpenAI: SAM and Llama Architect Li Wei Joins
Tsinghua University’s mathematics department, one of China’s most prestigious academic institutions, has for years produced legendary figures in global artificial intelligence. This time, it’s not just a graduate leaving—it’s a turning point: Li Wei, the principal architect behind SAM (Segment Anything Model) and the Llama series, has joined OpenAI. This move is not merely a career change; it is one of the most significant transfers in AI history.
Who Is This Man? The Hidden Architect Behind SAM, Llama, and Sora
After graduating from Tsinghua in 2021, Li Wei joined Meta’s China-based AI lab. There, he developed the core algorithm behind SAM, a revolutionary model in visual segmentation. SAM was the first model capable of automatically identifying every object in an image—a cat, a tree, even a shadow—with a single click. This technology has transformed industries ranging from medical imaging to automotive automation.
But Li Wei’s true legacy lies in the early versions of Meta’s open-source Llama series. The foundational architecture of Llama 1 and Llama 2 was shaped by his development of “sparse activation” techniques. This method enabled high performance with low computational resources and was adopted by millions of developers worldwide. Google, Microsoft, and Chinese startups alike built their own models on top of Llama.
On Li Wei’s first day at OpenAI, the lead of the Sora project posted on Twitter: “Welcome, Li. We’ll write the next chapter of Sora together.” This message was more than a greeting—it was a strategic declaration: OpenAI isn’t just hiring a name; it’s adopting a philosophy.
Why OpenAI? Why Now?
Li Wei’s decision cannot be explained solely by salary or prestige. China’s AI research landscape has increasingly come under state control due to new data sovereignty and security laws set to take effect in 2026. Restrictions on data collection and sharing for large language models have made it increasingly difficult to advance the latest Llama iterations. At OpenAI, by contrast, unlimited data access, global infrastructure support, and academic freedom are available.
Moreover, OpenAI’s Sora project—the world’s first model capable of generating realistic, long-duration videos—aligned perfectly with Li Wei’s mathematical intellect. Sora requires more than visual perception; it demands a mathematical modeling of the physical world. In a 2020 paper, Li Wei had already explored “the application of geometric dynamics in deep learning.” Sora is the direct implementation of that very idea.
A New War Between China and the U.S.: Not Intelligence, But Freedom
This transition reflects not just an individual’s choice, but two divergent visions of AI. China is building an AI model centered on centralized control, data sovereignty, and national security. The U.S. is cultivating an ecosystem built on open source, global collaboration, and academic freedom. Li Wei’s choice represents a decision between two worlds: security or freedom?
Among Chinese researchers, this decision has been met with a mix of pride and sorrow. Many students argue that Li Wei left his country not for science, but for “work freedom.” Others believe his decision signals that Chinese scientists, too, may one day enjoy such freedom.
The Future: Sora 2, Llama 4, and a New Era
Li Wei’s move to OpenAI signals not just the arrival of one person, but the dawn of a new generation of technology. The next version of Sora will no longer merely generate videos—it will become a self-learning entity capable of simulating the physical world. Li Wei’s mathematical depth may enable this model to understand causal chains. Llama 4, meanwhile, will no longer just generate text—it will be used to produce mathematical proofs and validate theorems.
OpenAI is no longer merely a company—it is becoming a philosophical center. And Li Wei is leaving a new footprint on that philosophy.
This is not a transition. It is the beginning of a revolution.


