Key Takeaways
- DeepSeek, Alibaba, and Zhipu AI want Nvidia H200 GPUs to avoid falling behind the frontier set by American competitors.
- Beijing actively pressures domestic AI developers toward local suppliers like Huawei, despite production bottlenecks that slow down training.
- Chinese industrial policy rejects short-term GDP growth and stock market returns, focusing resources on high-tech self-reliance and “new quality productive forces.” (play from 33:36)
- The strategy forces Chinese founders into a trade-off between competing on model performance today and building a protected domestic supply chain tomorrow.
The Fight Between Frontier Models and Domestic Silicon
Commercial AI labs in China face an immediate operational dilemma. When engineering teams at DeepSeek, Alibaba, or Zhipu AI build software, their benchmark is the global frontier. Falling six months behind American competitors means losing developers, enterprise clients, and mindshare. To keep pace, Chinese developers prefer to build on Nvidia hardware.
As Kyle Chan put it, “There are the Chinese AI companies, like DeepSeek or Zhipu AI or Alibaba, who they they want to build better and better models. They would probably prefer to just keep building on Nvidia chips if they could.” (play from 30:14) When export controls shifted to allow certain hardware into China, demand spiked immediately. Chan noted that “you see this tension where you have the Chinese AI companies clamoring to get the H200s that Trump allowed to be exported to China.” (play from 30:46)
Buying American accelerators clashes with Beijing's plan. Every dollar spent on Nvidia processors is a dollar withheld from domestic alternatives like Huawei. Domestic chips face production bottlenecks and smaller software toolsets. If Chinese labs do not adopt domestic hardware, those tools cannot improve. If labs are forced onto domestic chips too early, their training speeds slow down and the gap with American models widens.
Beijing Prioritizes Supply Chains Over Corporate Profits
Western investors often misinterpret Chinese policy because they assume Beijing plays by Wall Street rules. They look for consumer stimulus, rising equity prices, or quarterly GDP targets. When aggressive state-backed expansion floods sectors with capital and depresses corporate profit margins, outside observers call it market failure.
Chan argues that this critique misreads the state's objective: “If you start with the premise that the goal is not GDP growth in and of itself, the goal is high-tech self-reliance of sort of like winning the industries of the future. If you start with that premise, everything starts to make a lot more sense.” (play from 33:00)
Xi Jinping's policy focus centers on what Beijing terms "new quality productive forces." The state treats semiconductor fabrication, advanced packaging, and domestic AI chips as strategic assets rather than profit engines. Chan outlined the core tension: “And in this specific case, you have this trade-off between today, can you close the gap with the US AI frontier? Can you narrow that gap? Or are you focused on tomorrow, being able to build a fully self-reliant semiconductor supply chain.” (play from 31:21)
Lagging behind the global frontier is an acceptable price for Beijing if it builds an independent technology stack. As Chan explained, “For the Chinese central government, the goal is not just to have really incredible models today. The goal is to build up that kind of independence and sovereignty essentially in supply chains in the long run.” (play from 30:32)
What to Do With This
Audit your compute infrastructure for geopolitical and vendor exposure this week. If your core product relies entirely on a single proprietary model API or specific foreign hardware that could be restricted by export controls or compliance rules, build a secondary fallback pipeline. Run a performance test using an open-weights model on alternate cloud hardware, and calculate the exact latency and cost penalties so your roadmap survives regulatory shocks.