If the US decides to allow H200 sales to China, Chris, what is the risk that this choice might in some way support China’s military AI ambitions? Based on open-source information, we know that China has already repeatedly used US-origin AI chips for military applications. There has been extensive research on this topic. China is also demanding larger quantities of AI chips. However, it’s not just about which specific chip is used by the military; it’s about the overall strength of the AI ecosystem. Ultimately, long-term technological progress is driven by the vitality of the entire ecosystem. Therefore, any action that helps to build up Chinese AI companies or infrastructure essentially fuels that ecosystem and strengthens China’s position as a competitor in the long run. Chris, a frequently asked question is: who is leading the AI race, the US or China?How would you respond to that, and how has the situation changed compared to six or twelve months ago? Looking at the AI models themselves, US companies have a slight lead, but it is a very narrow margin. China is right behind. Whether you compare DeepMind versus OpenAI or Alibaba’s PAI versus Anthropic, the difference is minimal.

The pace of progress in AI models is remarkably rapid. So, in terms of models, the US does not hold a very strong advantage. The real edge lies at the chip level—specifically, the ability to deploy large volumes of cutting-edge chips in the most advanced data centers. The US, along with its allies, especially Taiwan, holds a commanding lead here. That’s why there is a strong focus in the US on preserving this lead and leveraging it to support the next generation of American tech companies. Finally, Chris, turning to the US domestic front, there is significant bipartisan interest in expanding domestic chip manufacturing to reduce chip dependency. How would you assess the progress we’re making in that area? There have been important investments from US firms as well as foreign companies, including Taiwanese and Korean firms, inside the United States. However, demand for chips has increased in parallel with these investments. So while there has been some progress, there is still a long way to go. At present, the most advanced AI chips are produced predominantly in Taiwan and Korea.