American AI firms want it both ways in limiting, profiting off China
The piece argues US AI firms are pushing tougher China controls while still building for China-adjacent markets.
Intelligence analysis by GPT-5.4 Mini
Jeffrey Wu says US AI firms are publicly backing stricter export controls on China while privately hiring and expanding around markets they cannot officially serve. He frames that as a hedge between security politics and growth.
Some AI companies are acting like a student who says a rival should not be allowed near the best tools, while still trying to stand close enough to use the same room.
The story says these companies want the US to keep a big lead in AI by making it harder for China to get key technology. At the same time, they still try to build teams and skills near places connected to China.
It is like wanting to win a race while also making sure the other runners slow down. The article says that split behavior shows they are trying to protect both power and profit.
Analysis
Two contests at once
The article argues that American AI firms are operating in two different races at the same time. One is a frontier capability contest, where the prize is who can build the strongest models and secure the compute needed to keep improving them. In that frame, China is treated as a strategic rival and tighter export controls are presented as necessary to preserve a US lead.
Jeffrey Wu points to Anthropic’s policy paper and lobbying disclosures as evidence of that posture. The company urged the US and its allies to protect a lead in frontier AI, and it also spent more on federal lobbying, with export controls and AI national security among the topics listed. The article presents this as an industry argument that scarcity matters: if access to advanced chips and systems is constrained, the US can try to preserve technological advantage.
The commercial hedge
At the same time, Wu says the same companies are behaving as if China still matters commercially, even when they do not officially sell services there. He cites OpenAI hiring Mandarin-speaking developer experience engineers in Singapore as an example of building language and developer ecosystems around China-adjacent markets. The point is not that these firms are openly serving mainland China, Hong Kong, or Macau, but that they are positioning themselves near those markets while asking Washington to narrow access for rivals.
The article’s broader argument is that this is not just corporate inconsistency. It reflects an industry trying to satisfy two competing incentives: support tougher controls for national security reasons, while still keeping a path to growth in nearby markets that could be affected by those same controls. The result is a hedging strategy that looks pragmatic from a business angle, but politically contradictory from a policy angle.
Key points
- The article says US AI firms are pushing for tougher limits on China while still courting China-adjacent business opportunities.
- Anthropic is cited as urging the US and allies to preserve a lead in frontier AI and as increasing federal lobbying.
- OpenAI is cited as hiring Mandarin-speaking developer experience engineers in Singapore despite not officially serving mainland China, Hong Kong, or Macau.
- The piece argues the industry is split between security-driven scarcity and growth-driven market expansion.
- The central tension is that the same controls that help firms politically may also limit their commercial reach.
