The U.S.–China AI contest is entering a new phase: competition over who controls not just the frontier, but its global diffusion.
On October 16, China's Ministry of Commerce and Ministry of Science and Technology announced proposed export restrictions covering 40 strategic technologies, including 23 AI technologies involving frontier models, autonomous agents, computing clusters and information processing.
Crucially, the announcement introduces immediate pre-export registration requirements for covered technology transfers, while the broader restrictions remain under consultation until November 15.
Why this is consequential
A second strategic objective
Until recently, China's international AI strategy relied heavily on distributing capable, inexpensive models to accelerate adoption and challenge the economics of American proprietary systems.
Beijing is now adding a second strategic objective: protecting sensitive AI intellectual property and retaining greater control over cross-border technology transfers.
This is not the abandonment of China's open-model strategy.
The proposed framework explicitly excludes technologies already in the public domain and provides simplified licensing for qualifying open-source releases. API access and certain other services are also excluded from the proposed technology-transfer restrictions.
The emerging structure is therefore more sophisticated: open diffusion for ecosystem expansion, selective control for strategic technological advantage.
The second-order effects
Three consequences
Three consequences deserve attention.
First, AI becomes a geopolitical instrument on both sides of the Pacific. American restrictions on advanced computing face an emerging Chinese framework governing the transfer of selected AI technologies. Both countries are increasingly treating parts of the AI stack as strategic assets.
Second, the competitive battleground shifts toward the economics of deployment. Chinese providers can continue expanding adoption through accessible models and inexpensive inference, while potentially retaining tighter control over strategically sensitive capabilities.
Third, technological sovereignty becomes more expensive. China's continued dependence on Nvidia technology illustrates the difficulty of replacing an established accelerated-computing ecosystem. The Financial Times reported on October 13 that Chinese companies remain reliant on Nvidia products despite expanding domestic alternatives.
These developments strengthen the prospect of two increasingly distinct AI ecosystems, although commercial interdependence remains substantial.
Implications for the AI Supercycle
Three sources of value
The emerging competitive structure increasingly separates three sources of economic value:
• Frontier innovation: Advanced models, research capabilities and accelerated-computing infrastructure, where the United States retains important advantages. • Industrial diffusion: Affordable inference, adaptable models and large-scale deployment, where Chinese competition is increasingly consequential. • Economic capture: The ability to convert AI capabilities into productivity improvements, enterprise revenues and sustainable returns on invested capital.
For investors, this distinction is fundamental.
The winners of the next phase will not necessarily be those financing the largest training clusters. They may be the companies translating increasingly accessible intelligence into recurring industrial cash flows.
What changes in our thesis?
One important refinement
The core thesis is strengthened, but it requires one important refinement.
China is not simply competing to make frontier intelligence cheaper. It is also developing a framework for governing which strategically sensitive capabilities can move abroad.
This introduces a new variable into the U.S.–China AI competition: control over technological diffusion. The decisive test will be whether China can preserve the international adoption advantages of open models while introducing selective export controls.
The November 15 consultation deadline is therefore an important policy milestone—not evidence that the proposed restrictions have already been finalized.
Sources: China's Ministry of Commerce announcement · Reuters, October 16
The United States leads much of the technological frontier; China is challenging the economics of its diffusion—and increasingly asserting control over the terms of that diffusion.