The AI Trap: Why Washington's China Tech Strategy Is Hitting Its Own Ceiling
The Trump administration signed an executive order on 2 June requiring frontier AI companies to share their models with federal agencies before public release — a measure framed as cybersecurity and national security, but understood across government and industry as something closer to export control by another name. The same day, Secretary of State Marco Rubio was in Beijing telling reporters that Washington and Beijing must keep talking despite deep differences. Both moves are real. Neither is sufficient on its own. Taken together, they expose a structural incoherence that has defined US-China tech policy for the better part of a decade: the belief that America can both constrain Chinese advancement and rely on Chinese engagement to sustain American technological leadership.
That belief is becoming harder to maintain.
The Order and What It Actually Does
The executive order signed on 2 June directs companies developing advanced AI models to submit those models to federal review ahead of full release. Federal agencies are simultaneously instructed to develop cybersecurity standards for frontier AI systems — a recognition inside government that the most powerful models create national security exposure precisely because they are powerful. The order does not explicitly name China. It does not need to. The provenance of the threat is understood by every actor in the room.
Companies developing frontier models — OpenAI, Anthropic, Google DeepMind, Meta's AI division — will face a new pre-release gatekeeping requirement. Industry executives who have seen early drafts describe it as workable but consequential: not a hard ban on deployment, but a friction layer that adds institutional risk to the already high-risk business of shipping powerful AI. The order's language allows the President to designate certain model capabilities as restricted if they pose national security concerns. That designation authority, applied to the right benchmark, would effectively bar release of any model above a certain computational threshold.
The administration framed this as a defensive measure. The Chinese government's likely read will be something else entirely.
Rubio in Beijing: The Diplomatic Track
In a joint press appearance with Chinese officials on 2 June, Rubio said the US and China must maintain dialogue even as the two powers hold fundamentally different systems and interests. "The fact that we have different values doesn't mean we shouldn't talk," Rubio said, according to the South China Morning Post. "It means we must talk more, not less." The framing was careful: talking serves American interests, not as a concession to Beijing but as a risk-management necessity for Washington. The administration has used similar language in recent months — acknowledging the rivalry while insisting that unmanaged competition risks the kind of misunderstanding that could produce a crisis neither side wants.
Rubio's visit came against a backdrop of elevated tensions. American officials have made clear in private briefings that they see Chinese AI development as the central long-term challenge to American technological dominance. The export controls on advanced chips — Nvidia H100s and successors — remain in place. The question of whether to extend restrictions to consumer-grade AI hardware remains under review. And the Biden-era framework that required American cloud providers to report when foreign actors used their infrastructure to train frontier AI models is being expanded, not wound back.
Chinese officials have not been passive. Beijing has called for strategic stability talks, proposed AI governance frameworks under UN auspices, and insisted — through MFA briefings and state media — that China is not seeking to displace American technological leadership but rather to develop its own capacity. That framing finds some purchase in developing-world capitals where American technology governance is viewed with more suspicion than Chinese alternatives. In Africa, Southeast Asia, and Latin America, the question of who sets the rules for AI is not settled, and Chinese infrastructure — built through Belt and Road frameworks and private investment — offers an alternative to American-controlled systems.
What China Is Actually Doing
The Reuters reporting on Chinese oil stockpiling — imports at decade lows as Beijing draws down strategic reserves — illustrates a broader strategic posture that has been consolidating for two years. China is building redundancy into its economic architecture, reducing exposure to the kind of supply chain disruption that export controls on advanced semiconductors have made a policy tool. The response is not simply to complain about American restrictions. It is to invest, at state direction and through directed private capital, in making those restrictions irrelevant over a five-to-ten-year horizon.
The biotech sector is the clearest example. The South China Morning Post's reporting on 2 June identified biotech as the next likely arena for US-China competition, following semiconductors, AI hardware, and quantum computing. China has built world-class pharmaceutical research capacity in Shenzhen, Shanghai, and Beijing over the past decade. Chinese-developed biologics and gene therapies are moving into clinical trials at pace that has alarmed American national security planners. The dual-use character of advanced biotech — medical applications that can serve military medicine, bioweapon research, or population health management in crisis — makes it a sensitive category that sits awkwardly alongside the administration's stated desire for therapeutic cooperation.
American officials are watching. The National Security Council has commissioned reviews on Chinese biotech investment in the United States. The Commerce Department has moved to restrict exports of certain biological research equipment. But the strategic logic differs from semiconductors: Chinese biotech does not depend on American input in the same way that Chinese AI training once depended on Nvidia chips. The restriction leverage is weaker, and the Chinese counter-response is more likely to succeed.
The oil stockpiling is not incidental. It signals that Beijing is preparing for a long period of friction — economic, technological, and potentially military. Drawing down imports while building domestic production capacity is the same playbook China ran with semiconductors, where massive state investment in SMIC and domestic fabrication infrastructure has produced chips that, while not at the leading edge, are sufficient for military and industrial applications. The time horizon for biotech self-sufficiency is longer than for semiconductors, but Beijing is betting on it.
The Structural Contradiction
The executive order on AI models, the export controls on chips, and the diplomatic pressure for Chinese restraint share a common premise: that American leverage over Chinese technological development is sufficient to shape Chinese choices. That premise is weakening.
The Trump administration's stated desire — that China slow its AI development to give American companies time to catch up — reflects a strategy that depends on Chinese willingness to be managed. It is not clear that Beijing shares that willingness. Chinese officials have watched American administrations restrict technology transfer, expand academic visa restrictions, block Chinese investment in American AI startups, and pressure allies to do the same. The accumulated weight of those measures, Chinese strategists believe, demonstrates that the American position is not temporary adjustment but strategic containment. That belief shapes Chinese behavior in ways that American officials find difficult to model.
If China interprets American policy as containment rather than competition management, the response will not be restraint. It will be acceleration — and parallel investment in making the American leverage irrelevant. The export controls on advanced chips have already produced that dynamic in semiconductors. Chinese AI labs are training on domestic hardware where possible, optimizing architectures to run on less capable chips, and treating the restrictions as a forcing function for indigenous semiconductor development. The timeline for full autonomy is long — five to ten years at minimum — but the direction is set.
The Stakes
The immediate stakes are defined by the AI competition itself. The country that develops and deploys the most capable AI systems first gains advantages in military applications, economic productivity, scientific research, and the soft power that comes from being the source of the technology that shapes the next decade. American officials have made this case explicitly, arguing that AI leadership is a national security imperative rather than a commercial preference. China has made the same argument in Beijing. Both are right, and both are positioning accordingly.
The executive order adds a new layer of uncertainty to that competition. Companies developing frontier AI in the United States will now navigate a regulatory requirement — share your model with federal agencies before release — that has no equivalent in China. Chinese AI labs face their own regulatory environment, but it is one they understand and have adapted to. The American order introduces friction at the moment when American companies are trying to stay ahead. Whether that friction is worth the security benefit is a genuine policy question, not a settled one.
Rubio's emphasis on dialogue suggests the administration does not want to sever the communication channels that might prevent miscalculation. That is right. But the structural forces driving the two powers apart — competition for AI leadership, divergent security interests in the Pacific, different models of technology governance — will not be managed by dialogue alone. The next twelve months will test whether the two sides can find stable terms of engagement for an AI competition that neither wants to lose and both know they cannot fully control.
The answer, for now, is probably not.
This publication covered the Trump AI executive order and Rubio's Beijing visit as parallel developments rather than contrasting narratives — the diplomatic and the coercive track running simultaneously, as they have for the duration of this administration. American wire coverage has tended to treat the executive order as a domestic policy story and Rubio's visit as a foreign policy story; this piece treats them as the same story.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- http://reut.rs/43fa5iG
- https://x.com/polymarket/status/1951030409814761473
- https://x.com/sprinterpress/status/1950998345676902400