Trump and Xi meet in Beijing with Elon, Jensen, and Tim Cook in tow — what's actually at stake

Trump-Xi Summit: The AI Gap Beijing Wants to Close

Trump brought Jensen Huang, Elon Musk, Tim Cook, and a roster of finance and defense executives to Beijing—a visible signal that the summit is about more than diplomatic theater. What's notable is who isn't there. Absent are the CEOs of the major U.S. AI labs: Dario Amodei from Anthropic, Sam Altman from OpenAI, and Demis Hassabis from Google DeepMind. That absence reveals a fundamental disconnect between the conversations happening in Silicon Valley and the geopolitical negotiation unfolding at the highest levels.

The semiconductor wedge

The Wall Street Journal's editorial framing of the summit centers on stability and tariffs, rare earths, and Taiwan. But the most significant technological tension is unspoken. Xi Jinping views AI as a decisive theater in competition with the United States and is trailing by roughly six months by most estimates. The administration's stated goal is to discuss AI "guardrails," but the editorial board is blunt: expect little from AI arms control. The real deterrent is U.S. dominance in models and compute power.

Yet the administration's policy appears incoherent. Treasury Secretary Scott Bessent personally vets access to Anthropic's Claude Mythos model—the most advanced AI system with cybersecurity and exploit-finding capabilities—restricting it to roughly 50 organizations worldwide. Bessent himself likely briefed Japan's three mega banks (MUFG, Sumitomo Mitsui, and Mizuho) on Mythos access as part of a show of U.S. confidence in allied relationships.

At the same time, Jensen Huang is on Air Force One flying to Beijing to sell China advanced AI chips. The administration appears to be simultaneously restricting access to its most advanced AI models while enabling the export of the chips China needs to build its own.

The DeepSeek question

The more strategic question is what happens when China develops its own Mythos-equivalent capability. Dario Amodei has said the open-source community in China is roughly six months behind the frontier, and that timeline appears consistent with observed progress from U.S. labs. The U.S. had a high-stakes moment when the Department of Defense pressured Anthropic over Mythos access, with threats of supply-chain designation. That tension has since quieted—Mythos remains available on hyperscaler clouds and is being granted to allies.

But the dynamic in China operates at a different scale. DeepSeek, the company behind China's most advanced models, is controlled by a founder with majority voting control and minimal dilution. That's radically different from Anthropic, which has 15 co-founders, multiple funding rounds, and is moving toward public markets where the SEC exercises oversight. If and when China reaches parity on frontier AI, it will be through a tightly-held private company answering to the CCP—not a democratically-governed entity subject to regulatory scrutiny and public accountability.

The question worth watching is what China's "Mythos moment" will look like: Will DeepSeek face government pressure to consolidate control over compute resources? Will the CCP channel all advanced chips into a single national data center to train an AGI-competitive model? The U.S. government has the regulatory authority to do that but hasn't exercised it. China has the de facto authority and a track record of consolidating control once companies become strategically important.

What's actually at stake

The summit's public agenda—tariffs, farm goods, rare earths, Taiwan—obscures a deeper structural question: Can the U.S. maintain compute dominance while exporting the chips that enable Chinese parity? And if China does achieve a Mythos-level capability within the next year or two, what does that mean for U.S. leverage in a world where both superpowers have weaponizable AI systems and neither trusts the other?

None of that is on the official schedule. That's why the absence of Altman, Amodei, and Hassabis matters. They can't answer what a post-AGI U.S.-China relationship looks like because that conversation isn't being had at the table where the deals are struck.