Trump and Xi's summit and the AI superpower race

Chinese President Xi Jinping arrived on American soil on September 23, 2026, for a three-day state visit that American officials billed as one of the most consequential US-China encounters since Nixon’s 1972 opening. President Trump personally met Xi at the plane — an unusual departure from diplomatic protocol that the White House framed as a sign of mutual respect. According to BBC reporting, the summit’s central agenda item is artificial intelligence: both sides see it as the definitive arena in which 21st-century superpower status will be decided. Al Jazeera reported that Trump greeted Xi in a scene with red-carpet formality rarely seen outside full state dinners. The meeting takes place against a backdrop of NPR’s analysis identifying cybersecurity, Taiwan, and AI governance as the three central issues on the table.

The received wisdom

The mainstream foreign-policy consensus treats this summit with something approaching relief. After years of tariff warfare, chip embargoes, and near-misses over Taiwan, the mere fact that the two leaders are sitting across a table from one another is taken as evidence that great-power management still functions. Progressive commentators stress the urgency of cooperation: on climate, pandemic preparedness, and AI safety, the argument goes, neither Washington nor Beijing can afford the luxury of pure competition. The framing in most outlets positions this as a hopeful departure from confrontational rhetoric — a chance to establish guardrails, stabilise supply chains, and prevent the kind of miscalculation that turns rivalry into war. The implicit assumption is that more summitry is better, that engagement reduces risk, and that Trump’s personal warmth toward authoritarian leaders is, on balance, a useful lubricant in a dangerous world.

A different read

There is something worth preserving in the engagement-optimist view, but it papers over a structural reality that no amount of airport ceremony can dissolve: the United States and China are engaged in a genuine zero-sum competition over the commanding heights of artificial intelligence, and both governments know it.

Consider what BBC’s data-centre investigation found: China is building AI infrastructure at what the report calls “China speed” — data centres rising at a pace that has no Western parallel. Meanwhile, NPR’s pre-summit analysis notes that the stakes extend from semiconductor supply chains to the question of which nation’s values get embedded in the world’s most powerful AI systems. These are not problems that a warm handshake resolves.

The historical parallel worth invoking here is détente. When Nixon and Kissinger opened to China in 1972, the strategic logic was sound: splitting Beijing from Moscow altered the balance of the Cold War in America’s favour. But détente also carried costs that took decades to materialise — technology transfers, industrial hollowing-out, the gradual integration of an authoritarian economy into a liberal trading order on terms that the authoritarian power exploited with spectacular effectiveness. The question for 2026 is whether Trump’s summit produces a genuine strategic rebalancing or simply reprises the pattern of trading short-term diplomatic wins for long-term structural vulnerabilities.

Trump’s own behaviour offers a clue that is underappreciated. He has simultaneously revealed substantial personal share deals in big tech and AI companies while hosting the leader of China’s rival AI programme. Whatever one thinks of the ethics of that arrangement, it at least suggests that Trump views this summit less through the lens of cold-war containment than through the lens of deal-making — which is to say, he may be more willing than his predecessors to trade specific concessions for specific wins in domains he personally values. That is not obviously worse than the managerial consensus approach of coordinating endlessly through interagency processes while China builds its data centres.

The deeper challenge is what Al Jazeera’s UN correspondent flagged: major powers are increasingly acting unilaterally, treating multilateral institutions as theatre rather than constraint. In that environment, a bilateral summit is not supplementary to the international order — it increasingly is the international order, for better or worse. That puts an enormous premium on the quality of what is actually agreed in the room, as opposed to what the press conference tells the world was agreed.

The right-of-centre case for scepticism here is not isolationism — it is precision. A summit that produces vague commitments to AI safety cooperation while leaving the underlying structural competition intact is not a success; it is a delay mechanism that gives Beijing time to close the capability gap while Washington congratulates itself on statesmanship. The Reagan administration’s approach to Soviet arms control — insisting on verification, specificity, and enforcement mechanisms rather than atmospherics — offers a more useful template than either the hawk who wants no engagement or the dove who mistakes a dinner invitation for a strategic concession.

What to watch

Watch the specific language on semiconductor export controls: any softening of chip restrictions would be the most concrete concession Washington could make, and Beijing knows it. Watch whether Taiwan is addressed explicitly or papered over with studied ambiguity — the latter has historically emboldened rather than constrained Chinese adventurism. Watch the OpenAI and Anthropic UN push for global AI terms — if Washington and Beijing jointly endorse a multilateral AI governance framework, that is a genuinely significant development; if they produce separate communiqués with compatible-sounding language, it is noise. Finally, watch Trump’s post-summit domestic political framing: how he sells the outcome to his base will reveal whether this was strategy or spectacle.

— J