The Competition Doctrine Meets the Caribbean: How Trump's Cuba Play Intersects With the China AI Race

On the morning of 21 May 2026, standing on the South Lawn of the White House, President Donald Trump described a nation ninety miles from Florida as a place without electricity, without money, without anything. "It's a failed country, everybody knows it," he told reporters. That same day, administration officials were quietly circulating a different message internally: Cuba is not merely a failed state — it is a potential Chinese intelligence platform, a node in a broader challenge to US regional dominance that warrants both hardening and, paradoxically, continued engagement. A Reuters dispatch filed at 00:50 UTC on 22 May reported that Trump had postponed signing an artificial intelligence executive order, citing the need to ensure the United States retains its competitive edge over China. The Cuba messaging, and the AI order delay, are not unrelated. They are two fronts of the same strategic posture — one that treats every bilateral relationship as a subset of the China question, and that struggles to process the Caribbean island on its own terms.

The contradiction runs deep. Administration messaging, as reported by the South China Morning Post on 22 May, casts Cuba explicitly as a China-linked security concern: a growing Chinese military and intelligence footprint in a nation the US has embargoed for more than six decades. Yet the same administration, per reporting from the same news cycle, is still pushing for talks. Trump himself, on Polymarket and in public statements the same day, suggested he would be the one to "do something" on Cuba — language that implies engagement, not only pressure. This is not the coherent doctrine it pretends to be. It is a collision between a competition framework that names Beijing as the central threat and a hemispheric reality that includes a bankrupt island economy, a diaspora in south Florida that votes, and an oil market that has not waited for Washington's permission to do business with Havana.

The Intelligence Framing — What the Administration Is Actually Worried About

The security case against Cuban-China cooperation is not invented. Chinese surveillance infrastructure has expanded across the island in recent years. The signals intelligence facilities at Lourdes, which once served Soviet interests during the Cold War, have been partially reactivated in cooperation with Chinese technical personnel. Naval resupply agreements have given Chinese vessels increased access to Cuban ports. These are documented developments that US military and intelligence officials have briefed to Congress in unclassified and classified form. The concern is legible: a Chinese SIGINT node ninety miles from Key West would give Beijing real-time access to US naval traffic in the Gulf of Mexico and a portion of the southeast coastline.

But the framing matters. By casting every Cuban-Chinese contact as a national security emergency requiring a full-spectrum response, the administration forecloses options that might actually reduce the Chinese footprint — or at least manage it more surgically. The history of US Cuba policy is not encouraging on this front. The embargo, renewed and tightened across multiple administrations, has produced no regime change. It has produced an economy dependent on Venezuelan oil subsidies and Chinese infrastructure loans, and a generation of Cubans who have every reason to view Washington as the obstacle to their prosperity. The security threat, in other words, is real — but the policy posture designed to counter it has demonstrably worsened the conditions that make the threat viable.

The Reuters reporting on the AI executive order delay offers a parallel case study. Trump postponed signing the order, sources said, to review whether it would give China any additional competitive advantage in the race to develop frontier artificial intelligence models. The delay itself is unremarkable — executive orders routinely undergo review — but the stated rationale reveals how far the competition frame has colonized every corner of the agenda. Even domestic AI governance is now filtered through the question of what it means for US-China parity. That is a coherent strategic concern. It is also a frame that makes it harder to ask what responsible AI governance actually requires domestically, independent of the rivalry with Beijing.

The Diplomatic Opening — Who Wants to Talk and Why

The administration is simultaneously pushing for talks. The South China Morning Post reported on 22 May that US officials, while publicly framing Cuba as a security threat, have not closed the diplomatic channel. The nature of those contacts is unclear — whether they involve back-channel intermediaries, third-country facilitators, or direct talks through the Swiss protecting power in Havana — but the intent appears genuine. Trump himself, via the Polymarket post on 21 May, indicated personal interest in being the figure who achieves something on Cuba. That language carries domestic political weight. The Cuban-American vote in Florida remains consequential, and a president who can credibly claim to have extracted concessions from Havana — even modest ones — gains something with an audience that has been promised results for sixty years and received none.

The question is what a deal would look like. Chinese infrastructure presence on the island complicates any negotiation because it raises the threshold of what the US would have to offer in exchange for Cuban concessions. Havana, which has watched Venezuela's Maduro use Chinese loans to insulate his government from US sanctions pressure, understands that Beijing is now an alternative to Washington as a patron. The negotiating leverage of the United States, therefore, is not what it was when Cuba's only option was Soviet subsidy and that subsidy had a twenty-four-hour shelf life. The administration would need to offer something substantive — sanctions relief, removal from the state sponsors of terrorism list, restored Remittance flows — and would need to do so while explaining to a domestic audience why engaging a "failed country" aligned with China is consistent with a maximalist competitive posture.

There is also the question of what Havana actually wants. Cuban officials, in private communications that have surfaced in academic and think-tank literature over the past two years, have consistently signaled a desire for normalized US relations, not as an ideological project but as an economic survival strategy. The current government is not ideologically Marxist in any meaningful sense; it is a rent-seeking apparatus that has survived by managing the intersection of Soviet, Venezuelan, and Chinese subsidies. Normalization would give it a new source of rents. Whether it would also require the kind of political opening that the regime cannot survive — that is the structural question no Cuban opening has ever resolved.

The Latin American Context — A Region That Has Already Moved On

Whatever Washington decides about Cuba, the rest of the hemisphere has already moved. Brazil, under successive administrations, has expanded trade and diplomatic ties with Havana without framing that expansion as a challenge to US authority. Mexico's state-owned oil company has done business with Cuban partners through third-country intermediaries. Argentina's new government, despite its alignment with Washington on most foreign policy questions, has not prioritized Cuban human rights as a bilateral issue. The regional consensus — insofar as such a thing exists — treats Cuba as a neighbour to be managed, not a threat to be contained.

This is not a reflexively anti-American posture. It reflects a calculation that US sanctions policy has failed, that the Cuban people bear the cost of that failure, and that Latin American governments have no interest in participating in a renewed US pressure campaign that would alienate their own voters without producing results. The administration has not offered these governments a compelling alternative framework — one that might combine the security concern about Chinese intelligence infrastructure with a practical path toward reduced Cuban dependency on Beijing. That kind of approach would require nuance that the current framing, which flattens every issue into the China competition, does not accommodate.

The AI order postponement, again, is instructive here. The administration is effectively treating frontier AI development as a zero-sum contest with China, which means it has incentives to restrict information flows, limit academic collaboration, and constrain the open-weights model ecosystem that has driven much of the innovation in the field. That posture has consequences for US relationships with Latin American governments that are themselves navigating the AI transition — countries that will need to choose between US and Chinese technology stacks, and that will make that choice partly on the basis of which power offers them genuine partnership rather than geopolitical subscription. A US policy that treats AI governance as a weapons-adjacent competition discipline may accelerate the very Chinese influence in the hemisphere that the Cuba framing is nominally designed to counter.

The AI Executive Order — What We Know and What Remains Unclear

The Reuters report of 22 May is thin on specifics about the content of the postponed order. Sources cited only the rationale — ensuring US competitive advantage over China — and the fact of the postponement itself. The order was originally scheduled for signing, then pulled for review. Administration officials declined to specify the duration of the review or the specific provisions under reconsideration.

What can be inferred is structural. An AI executive order from this administration would likely involve export controls on advanced semiconductors, restrictions on data flows to Chinese-owned platforms, procurement preferences for US-based AI infrastructure, and possibly some form of investment screening on AI-adjacent transactions. Those are the levers that have defined the administration's approach to the China tech relationship since the first term. Whether the order would also include domestic governance provisions — safety standards, transparency requirements, federal procurement rules — is unclear from the public reporting.

The competitive framing is not self-evidently compatible with good domestic AI governance. If the goal is to beat China to AGI, or to the most capable frontier models, then the incentive structure points toward speed, toward minimal friction, toward absorbing safety research as a cost rather than a constraint. If the goal is to ensure that AI systems deployed in the United States are safe and reliable, then the incentive structure points in a different direction — toward the kind of regulatory infrastructure that the current administration has shown no appetite to build. The postponement, by the administration's own framing, was driven by competitive anxiety, not by a concern that the order needed refinement on safety or accountability grounds. That tells us something about the hierarchy of concerns inside the White House as of 22 May 2026.

Stakes — What a Failed Cuban Opening Would Cost

If the administration attempts and fails to secure a meaningful Cuban diplomatic opening — whether because Havana's Chinese patron objects, because the security infrastructure cannot be unpicked, or because domestic political constraints prevent offering the necessary concessions — the consequences will be asymmetric. Cuba will remain as it is, which is to say, increasingly Chinese-dependent, increasingly desperate, and increasingly irrelevant to a Washington that has exhausted its own ideas. The China presence will grow incrementally. The humanitarian damage will continue. And the administration will have demonstrated that the competition framework, applied to every foreign policy question, does not produce solutions to problems that are not fundamentally about China.

The alternative is more demanding. It requires treating Cuba as a country with its own logic, its own internal politics, its own survival calculations — and accepting that the most effective way to reduce Chinese influence in the Caribbean may be to give Cuba a reason to reduce its own dependence on Chinese patronage. That requires engagement, which requires accepting that the embargo has failed, which requires a kind of intellectual honesty about US Cuba policy that no administration in sixty years has managed. The AI executive order delay suggests this administration thinks in terms of competition. The Cuba messaging suggests it is being forced, by the stubbornness of the facts on the ground, to consider something more complex. Whether it can move from consideration to policy is the open question — and it is the question that will determine whether the hemisphere in 2030 looks more like a US sphere of influence or more like contested ground between two great powers who have not solved the problem of what to do with a bankrupt island ninety miles from Florida.

This publication approached the Cuba-China overlap by leading with the security framing that US officials have publicly deployed, while foregrounding the structural contradictions — particularly the simultaneous engagement signals — that the reporting itself contains. The dominant US wire coverage focused on the AI order delay and the "failed country" quote as separate stories; this piece treated them as nodes of the same strategic posture.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• http://reut.rs/4wHUZQb
• https://x.com/unusual_whales/status/1922751965175226576
• https://x.com/polymarket/status/1922692958533148790