Less boom, more border: what the Midterms mean for the US-China AI race

“National data centre moratorium now,” declared Senator Bernie Sanders, one day after the Democrats’ sweeping victory. What followed turned out to be Wall Street’s worst day since 2020, while Beijing called the whole spectacle a “domestic affair”. Three capitals, one midterm, and a question that will outlast the ballot counts: does America still want to win the AI race, and at what cost?

Tuesday’s results were historic. Democrats secured 254 House seats and 54 in the Senate, their first full control of Congress in three years. Beyond scale, the composition of their victory is illuminating. Exit polls revealed 56% of voters calling AI and data centres a “crucial” or “major” factor in their vote, and 76% opposed a new data centre in their community. Of 45 data centre ballot initiatives nationwide, 39 AI-restricting measures were passed. The winners are not discreet about their intentions. Michigan’s Abdul El-Sayed has called for a “full-on pause” on frontier models, two-year moratoria, and an end to subsidies and tax breaks on AI companies, vowing to bring the AI billionaire class back down to earth. 

The mandate is real–a license to slow–but not a verdict on the global race for technological supremacy. 

America will build slower, but that does not settle the China question. Goldman Sachs estimates that areas where Democratic lawmakers are likely to restrict data centres account for around 45% of planned US capacity. Texas, home to roughly a quarter of planned development, has a newly elected governor promising to ban new approvals outright. Analysts now expect as much as a fifth of planned AI data centre capacity to vanish from the roadmap.

Yet the slowdown has hard edges. Federal legislation crumbles under Trump’s veto power, and the President fears that any easing off the accelerator will jeopardise America’s pursuit of “global dominance”. Restraint is not partisan either: AI remains a strategic priority for both parties. James Talarico, the Democrat who flipped Texas’ Senate seat, campaigned on protecting “America’s leadership in artificial intelligence”, even as he simultaneously promised to tame the industry. What a Democratic Congress is striding towards is not the pace of the race but the rulebook, and the rulebook is precisely where China lives.

So what can a Democratic legislature realistically do? Robert Garcia, the likely next chair of the House Oversight Committee, sums it up in two words: “Lawyer Up”. Rising representatives intend to use their newfound authority to issue subpoenas, investigate Elon Musk’s dealings with the federal government, and force tech leaders to testify under oath. Analysts expect these levers to push China policy in a “more hawkish direction” through legislation focused on economic security and Taiwan, with clear intent to tighten export controls on technology.

Taiwan is the litmus test. Trump has delayed a $14bn arms package for Taipei to preserve a fragile truce with Xi Jinping, an area where the Democrats are expected to challenge Beijing the hardest. As the AI buildout decelerates, the strategic tradeoff hinges on conditions explicitly outlined by Anthropic’s CEO, Dario Amodei: aggressive restrictions on Beijing’s access to advanced chips and a crackdown on Chinese “distillation” of American models. Amodei justifies this move as a sensible guard against naivety; the Global Times calls the same menu a “Cold War playbook” designed to contain China and preserve geopolitical dominance. Chinese experts have since warned that linking AI safety measures to US-China competition makes international cooperation feel impossible.

With Congress positioned to formalise this trade-off, the next two years will likely answer a pressing question: how does Beijing respond to an America that competes more quietly but restricts far louder?

So far, with procedure rather than protest. Foreign ministry spokesperson Mao Ning called the midterms “a domestic affair”, reaffirming a mutual consensus between the two presidents. Chinese analysts read the blue sweep as the usual pendulum swing of US politics, while awaiting louder congressional rhetoric on technology and human rights. Beijing insists that stable ties founded on fairness and reciprocity will benefit everyone.

This new Congress may look confrontational on the technology front, but the shift in Democratic power potentially opens a new door: a formal bilateral AI safety channel, with common model evaluations and mutual mechanisms to flag malicious use or unexpected system behaviour. With a lame-duck President, a Congress set on restrictions could also set its sights on something Washington has long refused to build: a common ground born of the acute risks AI poses.  

So does the Democratic sweep weaken America’s desperation for AI dominance? I would argue it has shaken the rhetoric, not the resolve. The buildout will slow, the AI billionaire class has lost its Congress, and the winner-takes-all AI-race language has been shelved by a party that ran against it. But the rivalry behind the machinery remains untouched: AI is a strategic priority for both parties, the veto pen remains in the hand of the man who declared the competition, and tech companies, wary of overcorrecting, are preparing for a Republican revival in 2028. 

Three tests will show which America emerges. First, whether the Shenzhen and Miami summits reveal if the Trump-Xi truce survives Congress. Second, whether Texas, Georgia, and Michigan actually enforce their moratoria. And third, whether the new Congress formally establishes an AI safety channel with Beijing or merely delivers a string of hearings. That last test will indicate whether the world’s two AI superpowers will race to the bottom or negotiate the edges of the cliff.

Image: Wikimedia Commons/Joli Rumi

Licence. 

No image changes made.