American tech giants are invoking the Chinese specter to avoid any regulation. Yet the United States enjoys a massive advantage in computing power, energy, and talent.

For some time now, a narrative has taken hold in Western capitals, particularly in the United States. We are repeatedly told that the West is engaged in a frantic race against China to dominate artificial intelligence, and that losing this race would have catastrophic consequences for our societies and values. This narrative is simple, anxiety-inducing, and effective—but it contains a major flaw: it does not reflect the global technological reality. Instead, it may serve other interests.

First, let us try to understand where we stand. According to several studies, including those by D. Kokotajlo, progress in AI rests on three pillars. The first is “compute,” the raw computing capacity required to train advanced models. The second is access to abundant and stable energy, since each new generation of models consumes ever-increasing amounts of electricity. The third is human talent, indispensable for designing, fine-tuning, and supervising these systems. Without compute, no models. Without energy, no compute. Without talent, no progress. And on all three pillars, the United States currently enjoys a massive structural advantage.

Artificial intelligence

The United States possesses roughly five times more computing power than China, largely thanks to Taiwan, where TSMC manufactures the world’s most advanced chips using American equipment. Without these components, China cannot train comparable models. The American advantage also rests on energy. Its vast energy mix allows it to power data centers at costs well below those of China or Europe. The United States also has underutilized gas power plants that can be mobilized quickly. China, by contrast, remains constrained by local grid saturation and a heavy reliance on coal. As for talent, leading AI researchers are primarily based in the United States, which attracts top profiles trained in Europe, India, or China.

U.S. tech leaders speak of an existential threat and claim that any regulation would mean losing the race, while deploying massive lobbying efforts. This rhetoric is not without echoes of the Cold War, when the military-industrial complex amplified Soviet power to secure budgets. Presenting AI as vital makes it possible to capture public contracts while weakening democratic safeguards.

The confrontation is less between Washington and Beijing than between industry giants and democratic institutions. In California, the ambitious SB 1047 bill was buried under industry pressure and replaced by the TFAIA (Transparency in Frontier Artificial Intelligence Act), a hollowed-out version that changes almost nothing in corporate practices. Yet the risks are real. Industry leaders themselves acknowledge that an uncontrolled AI could threaten global security, with Sam Altman even evoking an extinction-level risk. How, then, can a strategy that accelerates this race while undermining democracy be justified?

Switzerland does not need to imitate U.S. deregulation or Asian rigidity. It can choose a clear technological path: invest in compute, secure energy supplies, attract talent, independently test models, and require a minimum level of transparency. This is how an open and liberal country can regulate AI without stifling it—by strengthening both trust and innovation.Et si la vraie menace de l’IA ne venait pas de «l’autre»?

Guest contributor: Nicolas Ramseier, President of the Geneva Center for Neutrality.