The Trump administration may have a defensible case for loosening export controls on advanced chips — provided that superintelligence is not imminent. That is the argument advanced by the Brookings Institution, which contends that the timeline for achieving superintelligent AI systems should anchor the debate over how tightly Washington restricts the sale of cutting-edge semiconductors abroad.

The argument turns on a single premise. If artificial superintelligence remains a distant prospect rather than a near-term event, then the strategic value of hoarding advanced computing hardware through export restrictions weakens considerably. Controls designed to deny adversaries a decisive technological leap make sense only if that leap could plausibly arrive soon, Brookings argues. Under a longer timeline, the commercial and diplomatic costs of strict controls begin to outweigh their security benefits.

This framing lands in the middle of a live policy fight. Export controls on advanced chips — the accelerators and associated hardware at the frontier of AI compute — have become one of the most contested instruments in US technology policy since their expansion earlier in the decade. Policymakers have used them to slow rival states' access to leading-edge silicon, while chipmakers have complained that the restrictions hand market share to competitors and push buyers toward alternative suppliers.

The Brookings position effectively reframes the question. Rather than asking whether controls are tough enough, it asks whether the threat model justifying them is realistic. If superintelligence is not close, the case for sweeping restrictions on advanced AI chips weakens, and the Trump administration's willingness to loosen them becomes harder to dismiss as simply deregulatory reflex.

The stakes for the semiconductor industry are direct. Export rules govern which customers can buy top-tier AI accelerators, in what volumes, and through what licensing regimes. Looser controls would expand the addressable market for US chipmakers; tighter ones would constrain revenue but concentrate advanced compute in allied hands. Where the superintelligence timeline lands is therefore not just a philosophy-of-mind debate — it is a variable that shapes demand forecasts, fab utilization decisions and pricing across the AI silicon supply chain.

The counterargument is obvious. If superintelligence does arrive sooner than expected, hardware controls would be among the few levers capable of slowing its diffusion to adversarial states. That asymmetry — modest commercial cost against potentially decisive strategic loss — has motivated the hawks in this debate, and Brookings does not dissolve it. The think tank's contribution is conditional: loosen controls if, and only if, the short-timeline scenario lacks credible support.

That conditionality matters for an industry that plans in five-year cycles. Chipmakers, equipment suppliers and cloud providers all price policy risk into their roadmaps, and the direction of travel in Washington has oscillated between restriction and relaxation for years. An argument that ties export policy to a testable premise — how soon superintelligent systems could actually arrive — at least gives the industry a framework for anticipating which way the rules will move next.

How the administration weighs that framework against the counterargument about irreversible strategic loss will determine whether the current loosening marks a durable policy shift or another swing of the pendulum in the fight over advanced compute.