The Council on Foreign Relations has delivered a blunt assessment of the new US policy governing AI chip exports to China, calling the framework "strategically incoherent and unenforceable" in an analysis published this week.

The critique lands at a moment when export controls have become the central instrument of US semiconductor policy toward Beijing. Washington has spent years tightening restrictions on advanced AI accelerators and the tools needed to produce them, aiming to deny China the compute required for frontier artificial intelligence development. The CFR analysis argues that the latest iteration of this policy fails on both of its own terms: it does not add up as strategy, and it cannot be enforced in practice.

Those are two distinct charges, and the CFR piece presses both. The strategic objection holds that the policy's goals are internally inconsistent — that the measures as designed do not coherently serve the stated aim of constraining China's AI progress. The enforceability objection is narrower and arguably more damaging: a control regime that cannot be policed at the point of implementation provides the appearance of restriction without the substance.

For the semiconductor industry, the analysis speaks to a debate that has shadowed every round of controls since the first major restrictions on advanced chip sales to China. US chipmakers have watched their addressable market in China — historically among their largest revenue pools — narrow with each successive rule. Nvidia, AMD and other designers of high-end AI silicon have repeatedly had to rework product lines to comply with new performance thresholds, creating stripped-down variants for the Chinese market while Washington debates whether even those downgraded parts should be permitted.

The enforcement question cuts across the entire supply chain. AI accelerators are small, high-value components that move through global distribution channels with many intermediaries. Analysts have long warned that determined buyers can route acquisitions through third countries or gray-market resellers, and reports of restricted chips surfacing inside China have circulated since the earliest controls took effect. A policy judged unenforceable by a mainstream foreign-policy institution gives that concern institutional weight it previously lacked mainly in industry commentary.

The strategic-incoherence charge is the more sweeping of the two. It suggests the policy lacks a clear theory of victory — that restricting chip sales neither halts China's AI program nor preserves US competitiveness in a coherent way. Critics of earlier control rounds made similar arguments, contending that restrictions accelerate Chinese efforts to develop domestic alternatives while costing US firms revenue that funds their own research and development. The CFR analysis now places that argument in front of a policymaking audience rather than a purely industry one.

The piece arrives as pressure builds on multiple fronts. Chip designers continue pressing for clarity on what they may sell into China. Equipment makers face their own control regime governing fabrication tools. And allied governments are negotiating how far their companies must follow Washington's lead, since unilateral US restrictions lose force if advanced silicon and tooling can flow through third countries.

The Council on Foreign Relations does not set export policy, and the administration has shown no sign of retreat from controls it frames as essential to national security. But an assessment from a pillar of the foreign-policy establishment that the current approach is both incoherent and unenforceable is likely to sharpen congressional and industry debate over whether the next revision tightens enforcement, relaxes the rules, or restructures them entirely — with direct consequences for which AI chips US firms can ship to one of their largest markets.