A new report from Semafor documents what the semiconductor industry has long preferred not to tabulate: the American chip sector remains heavily dependent on China, despite four years of export controls and subsidy programs aimed at decoupling the two markets.

The finding lands awkwardly for Washington. The United States has spent the export-control era treating China as a strategic risk to be managed out of the supply chain. The report suggests the dependence runs in the opposite direction of official rhetoric, with US chipmakers still leaning on Chinese demand, Chinese manufacturing, and Chinese inputs at multiple points in the chain.

Semafor's publication of the analysis gives the industry's most uncomfortable structural fact a fresh citation. China is simultaneously the largest end market for semiconductors, a major assembly and test hub, and a growing supplier of mature-node capacity. American firms that design chips often cannot name a path to their revenue targets that excludes Chinese customers.

Why does the dependence persist?

Export controls have restricted specific advanced technologies, most notably leading-edge AI accelerators and the tools needed to make them. They have not restricted the bulk of the trade. Mature-node chips, automotive silicon, and consumer components continue to flow across the Pacific in both directions.

Chipmakers have also resisted a hard split for commercial reasons. Building parallel supply chains for two markets doubles cost, and no chief financial officer has volunteered for that experiment. The result is a sector that complies with the letter of the restrictions while keeping every commercial artery that the rules leave open.

What does the report change?

On its own, nothing operational. The report is an accounting of dependence rather than a policy proposal. Its value is in forcing the number into the open, where legislators and analysts can argue about it.

The timing matters. Washington is still distributing Chips and Science Act incentives, still tightening tool export rules, and still debating how far controls should extend down the technology ladder. Each of those debates assumes the industry can be steered away from China. The report's central claim is that, so far, it has not been.

Who carries the risk?

The dependence cuts both ways, which is precisely why it persists. US firms risk losing their largest market; Chinese firms risk losing their most advanced suppliers. Neither side has found a comfortable way to walk away from the arrangement, and both have spent the past several years paying for insurance they have not yet had to use.

For investors, the report is a reminder that US chip equities carry an unpriced geopolitical line item. For policymakers, it is evidence that subsidies alone do not redirect trade flows that remain profitable.

The next test will come from the enforcement calendar rather than the publication calendar. If Washington widens controls to cover more of the mature-node trade the report describes, the dependence it documents will begin to shrink on paper — and, more slowly, in the factories.