Two tracks, one still open
On July 24, 2026, the United States Trade Representative's final action in a set of forced-labor Section 301 investigations took effect, imposing additional duties of 10% or 12.5% on goods from 60 economies. Taiwan drew the most favorable outcome available: a 10% rate applied net of its existing most-favored-nation duty, the same treatment given to the European Union. Japan, South Korea, and Switzerland received the same net-of-MFN structure but at 12.5%; China and roughly 38 other economies face a flat 12.5% with no MFN offset (USTR; Focus Taiwan).
USTR credited Taiwan's commitments under its bilateral trade dialogue to enforce forced-labor import prohibitions — a real, if narrow, policy win for Taipei. But Taiwanese economists have been consistent that this is not the tariff track that matters. A separate Section 301 investigation, opened March 11, 2026, targets "structural excess capacity and production" across 16 economies, including Taiwan, China, Japan, South Korea, and the EU, and explicitly covers semiconductors alongside autos, steel, solar modules, and a dozen other sectors (USTR). Hearings ran through May and July; USTR's own timeline pointed to findings by the end of July. As of this writing, no determination has been published — the investigation is still pending, and Taiwan's chip sector is still waiting to learn whether it gets tariffed at all.
Why the probe worries Taipei more than the tariff did
Taiwan Institute of Economic Research economist Liu Pei-chen has said plainly that the overcapacity probe "remains the bigger concern" relative to the forced-labor tariff, even though Taiwan came out of the latter with the best rate on offer (Taipei Times; The Tribune). The reasoning is structural. TSMC's leading-edge nodes — the chips inside iPhones, Nvidia GPUs, and most AI accelerators — are not the products USTR is worried about; overcapacity concerns concentrate in mature-node production (28nm and above), the workhorse chips used in cars, appliances, and industrial electronics, where Chinese state-subsidized capacity has been depressing global prices for years. TSMC and Taiwan's mature-node foundries sit closer to that exposure than Taiwan's advanced fabs do.
The complication is that Taiwan barely exports finished chips to the US directly; most Taiwanese semiconductors leave the island embedded in laptops, phones, and industrial equipment assembled elsewhere. That means the real risk isn't a tariff line item on wafers — it's whether Washington decides to tariff finished electronics on the basis of the non-US-made chips inside them. Taiwan Institute of Economic Research president Chang Chien-yi has flagged exactly this scenario as carrying more economic weight than the forced-labor duties already in effect.
The case for the probe — and where it overreaches
The overcapacity investigation is not baseless protectionism dressed up in legal language. USTR's own initiation notice pointed to global manufacturing capacity utilization sitting between 75% and 75.9%, below the 80% level generally treated as healthy, and mature-node chips are a textbook case of a market distorted by non-market subsidy: Chinese foundries have expanded legacy-node capacity well beyond domestic demand, and the resulting price pressure genuinely threatens allied producers in Taiwan, Korea, and the US who compete on commercial terms rather than state support (Holland & Knight). An investigation aimed narrowly at that distortion — with remedies scoped to the actual source of the overcapacity — would be a defensible use of trade law.
That is not what USTR built. Bundling Taiwan, Japan, South Korea, and the EU into the same 16-economy investigation as China treats US allies whose foundries and automakers are themselves squeezed by Chinese overcapacity as part of the problem to be solved, rather than partners in solving it. Taiwan's mature-node exposure is a byproduct of proximity to a genuine Chinese distortion, not evidence that Taiwan itself is flooding the US market. A tariff regime that cannot distinguish between the two risks punishing the exact firms — TSMC and its mature-node peers — that the US has spent the past five years courting into Arizona fabs precisely because Washington wants supply chains that don't run through Beijing.
What to watch
Three things determine whether this becomes a real cost for Taiwan's chip sector rather than a paper threat. First, whether USTR's eventual determination distinguishes mature-node exposure by origin of subsidy rather than applying blanket sectoral tariffs. Second, whether any resulting duties attach to finished products containing non-US chips — the scenario Taiwanese officials view as most damaging, since it would reach far beyond direct wafer exports. Third, the timeline itself: a probe that missed its own end-of-July target leaves Taiwanese exporters pricing contracts into 2027 without knowing their tariff exposure, which is its own form of drag on investment decisions independent of whatever rate eventually lands. Congress and USTR should treat a fast, narrowly scoped resolution as the minimum bar for a policy aimed at a real problem — indefinite uncertainty for an ally's core export industry is not a cost-free way to pressure China.