Trump revived the idea of steep tariffs — 100% or even 200% — on Taiwanese chips, this time framing it as long-overdue rather than retaliation for a "stolen" industry. Taipei's calm response suggests it believes an existing MOU already carves out preferential treatment.
The strategic signal matters more than the number. Tariffs on semiconductors are a blunt instrument aimed at a precise goal: forcing capacity onto US soil. But chips are not commodities you can re-shore by decree. The leading-edge supply chain is a web of lithography, packaging, materials, and yield know-how concentrated in Taiwan and a handful of partners. A punitive tariff without domestic capacity simply taxes American buyers — from hyperscalers to automakers — and raises the cost of the AI buildout everyone else in today's headlines is racing to fund. Taiwan's confidence hints at a two-track world: firms that secured bilateral exemptions win pricing certainty, while everyone outside those deals absorbs the volatility. Expect accelerated "tariff arbitrage" — routing final assembly, IP ownership, and invoicing through jurisdictions with favorable terms.
The timing, just before Xi's expected US visit, turns chips into a bargaining chip. Trade rhetoric here is as much about the China negotiation as about Taiwan, which leaves the whole East Asian supply base exposed to decisions made over its head.
For Japan, this is not a spectator sport. TSMC's Kumamoto fabs, the Rapidus 2nm ambition, and Japan's dominance in equipment and materials — photoresists, silicon wafers, deposition and etch tools — mean any reshaping of chip trade flows lands directly on domestic balance sheets. If US policy pulls advanced capacity stateside, Japanese equipment and materials suppliers may benefit from a larger American fab footprint, but Japanese fabless and device makers face costlier, more fragmented sourcing. Tokyo's own subsidy strategy needs to assume tariffs are now a permanent variable, not a one-off shock.
For Japanese SIers and enterprise IT teams, the second-order effect is procurement risk. Server, GPU, and edge-device costs feed straight into cloud migration and AI project budgets; a tariff-driven price step-up can quietly blow up multi-year TCO models. The practical move now is scenario planning: build tariff sensitivity into hardware refresh and datacenter capex forecasts, diversify vendor and geographic exposure, and treat supply-chain geopolitics as a standing line item rather than an occasional headline.
