TSMC upgrades Central Taiwan fab

Semiconductor manufacturing is turning into a geography story as much as a technology story. That’s the real news here. TSMC is not just pushing its roadmap forward in Taiwan — it’s also rearranging where different classes of chips get made, and how quickly older fabs can be pulled upmarket. The result is a clearer picture of what the next few years look like: Taiwan stays the center of gravity for the very newest nodes, while the U.S. buildout gets bigger and more strategic. (trendforce.com) ### What changed in central Taiwan? The immediate move is a reported upgrade of a central Taiwan fab that had been associated with 28/22nm production, shifting it toward 4nm manufacturing. That matters because 4nm is not “bleeding edge” anymore, but it is still premium cap(trendforce.com)ll serving customers that need lots of mature-but-fast silicon. (trendforce.com) ### Where does 1.4nm fit? TSMC’s next major node after N2 is A14 — its 1.4nm-class process. TSMC itself has said A14 is on track for volume production in 2028, with gains of 10% to 15% speed at the same power, or 25% to 30% lower power at the same speed, plus more than 20% logic density improvement over N2. The reporting around central Taiwan lines up with that roadmap, pointing to trial or risk production in 2027 at Fab 25 in Taichung. (tsmc.com) ### Why reuse an older fab at all? Because fab construction is slow, brutal, and expensive. If TSMC can convert an existing 28/22nm site into 4nm capacity, it gets useful output faster than waiting for a greenfield project. Basically, this is a ladder: older nodes move up one rung, leading-edge fabs move up another, and the whole network carries more demand without every problem needing a brand-new building. (trendforce.com) ### Is the U.S. piece really that big? The official number today is $165 billion. TSMC announced in March 2025 that it would add $100 billion to its existing $65 billion Arizona commitment. That expansion covers three new fabs, two advanced packaging facilities, and a majo(trendforce.com)ned for N2 and A16 by the end of the decade. (pr.tsmc.com) ### So where does the $250 billion number come from? That looks more like an industry extrapolation than a current TSMC commitment. TrendForce tied it to comments from TSMC executive Cliff Hou at the 2026 SelectUSA summit, where he said the company is prepared for growth from new business opportunities. But the company’s own public baseline remains $165 billion, not $250 billion. So the righ(pr.tsmc.com)e, but not yet formalized at that headline number. (trendforce.com) ### Why does this matter for customers? Because the scarce thing is no longer just the smallest transistor. It’s the whole stack — wafers, advanced packaging, power delivery, proximity to customers, and the ability to ramp without bottlenecks. TSMC’s own symposium material now (trendforce.com)t just announcing a better node name. (tsmc.com) ### Does Taiwan still matter most? Yes. The central Taiwan upgrade actually underlines that. A14 remains tied to Taiwan’s manufacturing base, and TrendForce notes that 1.4nm is not part of the current U.S. investment plan. Arizona is getting more advanced, but Taiwan is still where TSMC is concentrating the earliest frontier production. (trendforce.com)2nm-fab-to-4nm-phase-2-1-4nm-trial-production-may-begin-in-3q27/)) ### Bottom line? This is TSMC widening the funnel. Central Taiwan gets more useful right now with 4nm, Taichung stays the launchpad for A14 in 2027-2028, and the U.S. expansion keeps growing around that core. The catch is that the strategic race now lives in orchestration — not just who can print the tiniest features first. (trendforce.com)