AINEWS
Behind the $265B US chip factory pledge is a $65B track record of two finished fabs and a CEO who won't commit to a schedule.
TSMC put a $265 billion number on US chip-factory expansion this week, then on the same Q2 2026 earnings call declined to give a firm schedule for any of the new sites. "If you ask me to give you a firm schedule, no, we don't have it today, but we do have a plan," CEO CC Wei told analysts, per the Yahoo Finance earnings transcript. A public capex pledge and a binding construction schedule are different documents. The $100B tranche TSMC announced this week brings the cumulative US pledge to $265B across 12 sites, The Register reports. Reading the next 18 months of news as evidence, not as more pledges, requires a working test.
The test starts with what TSMC has already built. Since the first Arizona fab was announced in the first Trump administration, TSMC has finished two facilities, broken ground on a third, and spent about $65B, per The Register's compilation. The first Arizona fab came online in late 2024 with Apple and Nvidia as the headline customers. The second, a 3nm node, is not scheduled to come online until the second half of 2027, per TSMC's own Arizona documentation. Volume production at the third Arizona fab is described in the same documents as "end of the decade."
The March 2025 tranche is the closest analogue for the $100B announced this week. That earlier round promised $100B of US fab expansion on top of prior commitments and landed alongside tariff threats, The Register reported at the time. Eighteen months later, the additional 900 acres the project needed were only acquired earlier in 2026, and zero new facilities from that tranche are complete. The build-and-ramp cycle for a leading-edge fab runs four to five years per site. Even on a clean schedule, the $100B announced this week would not produce wafers until the 2030s.
The labor constraint is the second half of the test. A McKinsey analysis surfaced by the Los Angeles Times projects a 157,000 skilled-worker shortfall in the US chip industry by the time TSMC's third Arizona fab reaches volume production. That is a forecast, not a realized shortage, but it lands directly on the buildout the new $100B tranche assumes.
The Intel precedent is the closest comparison for the same dollar signs and the same four-to-five-year clock. Intel committed $30B to Arizona, €30B to Magdeburg, $25B to Israel, and $20B to Ohio, per The Register's tally of the chipmaker's stalled projects. Arizona delivered one of two planned fabs. Magdeburg was cancelled in July 2025. Ohio was pushed into the 2030s. Israel was delayed indefinitely. Same industry, similar timelines, similar dollar signs, different outcomes.
The Trump administration framed the new $100B tranche as part of a separate $100B federal semiconductor commitment secured this month. Tax credits and federal underwriting change the math on which sites get built first, but they do not shorten the four-to-five-year fab cycle.
So what counts as evidence the $265B is becoming a schedule. Simultaneous equipment orders at multiple new sites. An earlier-than-H2-2027 first-wafer date for the 3nm Arizona fab. Large-scale Arizona hiring that lands ahead of the McKinsey shortfall curve. And the negative side: a Magdeburg-style cancellation, deferred tool moves, or renegotiated tax credits. The Q2 2026 print, with revenue above $40.2B and profit above $22B per the Yahoo Finance transcript, is the cash flow that makes any of those schedules possible. Wei's refusal to commit to one is the live signal that the headline number is not yet a plan.