TSMC · one lens
Hamilton Helmer on TSMC
Microsoft’s N2P chip supports TSMC’s process case, while an interposer deal and Terafab’s separate plans leave the size of its cost barrier unproven.
Helmer’s 7 Powers asks whether a company has a Superior and Significant cash-flow Benefit, then whether a Barrier makes that Benefit Sustainable. TSMC’s record still supports Scale Economies and Process Power. The October reports sharpen the distinction: customers keep choosing its advanced process, while neither new spending nor a new contract measures the cost gap a challenger must overcome.
TSMC raised its 2026 capital-spending guidance to $60B–$64B from an earlier $52B–$56B range for the same year, per Bloomberg’s coverage. TSMC can spread fixed fab costs over its leading wafer volume; a smaller foundry would need comparable volume to match that unit-cost Benefit. The TEXXR record does not give matched unit costs or rival wafer volumes, however, so it cannot calculate Helmer’s surplus leader margin. GlobalFoundries’ October deal adds a useful boundary: GlobalFoundries agreed to provide TSMC silicon-interposer capacity under a five-year, $2B contract at its Malta facility. GlobalFoundries is supplying a packaging input, not demonstrating a rival advanced-wafer process.
The Verge reports that Microsoft’s newly unveiled Maia 300 uses TSMC’s N2P process and is intended to become Microsoft’s primary AI inference system. Microsoft’s design choice supports TSMC’s Process Power case, but the announcement does not disclose yields or prove a durable unit-cost advantage. Reuters reports that AMD chief Lisa Su wants more advanced wafer capacity as AMD plans to increase supply in 2027; AMD’s stated need is not a measurement of TSMC’s pricing power.
The Texas reports test a different claim. Culpium reports that TSMC executives met Texas officials about overseas operations in the first half of 2026. Bloomberg reports that Elon Musk ruled out a TSMC operational role in Terafab. Neither report establishes that Terafab can replicate TSMC’s process, and Culpium’s account does not establish a new TSMC fab.
The TEXXR coverage record counts 30 TSMC articles in the completed third quarter of 2026, against an eight-quarter baseline averaging 29.2 articles per quarter. Those counts measure indexed coverage, not wafer orders. TSMC faces its next earnings report on October 15; that report may offer a better test of the Benefit than article volume does.
The seven Powers, one by one
| Power | Status | Why |
|---|---|---|
| Scale Economies | held | 2026 capex guidance rose to $60B–$64B from $52B–$56B against TSMC’s leading wafer volume, though rival unit costs remain undisclosed |
| Network Economies | absent | not argued in the record |
| Counter Positioning | absent | not argued in the record |
| Switching Costs | absent | not argued in the record |
| Branding | absent | not argued in the record |
| Cornered Resource | absent | not argued in the record |
| Process Power | held | Microsoft chose TSMC’s N2P process for Maia 300, while the fresh reports show no rival replicating TSMC’s advanced-wafer operations |
An editorial application of Hamilton Helmer's published framework — not their actual view, affiliation, or advice.
Where this lens comes from
Hamilton Helmer's framework is set out at the lens page, drawn from 7 Powers. This page applies it to TSMC and nothing else — the company's full record is in the dossier.