Taiwan Semiconductor Manufacturing Company added $28 billion in market capitalization across three trading sessions this week, reaching $982 per ADR on Thursday before settling at $971. The move follows confirmation that TSMC's Arizona Fab 21 achieved first 3-nanometer wafer production in December, two quarters ahead of internal guidance. Flputnam Investment Management disclosed a 1.2-million-share position increase in quarterly filings, part of a broader reallocation by multi-family offices and sovereign wealth managers.

The rally reflects a fundamental repricing of geopolitical risk premium. TSMC controls 92 percent of the world's sub-5-nanometer logic production, all of it concentrated in Taiwan's Hsinchu and Tainan campuses until this quarter. The Arizona facility now produces 20,000 wafer starts per month at the N3E node, with a second phase targeting 100,000 monthly starts by late 2025. Japan's Kumamoto Fab 1 reached full 28-nanometer capacity in November, serving automotive and industrial clients previously dependent on cross-strait logistics. These are not symbolic hedges. They are live production assets reducing single-point failure risk in the semiconductor supply chain.

The timing matters. The U.S. CHIPS Act disbursed $6.6 billion in direct grants to TSMC in September, contingent on domestic production milestones now being met ahead of schedule. European allocators, particularly those managing exposure to German automotive and Dutch equipment manufacturers, have been rotating out of Samsung and Intel foundry proxies into TSMC on the thesis that only one company has demonstrated 80-percent-plus yield rates at the bleeding edge. Samsung's 3-nanometer GAA process remains stuck at 60 percent yields, per supply-chain data from Hsinchu equipment vendors. Intel's 18A node is sampling but not in volume production. The gap is widening, not closing.

What operators miss: TSMC's pricing power is compressing, not expanding. The company's N3 wafers are priced at roughly $16,000 per unit, a 12-percent discount to the N5 node's peak pricing in 2021 when hyperscaler demand was unconstrained. Gross margins dropped 180 basis points year-over-year in the most recent quarter, to 53.1 percent, as Arizona labor costs run four times Taiwan levels and Japan subsidies required margin concessions to secure government co-investment. The stock is not rallying on margin expansion. It is rallying because the alternative to TSMC is no alternative at all.

Allocators should track three events in the next six months. First, TSMC's April earnings call will disclose 2-nanometer customer commitments and whether Apple has locked volume at the new Hsinchu P2 fab. Second, the U.S. Commerce Department's next CHIPS Act tranche, expected in May, will signal whether Arizona Fab 22's 2-nanometer phase receives the additional $8 billion in federal backing TSMC has privately lobbied for. Third, any cross-strait incident that disrupts the Taiwan Strait shipping lanes for more than 72 hours will test whether 120,000 combined wafer starts per month in Arizona and Japan can sustain even minimal supply to hyperscalers and automotive Tier 1s. It cannot, but the market has not priced that gap.

TSMC's January revenue came in at $7.3 billion, up 18 percent sequentially and the highest January figure in company history. The stock trades at 28 times forward earnings, elevated but defensible if you believe compute demand holds and no credible competitor emerges at 3-nanometer or below in the next thirty-six months. The latter is not a bet. It is observable fact.