93% probability TSMC's Q3 2026 revenue reaches at least USD 43.0bn. The company guided USD 44.6-45.8bn, up 12% sequentially and 37% year-on-year at midpoint, on a steep 2-nanometre ramp. Paired deliberately with P-235 on ASML: same AI wave, different link in the chain.
TSMC guided third-quarter 2026 revenue of US$44.6bn to US$45.8bn, a 12% sequential and 37% year-on-year increase at midpoint, with gross margin of 65-67% and operating margin of 56-58%. Gross margin is expected to fall 1.7 percentage points on 2nm ramp-up dilution, partially offset by strong demand and cost improvements. Second-quarter profit rose 77% year-on-year, and June revenue rose 68%. The company raised its 2026 revenue guidance and announced an additional USD 100bn of investment in Arizona. Source: TSMC Q2 2026 results and Q3 2026 guidance.
We lock a binary: TSMC's reported revenue for the third quarter of 2026 is USD 43.0bn or higher. Confidence 93%.
Guidance is USD 44.6-45.8bn. Our threshold sits USD 1.6bn, or 3.6%, below the floor of that range.
Most companies leave a quarter dark until it is finished. TSMC publishes net revenue monthly, which means that by the lock date two of the three months of Q3 have not only happened but have largely been disclosed. A guidance-anchored lock on a quarter this far advanced is a much smaller claim than the same lock made in the first week of the period.
The demand side is not the question here. Second-quarter profit rose 77% year-on-year and June revenue rose 68%, on leading-edge logic demand that the entire AI build-out depends on. The real uncertainty is the 2-nanometre ramp, which management itself flags as diluting gross margin by 1.7 points — new nodes bring yield and throughput risk, and a stumble there shows up in shippable wafers.
The residual 7% is that ramp, plus the Taiwan dollar. TSMC reports in New Taiwan dollars and the widely cited guidance is in US dollars at an assumed exchange rate, so a sharp currency move can shift the US-dollar figure without anything changing in the fab. The scoring line below fixes which figure governs so this cannot be argued after the fact.
P-235 locks ASML's full-year net sales; this locks TSMC's third quarter. These are not independent draws and we will not present them as if they were. Both depend on AI capital expenditure. What differs is position in the supply chain and how much risk has already been retired: TSMC turns wafers on capacity that is already installed and is nearly through the quarter being forecast, while ASML must ship and have accepted a steep second-half ramp of new tools.
So the pair carries information the individual locks do not. Both HIT is the consensus outcome and says little. TSMC HIT with ASML MISS would say the AI build-out is currently being served by existing installed capacity running harder, not by new capacity being added — an early sign of a digestion phase in equipment. Both MISS would say the demand story itself has cracked, which nothing in the current data supports.
RAOSCAFF locks P-240 on 2026-08-29, before the Q3 release. Scored against consolidated revenue as reported by TSMC for the three months to 30 September 2026.
Two of the quarter's three months were already disclosed through TSMC's monthly revenue releases when this was locked. Correlated with P-235 by design; a split resolution is the informative one.