TSMC reported third-quarter 2026 revenue of NT$1,494.24 billion, or US$46.72 billion, on 15 October, according to its filing with the US Securities and Exchange Commission. Revenue rose 41.2% from a year earlier and 16.2% from the second quarter in US dollars, and net income reached NT$763.06 billion, up 68.7% year on year. TheStreet reported that the stock nonetheless fell 4.14% that day, citing a sequential decline in gross margin and a broader sell-off in AI-related stocks. A record quarter and a falling share price are not contradictory once you read the filing against what TSMC itself guided three months ago. This piece does that, and is explicit about what the filing leaves out.

What the filing shows

The headline numbers come from TSMC's Form 6-K. Gross margin was 66.4%, operating margin 58.7% and net margin 51.1%. Diluted earnings per share were NT$29.43, equivalent to US$4.60 per ADR unit. Gross profit was NT$992.38 billion and income from operations NT$876.45 billion. One footnote matters for how much weight to put on any of this: the 6-K states that the third-quarter figures have not been approved by TSMC's board. The release does not say whether they are audited. Final numbers can differ from these.

Against TSMC's own July guidance, the quarter was a beat. Xenospectrum reports that the July outlook was revenue of US$44.6 billion to US$45.8 billion at an assumed exchange rate of NT$32 per US dollar, a gross margin of 65% to 67% and an operating margin of 56% to 58%. Actual revenue of US$46.72 billion was about 2.0% above the top of that range, and the implied exchange rate in the filing, about NT$31.98 per dollar, was almost identical to the guidance assumption, so the beat is not a currency effect. Gross margin of 66.4% landed above the 66% midpoint and operating margin of 58.7% above the 57% midpoint. On its own guidance, then, TSMC beat on every line.

The growth arithmetic: what the Q4 guide implies

For the fourth quarter, TSMC guided revenue of US$52.3 billion to US$54.3 billion, a gross margin of 65% to 67%, an operating margin of 56.5% to 58.5%, at an assumption of NT$31.8 per dollar. Measured against Q3's US$46.72 billion, the revenue range implies sequential growth of 11.9% to 16.2%, with a midpoint of 14.1%. That is where a deceleration reading comes from: even the top of the range only matches the 16.2% TSMC just delivered, and the midpoint is two points below it.

But a fair reading has to account for how TSMC guides. Second-quarter revenue in dollars works out to about US$40.2 billion by our calculation from the reported 16.2% growth, so July's range implied Q3 growth of roughly 10.9% to 13.9%, with a midpoint of 12.4%. The company then delivered 16.2%, about 2 points above the top of its own range. If Q4 outperforms the top of its range by the same margin, revenue would be roughly US$55.4 billion, or about 18.6% above Q3. On that pattern the Q4 midpoint of 14.1% is actually a higher implied growth rate than July's Q3 midpoint of 12.4%. We do not claim TSMC will repeat the beat; we only note that comparing a guide with an outcome overstates the slowdown, and that the market was probably comparing the guide with expectations that were higher than either.

The margin path

Margin is the clearer source of disappointment. Gross margin was 67.7% in the second quarter, a figure that TSMC's 6-K does not state directly but that follows from the table (gross profit of NT$860.31 billion on revenue of NT$1,270.38 billion) and that TheStreet also cites. It fell to 66.4% in Q3, a drop of 1.3 points, and the Q4 guide of 65% to 67% has a midpoint of 66.0%, a further 0.4 points lower. Operating margin tells the same story: 60.3% in Q2, 58.7% in Q3 and a guided midpoint of 57.5% for Q4, 1.2 points below Q3.

Lower margins on higher revenue still mean more profit in absolute terms. At the Q4 midpoint of US$53.3 billion and a 66% gross margin, gross profit would be about US$35.2 billion, against roughly US$31.0 billion in Q3, a rise of about 13%. Profit dollars are still growing at a double-digit pace. What changed is the rate at which each extra dollar of revenue converts to profit, which is the number an equity valuation built on a premium margin is most sensitive to.

Net income grew only 8.0% on the quarter even though gross profit rose 15.4% and operating income rose 14.3%, according to the 6-K. The gap sits below the operating line. Pre-tax income rose just 5.6%. By our subtraction of operating income from pre-tax income, non-operating items contributed about NT$34.6 billion in Q3 against an implied figure near NT$96 billion in Q2 (we derive the Q2 pre-tax figure from the reported growth rate, so treat it as approximate). Xenospectrum notes that Q2's net margin of 55.6% was lifted by roughly NT$63 billion of non-operating gains. Part of the apparent profit slowdown, in other words, is a tough comparison with a flattered quarter, not weaker core operations.

What the node mix says, and what nobody has explained

The 6-K gives wafer revenue by technology. In Q3, 2-nanometre was 8%, 3-nanometre 33%, 5-nanometre 28% and 7-nanometre 10%, so 7-nanometre and more advanced nodes made up 79% of wafer revenue. Xenospectrum reports that 2-nanometre was only 3% in Q2, 3-nanometre 30% and 5-nanometre 33%. If those earlier figures are right, 2-nanometre grew from 3% to 8% of wafer revenue in one quarter, while 5-nanometre fell five points. A new node typically carries a lower margin early in its life because yields and utilisation are still improving. Xenospectrum, summarising the July call, reports that chief financial officer Wendell Huang said the 2-nanometre ramp would weigh on gross margin by about three to four percentage points, partly offset by productivity gains and cost reductions, and that overseas fab start-up costs also play a part.

That explanation is plausible and consistent with the numbers, but we are relaying it second-hand, and the 6-K itself does not attribute the margin move to any cause. The filing refers to demand for leading-edge technologies and a steep 2-nanometre ramp expected to support Q4, and it stops there. The same source puts the 2026 capital expenditure plan at US$60 billion to US$64 billion, raised from US$52 billion to US$56 billion, with 70% to 80% for advanced processes; the 6-K extract we reviewed contains no capex figure, so treat that as reported, not confirmed here.

Reconciling record results with a falling price

Putting the pieces together gives a more modest story than a miss. TSMC beat its own guidance on revenue, gross margin and operating margin. The Q4 guide implies slower sequential growth than Q3 delivered and a gross margin about 0.4 points below Q3 at the midpoint, but TSMC's recent habit is to beat the top of its range. The share price fell 4.14% on a day TheStreet describes as a broad AI-stock sell-off, so some of the move cannot be attributed to TSMC at all. Our analysis cannot separate the market-wide component from the company-specific one, and no figure in the filing does either.

The questions we would hold open are about the next two quarters, not this one. How quickly does the 2-nanometre share rise, and does the dilution on gross margin deepen before yields improve? Does the Q4 revenue beat the top of the range, as Q3 did? And when TSMC's board approves the final numbers, do they match the preliminary ones? Those are the points on which a reader can check TSMC's next report against this one.