A week of nervous AI trading ended on Thursday with the numbers from the company that makes most of the world's advanced chips. Taiwan Semiconductor Manufacturing Co. reported record third-quarter profit, topped its own revenue guidance and raised its forecast for the current quarter, while its chairman said the gap between AI demand and supply is still quite large.

It is the cleanest reading yet of whether the AI build-out is slowing. On Oct. 15, at least, it was not.

The Quarter

TSMC's revenue reached NT$1,494.24 billion, or $46.72 billion, up 41.2% in US dollars from a year earlier and 16.2% from the second quarter, according to the company's filing with the US Securities and Exchange Commission. That was above the $44.6 billion to $45.8 billion range it had guided in July.

Net income rose 68.7% to NT$763.06 billion, and diluted earnings per share were NT$29.43, or $4.60 per US-listed share. CNBC noted that it was a sixth straight quarter of record profit and that results beat analyst estimates.

Margins and the 2-Nanometer Ramp

Gross margin was 66.4%, within the 65% to 67% guided range but 1.3 percentage points lower than in the second quarter, which one analysis attributes mainly to the start of mass production of the 2-nanometer process. Operating margin was 58.7%, above the 56% to 58% range TSMC had forecast.

The technology mix shows where revenue comes from. Shipments of 3-nanometer chips made up 33% of wafer revenue, 5-nanometer 28%, 7-nanometer 10% and the new 2-nanometer 8%, according to the company's release. Customers are paying for the newest, most advanced manufacturing, and TSMC is selling it at margins few manufacturers match.

Guidance for the Fourth Quarter

TSMC expects fourth-quarter revenue of $52.3 billion to $54.3 billion, with a gross margin of 65% to 67% and an operating margin of 56.5% to 58.5%. At the midpoint that is about 14% growth over the third quarter, according to a summary of the call. For the full year, summaries put TSMC's expected revenue growth at about 44% in US dollars, and its capital spending budget remained $60 billion to $64 billion.

Capital spending is itself a signal. A company that plans to spend $60 billion or more on equipment in a year does so only when it is confident the plants will be full.

The Supply Gap

High-performance computing, the category that includes AI accelerators, rose 22% from the previous quarter to 68% of revenue, TSMC said. Chairman and CEO C.C. Wei told investors that the gap between demand and supply remains still quite big and that the company is doing everything it can to narrow it, according to BigGo Finance's report on the call.

CNBC added that TSMC pointed to growing adoption of agentic AI and open-source AI models as drivers, and to rising demand for processors and networking chips that support AI applications.

Reading It Against This Month's Jitters

The report lands after a rough fortnight for chip stocks. A report on OpenAI's revenue unsettled the sector earlier in the month, and hot inflation data pushed semiconductor shares down for six straight sessions. Blogerroom reported that the Nasdaq had hit a record on chip strength in September, and that Oracle had flagged delays at a data center in its Stargate project, the sort of customer-side hiccup that investors have been watching. The Nasdaq's earlier record run showed how much was priced in.

My view: TSMC's numbers should calm fears about demand, but they do not answer the question that unsettled markets. TSMC sells wafers to chip designers, who sell to cloud companies and AI developers. If trouble comes, it will appear first with the customers who must finance data centers at 5% Treasury yields, and only later in TSMC's order book. A quarter of record results tells you the chain is moving. It does not tell you whether every link can pay.

What to Watch Next

First, 2-nanometer margins: the technology will dilute profitability while it ramps, and TSMC has guided that to continue. Second, customers' spending plans, especially from the largest AI developers and cloud providers. Third, TSMC's own warning that supply cannot yet meet demand: if it is right, constraints on capacity, not demand, will set the pace of AI growth through next year. The next quarterly report is due in January.

This article is general information, not financial advice.