TSMC Posts Record Profit but Turns Cautious on 2027 as AI Shares Slide

TSMC posted another record quarterly profit on October 15 but warned that macroeconomic uncertainty warranted caution in planning for 2027, according to Reuters. That combination put a limit on the reassurance investors could take from the report during the broader AI-stock selloff: current chip demand remained strong, while management was not treating the next year as free of risk.

Net income rose 68.7% from a year earlier to NT$763.06 billion on revenue of NT$1,494.24 billion, the company reported. The results show business reaching the chip factory; they cannot by themselves establish the profitability of the applications those chips will eventually run.

Revenue in U.S. dollars was $46.72 billion, up 41.2% from a year earlier. The company’s reported revenue growth in New Taiwan dollars was 50.9%. Those are different currency presentations of the business, not competing estimates that can be substituted without explanation.

The next manufacturing step is already contributing

The company said 2-nanometer shipments represented 8% of quarterly wafer revenue, with 3-nanometer at 33%. Technologies at 7 nanometers and below accounted for 79%. Gross margin was 66.4%. The process mix offers a more concrete measure of advanced manufacturing activity than a forecast about how many people might eventually use AI software.

Process names are not a direct measure of the value of an entire finished system. Chip design, packaging, memory, networking and software all affect what customers can do with the silicon. The manufacturing mix is important evidence, but it is one part of a longer production chain.

A customer also needs the right chip in the right quantity at the right time. Capacity on a process that does not fit a particular design is not interchangeable with capacity on one that does. That makes the composition of a manufacturer’s output relevant alongside its headline revenue.

Guidance describes an expectation, not a booked result

TSMC forecast fourth-quarter revenue of $52.3 billion to $54.3 billion and a gross margin of 65% to 67%. That puts the revenue midpoint at $53.3 billion, about 14.1% above the third quarter, calculated from the company’s figures. The fourth-quarter numbers remain guidance, rather than revenue already earned.

The distinction matters when using supplier results to assess the AI cycle. Current sales support a claim that manufacturing demand has been strong. Guidance adds management’s view of what comes next. Neither removes the possibility of later changes in customers’ orders, capacity needs or financing.

For an infrastructure buyer, the business decision sits further down the chain. It must convert access to chips into computing services customers want, at prices that cover equipment, power, operations and capital. TSMC can earn a manufacturing margin before that downstream investment has delivered its ultimate return.

Currency comparisons need a consistent basis

An investor comparing TSMC with another supplier should keep reporting currencies and periods aligned. A change in exchange rates can make the same underlying sales growth appear different in dollar and local-currency presentations. Neither measure is inherently wrong; the analysis needs to identify which one it uses before drawing a comparison.

Strong factory results can coexist with weaker AI shares

That timing helps explain why a favorable supplier report does not settle every concern behind a market selloff. Investors in a manufacturer, a cloud operator and an AI application developer own different claims on different cash flows. They can reasonably reassess them in different directions.

A manufacturer’s quarterly growth also should not be read as proof that all of its sales are attributable to AI. Its technology serves multiple uses. The results provide evidence about advanced-chip production without supplying a complete breakdown of the eventual economic value created by each customer.

The next useful comparison is between the factory’s expansion and the pace at which customers deploy, use and pay for the resulting computing capacity. TSMC has delivered a strong quarter and expects another step up. Whether the rest of the AI industry earns enough to sustain that pace remains a separate question.