TSMC Posted Record $46.7 Billion Revenue. Thursday Will Show What the Market Priced In.

The world knows almost exactly how much money TSMC made last quarter, a full week before the company holds its earnings call. The Taiwanese chipmaker reported September revenue of NT$511.86 billion on October 8, which allowed Reuters to calculate third-quarter revenue of roughly NT$1.49 trillion, about $46.71 billion. That is the largest quarterly revenue in the history of the company, and roughly 50 percent more than the same quarter a year earlier.

Because TSMC discloses sales every month, the headline number rarely surprises. September alone was up 54.6 percent year over year, though it slipped 0.6 percent from August, and revenue for January through September reached NT$3.89873 trillion, up 41.1 percent from the first nine months of 2025. Even so, the quarter beat a consensus of 19 analysts compiled by LSEG and came in above TSMC’s own July guidance of $44.6 billion to $45.8 billion. When a company guides up and still beats its own bar, investors pay attention.

Thursday is when it gets interesting

The full results arrive Thursday, October 15, with an earnings call scheduled for 0600 GMT. Revenue is known; the profit figure is not. Analysts tracked by LSEG expect net profit to surge 65 percent to NT$746.3 billion, about $23.4 billion, which would be TSMC’s sixth consecutive quarter of record earnings and its 11th straight quarter of profit growth. Anything above NT$706.56 billion sets a new all-time quarterly record.

The call matters more than the profit line. Management will give its fourth-quarter outlook and, analysts hope, an updated view of the full year. In an October 6 report, Morgan Stanley analyst Charlie Chan wrote that he expects TSMC to raise its 2026 revenue growth forecast to the low-to-mid 40 percent range, from current guidance of more than 40 percent, driven by strong demand for its 3-nanometer chips. Guidance raises have become a ritual at these calls. The ritual continues only as long as demand does.

When the company that fabricates nearly every AI accelerator on Earth posts its best quarter ever, the debate is no longer whether the AI buildout is real. It is whether the stock market has priced the second half of it.

The demand side looks unbreakable, for now

The drivers are visible across the industry. TSMC is a key supplier to Nvidia and Apple, and its leading-edge nodes are where nearly every AI training and inference chip gets built. Samsung, the closest rival, projected its own milestone on the same day as TSMC’s revenue release: a quarterly operating profit above 100 trillion won, roughly $74.4 billion, driven by strong memory sales to AI data centers. That would be the first time a technology company has crossed that threshold in a single quarter.

Packaging is now as important as the chips themselves. TSMC’s advanced chip-on-wafer-on-substrate (CoWoS) packaging technology connects processors and high-bandwidth memory inside a single package, and it has become one of the industry’s binding constraints. Demand for CoWoS capacity has outrun TSMC’s ability to add it, which is why a five-year agreement announced this month for GlobalFoundries to make silicon interposers for TSMC in New York was treated as genuine industry news.

TSMC responded to the pressure in July by raising its 2026 capital expenditure forecast to between $60 billion and $64 billion, from an earlier range of $52 billion to $56 billion. Spending of that size is the company’s way of saying it believes current demand has years to run.

The expansion question comes to a head

Expect analysts to press management on the United States. TSMC has committed $265 billion to building factories in Arizona, including an additional $100 billion announced in July, and Reuters reported in September that the company is evaluating a potential investment in Texas. Opinions are split on whether Texas is urgent. Morgan Stanley’s Chan argues that limited room at the Arizona site within five years could make expansion necessary; Morningstar analyst Phelix Lee sees less urgency while some Arizona fabs are still unbuilt, describing Texas as a sensible location for a 2030s or 2040s expansion plan and possibly a diplomatic lever in trade policy.

The political dimension is real. Fabs are now strategic assets, and every TSMC investment announcement is read in two capitals at once. The company has navigated that pressure well so far, but the questions get harder as the spending numbers get bigger.

What to watch on Thursday

Three things matter beyond the profit number. First, gross margins: 3-nanometer yields and pricing will show whether TSMC’s most advanced technology is still getting more profitable as volumes scale. Second, the capital expenditure outlook for 2027, which is the clearest statement of management’s confidence in the cycle. Third, any comment on CoWoS capacity additions and the GlobalFoundries partnership, which together determine how quickly the packaging bottleneck eases.

TSMC’s Taipei-listed shares have risen about 63 percent this year, roughly in line with the broader market, and the stock dipped 1.35 percent on the day the revenue record landed. That reaction tells its own story. Monthly reporting removes the element of surprise from the sales figure, so the stock trades on what management says next, not on what the company already sold. For a company worth about $2.1 trillion, nearly twice Samsung’s market value, the bar for surprise is extraordinarily high. Thursday’s call is where it gets set for the next quarter.