TSMC reported $46.7 billion in revenue for the third quarter of 2026, up 16.2 percent from the previous quarter, according to figures presented on its earnings call. The company then guided fourth-quarter revenue to a 58 percent jump year over year. The reason, in the company's own framing, is simple: AI demand is outrunning supply.

That combination — a record quarter plus an accelerating guide — is the rare kind of report that requires no editorial decoration. It is the strongest quarterly performance the foundry has disclosed this cycle, and the forward number suggests management expects the acceleration to continue rather than plateau.

What is driving the numbers?

The earnings call attributed the growth to artificial intelligence demand that the company cannot fully serve. Supply, not orders, is the binding constraint. For a foundry that sells manufacturing capacity rather than finished products, an explicit admission that demand exceeds supply signals two things at once: pricing power today and a rationale for continued capital spending tomorrow.

The 16.2 percent quarter-over-quarter gain is the more telling of the two figures. Sequential growth of that scale, on a revenue base already in the tens of billions, implies volume and price moving up together. The 58 percent year-over-year guide for the fourth quarter extends that curve into the December period.

Why does the guide matter more than the quarter?

Quarterly results confirm what already happened. Guidance reveals what management believes about the next three months, and TSMC's belief is apparently robust. A 58 percent year-over-year surge, committed to in an official outlook rather than floated as an aspiration, puts the company on track for a fourth quarter well above the $46.7 billion just reported — assuming the comparison base and seasonal patterns hold.

Investors tend to punish foundries that guide aggressively and deliver modestly. The willingness to publish this figure anyway is itself information.

How does this fit the AI supply picture?

TSMC sits at the center of the AI hardware supply chain: the accelerators that train and run large models are fabbed on its leading nodes, and hyperscalers compete for allocation. When the company states that AI demand outpaces supply, it is describing an industry-wide shortage of advanced wafer capacity, not a scheduling inconvenience at one plant.

The shortage explains the revenue. When customers cannot get enough wafers, they bid for the ones that exist, and the foundry's realized prices rise along with volumes. The 16.2 percent sequential figure is the arithmetic footprint of that dynamic.

What comes next?

The question for the quarters beyond Q4 is whether capacity additions catch up with demand or whether the gap persists into 2027. On the evidence of this call, TSMC expects demand to keep running ahead of what it can ship — and, on past form, it intends to charge accordingly.

By Nathan Brooks