SEMI Doubled Its Own 2030 Forecast, From $1tn to Over $2tn

Speaking in Taipei on 31 August, with the item filed at 09:53 UTC, SEMI put the semiconductor market above $1.5 trillion in 2026 and above $2 trillion by 2030. Reported as a datapoint, it is a big number. Reported as what it is — a revision — it is a much stranger one.

The number that got replaced

SEMI said on the record that its own projection for 2030 was $1 trillion about eighteen months ago. So the forecast for a fixed year has doubled in a year and a half. That is not a market growing; that is a model being rebuilt. And the sharper detail is in the near term: the 2026 figure of over $1.5tn already exceeds the retired 2030 target by roughly 50%. The old forecast was not merely low — it was overtaken four years early.

The control group

Equipment forecasts over the same period moved only about 10–16%. That gap is the most informative thing in the release. If tool spending has barely re-rated while the device market has doubled, the revision is being driven by price and mix — accelerators and HBM carrying far more revenue per wafer — rather than by a belief that the industry will make proportionally more silicon. A doubling in dollars on a roughly flat capacity plan is a statement about average selling prices.

What the common framing gets wrong

Headlines ran "semiconductor market to top $2tn by 2030" as though a fresh measurement had arrived. It is a forecast, and forecasts from trade bodies are directional rather than audited. The load-bearing fact is not the $2tn; it is that the same body's number for the same year was half that recently enough to be embarrassing, and it said so itself. Treat the new figure with the accuracy the old one earned.

What to watch instead

Forecast revisions of this size usually resolve one of two ways: the demand proves durable and the equipment number catches up later, or pricing normalises and the dollar forecast retreats toward the physical one. The divergence between a doubled device forecast and a barely-moved equipment forecast is the thing to track, because only one of the two can be right for long.