ASML’s China surge meets a broader AI labour warning

A big percentage, a smaller share

ASML says its chip-equipment sales to China rose 109%, even as the US pushes to curb such exports. The company expects China to account for about 20% of total sales this year. That combination points to a market that is still large, still buying and still moving through the export-control perimeter.

A 109% increase is a ratio, not a volume. It tells us the change relative to the earlier level, not the absolute number of tools shipped. And a 20% expected share means four-fifths of ASML’s sales still sit outside China, even if China remains a large single market.

The rise is not a clean signal of underlying demand. A percentage gain can be magnified by a small prior base, and buyers can pull orders forward when they expect restrictions to tighten. The 20% share may also look smaller than an earlier peak if China previously accounted for more. The number, in other words, is loud but not self-explanatory.

This is why the phrase “despite” needs caution. A rise during an export-curbing push may show demand, but it may also show timing. If customers expect fewer sales later, current deliveries can rise before the gate closes. The same invoice can look like strength in one quarter and pull-forward in the next.

A surge can therefore mean confidence in demand, fear of cutoff, or both at once.

Raimondo moves the risk from tools to jobs

Former US commerce chief Raimondo warned that rapid and unbridled AI development has the potential to disrupt the US economy in a way comparable to the China shock that cost jobs. That is not a comment on ASML’s order book. It turns the lens from supply chains to labour markets and policy response.

Her warning is about domestic economic risk: what happens when a technology spreads faster than workers, firms or institutions can adjust. It does not require a judgment on whether export controls are working. It asks a different question: whether the US can absorb AI’s economic effects without repeating the job losses associated with the earlier China shock.

The accompanying data include two instrument readings: INFO_SECTOR_EMP at 2,739 on 1 September 2026 and SEMI_PPI at 28.472 on 1 August 2026. Without units, history or direction, those figures cannot say whether conditions are tightening or loosening. They are markers, not evidence of a trend.

The names suggest labour and prices, but the values carry no context. A single point can be noise, especially in a fast-moving chip cycle. If the series are indexes, the reader needs a base period; if they are prices or employment measures, they need units and a comparison. As presented, they add precision without a baseline.

The bottleneck is policy, not just lithography

The physical chain is straightforward. ASML sells equipment used to make chips; when buyers in China purchase more tools, they are buying capacity for future output. But the commercial flow is now filtered through export rules and customer expectations of future constraints. A surge can therefore mean confidence in demand, fear of cutoff, or both at once.

Raimondo’s warning broadens the frame again. If AI can repeat a China-style job shock, the policy question is not only who sells advanced tools, but who bears the cost when production shifts. That links chip strategy to employment and political tolerance for disruption.

For investors and policymakers, the useful conclusion is narrower than the headline. The ASML figures show China remains a significant buyer. Raimondo’s remarks show anxiety about AI’s domestic economic effects. The two belong to the same China-shock story only if trade controls, industrial capacity and labour markets are read together. The open question is whether policy can move fast enough to affect any of those before the next capacity wave arrives.