ASML shares rose 4% on Wednesday after the company raised its annual sales guidance for the second time this year, pointing to customers pressing ahead with AI chip production capacity. Two upward revisions within a single year carry a different signal than one: demand may be running faster than ASML anticipated, or the original forecast left itself room to raise.

What's changed

The case for reading this as a genuine demand signal rests on frequency. Companies rarely revise full-year guidance upward twice in quick succession without sustained order flow behind the decision. ASML's customers, the businesses expanding AI chip capacity, are doing so at a pace that has moved the equipment maker's own forecast higher twice in a matter of months.

That matters for how capital reads the AI hardware buildout. Equipment suppliers sit early in the supply chain. Their revenue reflects purchase decisions customers made well before a chip ships, so a second guidance raise signals that the buildout is still accelerating, at least as ASML sees it from its order book.

The counterargument

The risk is that two upgrades within a year also reveal a company that did not see its end-market clearly when it set the original forecast. If the starting guidance was anchored too low, the raises tell investors less about demand strength and more about forecasting conservatism at ASML itself. The read-through for the broader AI chip supply chain becomes harder to trust.

There is also a sequencing question. Customers ramping production today are placing orders months before chips ship. If downstream AI spending slows, the equipment buildout would lag that turn, leaving ASML's numbers resilient even as the cycle peaks. Wednesday's 4% gain does not price in that lag.

On balance

Two guidance increases in one calendar year is a hard data point. The AI chip production expansion that ASML's customers are running is large enough to move the forecast twice over. The line to watch is whether a third revision materializes: that would shift the story from a strong cycle to something more structural. For now, the facts support the move. The counterargument on cycle lag is real enough to keep in the margin.