Publication date: 14 October 2026
ASML says demand for its semiconductor-manufacturing equipment will exceed available capacity through 2028, with customers already discussing orders for 2029 and 2030. The warning shifts attention away from the most visible AI chip designers and towards a deeper constraint: the highly specialised machines required to turn advanced designs into physical chips.
Demand is running ahead of expansion
The Dutch company reported third-quarter net sales of €11.7 billion and net income of €5.0 billion. It expects fourth-quarter sales of €14.2 billion to €15.2 billion and lifted the lower end of its full-year outlook, forecasting 2026 sales of €44 billion to €45 billion.
ASML had previously said it planned to increase capacity by 30% in 2027 and by a further 30% in 2028. Chief executive Christophe Fouquet now says that will not be sufficient to meet customer demand. The company is considering additional measures while upgrading existing tools and introducing higher-throughput systems to help chipmakers produce more from installed equipment.
Its customers include TSMC, Samsung, Intel, SK Hynix and Micron. Their expansion plans are being driven by demand for processors and memory used in AI data centres. Some customers are seeking to reserve equipment years before new fabrication plants are ready.
Why lithography is different
Advanced chips require layers of microscopic patterns to be projected onto silicon. ASML dominates the market for the extreme-ultraviolet lithography systems used at the leading edge, and these machines are extraordinarily complex. Capacity cannot be added as quickly as software services or conventional server space because production depends on specialised components, precision engineering, qualified workers and a tightly coordinated supplier network.
That makes lithography an upstream choke point. A shortage of processors can prompt chip designers to place larger orders, but those orders do not create additional manufacturing capability unless foundries can obtain and install more production tools. The long lead time pushes strategic planning towards the end of the decade.
The boom contains its own risk
Strong forward demand gives ASML visibility and pricing power, but multi-year order discussions are not the same as guaranteed sales. AI infrastructure investment could slow, customers could postpone fabrication plants or export controls could restrict deliveries. ASML itself cautions that forward-looking plans depend on demand, supply-chain capacity, regulation and customers converting interest into orders.
There is also a coordination problem. Foundries do not want to invest too slowly and miss demand, but overbuilding could leave expensive factories underused if AI spending cools. ASML faces the same balance: expanding too cautiously prolongs scarcity, while expanding too aggressively could expose it when the cycle turns.
A strategic industrial constraint
The capacity warning matters to governments as well as investors. Countries offering subsidies for domestic chip plants may discover that finance and construction are not enough. Access to production equipment, maintenance expertise and supplier capacity can determine whether a new factory reaches commercial output on schedule.
The AI race is therefore not only a contest over models, data and chip design. It is also a contest over a small number of machines that take years to build and integrate. ASML’s order horizon to 2030 shows how quickly a digital boom can become an industrial planning challenge.