ASML raises 2026 sales outlook to €45 billion

ASML raised its 2026 sales outlook to €43-45 billion, up from €36-40 billion, because the companies that manufacture AI processors and memory chips are ordering more of its lithography machines, the tools that print circuit patterns onto silicon wafers (ASML, CNBC). That upgrade, roughly €5 billion at the midpoint, changes the revenue path of a company no advanced chipmaker on the planet can work without.

The reason is straightforward. AI models require enormous computing power. That drives demand for the most advanced processors and memory chips. The foundries and memory makers that build those chips (TSMC, Samsung, SK Hynix, Micron) must expand production lines, and expanding means buying ASML's equipment (Economic Times/Reuters).

Why ASML Has Pricing Power, Not Just Demand

Almost all expanded capacity for ASML's flagship EUV (extreme ultraviolet, its most advanced lithography technology) tools through 2027 is already spoken for. The company plans to raise 2027 production by 30% for both EUV and its older DUV (deep ultraviolet) systems (ASML, Biz Chosun). Expected gross margins climbed to 54-56%, from 51-53% (Yahoo Finance/Reuters). When your order book is full years out, you can charge accordingly.

Revenue from service and upgrades on already-deployed machines hit about €2.8 billion in Q2, roughly €300 million above expectations (Investing.com). Existing customers are paying more to squeeze extra output from tools they already own. This is a spending cycle in which chipmakers commit real money now for factories whose chips will ship later.

Four Countries, Four Layers of the Same Boom

ASML is headquartered in Veldhoven, but the revenue spreads across borders. In Germany, Carl Zeiss SMT supplies the precision optics at the heart of EUV systems (FAZ, Dr. Web). More machines sold means more Zeiss optics shipped. Berlin is also subsidising chip manufacturing directly: the TSMC factory in Dresden received around €5 billion in state support (ZEIT).

In Belgium, imec, the independent research hub in Leuven, runs pilot lines where chipmakers test new processes on ASML equipment before committing to mass production. A five-year agreement puts ASML's full tool portfolio inside imec's NanoIC pilot line (imec, Data Center Dynamics). In Ireland, Intel announced a €5 billion upgrade of its Leixlip fabrication campus for AI and high-performance computing (RTÉ, Reuters).

The Netherlands integrates the machine, Germany supplies precision inputs, Belgium validates next-generation processes, Ireland manufactures chips. Each captures a different layer of the AI hardware boom.

Where Europe's Grip Loosens

ASML's customers are almost entirely outside Europe. Most chips made with its machines are designed, packaged and deployed in Asia or the United States. Europe builds the tool nobody else can make, but the highest-value AI profits sit with chip designers like Nvidia, cloud platforms like Microsoft and Amazon, and the deployment businesses built on top of them. Europe captures the bottleneck, not the full value chain.

The order book is also concentrated. A handful of buyers drive most of the revenue. Each machine costs hundreds of millions of euros. A delayed order from one major customer can swing quarterly results.

Export controls add political risk. ASML says it has never delivered EUV machines or EUV-related parts to China (FAZ), but China still accounts for roughly 20% of expected 2026 sales through older DUV tools and services (ASML, Economic Times/Reuters). Tighter US-led restrictions could shrink that share, and the Netherlands does not set those rules alone.

Infrastructure is another constraint. Intel has flagged high electricity and construction costs in Ireland; an Irish Times editorial argued that energy, water and housing delivery is the real test for hosting projects at this scale (Irish Times).

ASML's upgraded outlook confirms that Europe dominates one of the hardest-to-replicate layers in the global AI economy, generating revenue and supplier demand across four member states today. It has not yet turned that dominance into broader AI industrial depth. The gap is measured in factories, skills and energy grids, and closing it depends on decisions that Brussels, Berlin, Dublin and The Hague have not yet made.