VELDHOVEN (Realist English). Dutch semiconductor equipment giant ASML published its Q3 2026 financial report on October 14 and once again raised its full-year sales outlook. CEO Christophe Fouquet noted in a statement that customers continue to raise their medium- and short-term demand forecasts for ASML equipment, based on their assessment of the scale and duration of AI-related investments.
Third-quarter results
ASML’s net sales in Q3 reached €11.662 billion, exceeding €9.326 billion in the previous quarter and approaching the upper end of the guidance range. Net profit rose from €2.918 billion in the previous quarter to €4.96 billion, beating market expectations of €4.165 billion.
Gross margin expanded from 54% in Q2 to 57.5%, mainly due to installed base service sales of €3.3 billion, which was above guidance.
The company expects Q4 net sales of between €14.2 billion and €15.2 billion, with a gross margin of around 57%. The full-year 2026 net sales forecast is €44 billion to €45 billion, with a gross margin of around 56%.
Capacity — the key bottleneck
Fouquet sent a clear signal in the report: demand growth has outpaced capacity expansion. “We currently expect that, under current plans, our supplies in both 2027 and 2028 will not be able to fully meet customer demand,” he said. “Therefore, we are studying how to further increase capacity in 2028 beyond the previously announced 30% growth.”
ASML had previously planned to increase low-NA EUV system output from approximately 65 units in 2026 by 30% — to about 85 units in 2027 — and by another 30% in 2028, to about 110 units. However, Fouquet’s latest statement indicates that even with this expansion plan, it will be insufficient to cover demand. The company reported that some customers intend to place equipment orders for 2029 and even 2030.
Structural demand driven by AI
ASML’s growth driver is the ongoing expansion of AI infrastructure investment. Advanced logic chips, high-bandwidth memory (HBM), and related silicon wafer production capacity for AI data centers are pushing customers to accelerate capacity expansion.
Fouquet noted in the Q2 report that ASML expects revenue growth from advanced logic chips of about 25% in 2026, and about 75% from memory. This trend continued in Q3. Key customers — TSMC, Samsung, SK Hynix, and Intel — are expanding capacity for AI processors and advanced memory.
Commercialization of High-NA EUV technology is also progressing. Intel has already used this technology to process more than 1 million wafers in research, testing, and production, including certain layers of its Panther Lake processor. Samsung plans to introduce this technology into mass production of advanced DRAM in 2028.
China market: share declining, absolute volume growing
ASML CFO Roger Dassen confirmed in Q2 that the company still expects China’s share of 2026 net sales to be around 20%, as previously forecast. Although the share has declined compared to previous years, thanks to the raised full-year sales forecast, the absolute volume of sales in China will also be higher than expected at the beginning of the year.
ASML is still prohibited from exporting EUV systems and the most advanced DUV equipment to China, but the company continues to supply older generations of DUV for semiconductor production by Chinese chipmakers for domestic applications.
Proposed US legislation could further tighten export restrictions on China, creating a persistent business risk.
What this means
ASML’s report reflects a structural contradiction: AI-driven demand is real and sustainable, but the physical bottleneck on the supply side — from optical systems to precision motion components, from light sources to control electronics — cannot be overcome in a short time. ASML is the only company in the world capable of commercially producing EUV lithography machines, and each unit of equipment depends on a highly specialized supply chain that cannot be rapidly expanded.
Fouquet’s statement about supply shortages in 2027–2028 is essentially a warning for the entire AI chip production chain. If ASML’s capacity cannot keep up, the advanced process expansion plans of TSMC, Samsung, and Intel will be constrained, which will be transmitted to chip designers such as Nvidia and AMD, and ultimately affect the pace of construction of global AI infrastructure.
ASML is studying how to increase capacity growth in 2028 by more than 30%, but equipment production cycles and supply chain constraints mean that even if a decision is made now, the effect will only manifest in a few years.
