NEW YORK — What happened
Around late 2026, ASML, the Dutch chip equipment giant, has successfully begun the commercial deployment of its first-generation High-NA EUV (Extreme Ultraviolet) lithography systems. These massive, multi-billion-dollar machines are crucial for manufacturing the most advanced semiconductor nodes, pushing transistor densities to unprecedented levels. Early systems have been installed in a handful of top-tier foundries, with the first production wafers expected to roll out in the coming year.
Why it matters
This marks a significant milestone in the quest for ever more powerful and efficient chips. High-NA EUV enables chipmakers to etch finer features onto silicon, leading to smaller, faster, and more power-efficient processors. This will directly impact everything from next-generation AI accelerators and data center CPUs to high-performance computing. However, the immense cost and complexity of these machines also mean that advanced chip manufacturing will remain concentrated among a very select few companies, further solidifying the oligopoly in leading-edge silicon production.
Deep dive
Traditional EUV uses a numerical aperture (NA) of 0.33. High-NA EUV systems boast an NA of 0.55, allowing for a resolution approximately 1.7 times finer. This leap is critical for fabricating features below 2nm, such as those anticipated for 1.8nm or A14 nodes. The technology involves incredibly precise optics, vacuum environments, and powerful laser-produced plasma light sources. Each machine requires hundreds of thousands of components and specialized infrastructure, making their deployment a multi-year, multi-billion-dollar undertaking for any chip manufacturer. The initial learning curve for these systems is steep, as is the cost of integrating them into existing fab lines.
Report check
Claims that High-NA EUV will immediately double transistor density or halve chip costs are overstated. While it enables higher density, the cost per transistor decrease is a long-term goal, not an immediate outcome. ASML has shipped initial systems for evaluation and process development, and some sources indicate early production runs are indeed underway for validation. However, widespread commercial volume production and significant yield improvements are still projected for late 2027 or 2028. Claims of exclusive access for certain foundries are largely verified due to the limited number of machines produced and their high cost.
Open questions
How quickly can chipmakers ramp up production and improve yields with High-NA EUV? What will be the ultimate cost impact on end-user devices? Will this technology further widen the gap between leading-edge and trailing-edge chip manufacturers, and what are the geopolitical implications of such concentrated expertise?