EUV Lithography

The EUV lithography supply chain bottleneck revolves around ASML's monopoly as the sole global supplier of extreme ultraviolet (EUV) lithography systems. Standard EUV with 0.33 numerical aperture (NA) is essential for manufacturing nodes from 7nm down to 3nm, while the transition to High-NA EUV (0.55 NA) has become the critical path for sub-2nm production. The extreme complexity of these systems—particularly their optical and light-source components—limits production scaling and leads to high lead times, constraining the advanced manufacturing roadmaps of leading-edge foundries and integrated device manufacturers.

Recent developments highlight persistent challenges. ASML's High-NA EUV systems are not yet ready for high-volume manufacturing (HVM), as confirmed by industry reports in May 2026. This delays advanced node roadmaps for TSMC, Intel, and Samsung Foundry. Additionally, curvilinear masks required for High-NA EUV push the limits of inspection and metrology, while mask technology faces converging challenges from tighter specifications and new materials. A research milestone—Imec's fabrication of a quantum dot qubit using High-NA EUV—demonstrates the technology's potential but remains far from commercial production.

Key players include ASML as the source, with Intel, Samsung Foundry, and TSMC as affected parties. These foundries are heavily reliant on ASML's capacity expansion plans, which target approximately 90 standard EUV units and 20 High-NA units by 2025-2026. However, the supply chain's intricate optics, light sources, and metrology components continue to constrain rapid scaling, forcing chipmakers to compete for limited tool availability and adjust their node timelines accordingly.

The outlook remains cautious. While ASML's capacity targets suggest gradual improvement, the mask ecosystem and High-NA readiness are notable bottlenecks. Standard EUV supply is expected to increase modestly, but the transition to High-NA for sub-2nm nodes faces multiple technical hurdles—including inspection and metrology gaps—that may push volume production into 2027 or later. The bottleneck is likely to persist as a structural constraint on advanced node scaling, with incremental improvements rather than dramatic breakthroughs in the near term.