TSMC, Samsung, Intel Plan High-NA EUV Adoption

ASML’s current EUV systems can print chips up to roughly 800 square millimeters, a physical limit that companies such as Nvidia and Google routinely approach when designing data-center processors. The larger photomasks are intended to let High-NA tools manufacture chips at that scale without stitching separate sections together.
Intel is already using ASML’s High-NA EUV equipment to manufacture laptop processors, while TSMC and Samsung have not yet deployed the tools in high-volume fabrication. Their planned adoption would therefore broaden High-NA’s use beyond Intel’s current production application.
ASML has said it plans to increase its overall EUV production capacity by about 30% in 2027. Barclays analysts said the new commitments could give ASML greater visibility for capacity planning and force the company to consider expanding beyond its recently announced targets.
The new High-NA tools are described as capable of printing features almost twice as small as those of the previous generation and fitting up to three times as many transistors on a chip, potentially increasing processing capability for AI hardware.
ASML estimates that moving to 12-inch photomasks could raise productivity by about 40%, because larger masks can reduce the complexity and defect risk associated with stitching together separate mask sections for large processors.
TSMC, Samsung, and Intel are joining forces with Dutch chipmaker ASML to adopt High-NA EUV lithography, a next-generation technology that prints smaller features and fits more transistors on chips for artificial intelligence. PNN Digital reports the three companies plan to transition from today's 6-inch photomasks to 12-inch versions, potentially boosting factory productivity by 40% and eliminating the need to stitch together smaller masks when manufacturing massive data-center processors.
TSMC aims to begin high-volume production using High-NA EUV in 2030, with a pilot line for 12-inch masks by 2031 and broader manufacturing in 2033. Samsung could move faster—targeting high-volume DRAM production by 2028, which would make it the first memory-chip maker to deploy the technology at scale. KSL.com notes these commitments give greater visibility for demand of ASML's roughly $400 million machines.
Today's EUV systems can only print chips up to roughly 800 square millimeters—a limit that AI-focused companies like Nvidia and Google frequently hit when designing data-center processors. When a chip exceeds this size, manufacturers must stitch together separate mask sections, adding complexity and defect risk. Larger 12-inch masks solve this by allowing High-NA tools to manufacture full chips without stitching, EE News Europe reports.
Intel has moved ahead of its competitors, already using ASML's High-NA EUV equipment to manufacture laptop processors on its Intel 18A technology node. Evertiq reports that products made on Intel 18A using High-NA for select layers are meeting or exceeding performance of comparable layers patterned with earlier methods. TSMC and Samsung, however, have not yet deployed High-NA tools in high-volume production, making their planned adoption a major step toward broader industry adoption.
High-NA EUV tools can print features almost twice as small as previous-generation machines and pack up to three times as many transistors onto a single chip. This density boost translates directly into more processing power for AI hardware—crucial as companies race to build faster AI accelerators. The technology also increases productivity by roughly 40% when using 12-inch masks, according to ASML estimates, because fewer stitching operations mean less complexity.
ASML plans to boost overall EUV production capacity by about 30% in 2027. Barclays analysts say these new commitments from TSMC, Samsung, and Intel could give ASML much clearer demand visibility and may push the company to expand beyond its recently announced capacity targets. With each High-NA machine costing roughly $400 million, securing multi-year orders from the world's largest chipmakers provides ASML with significant revenue certainty as the semiconductor industry accelerates AI-chip development.
Publishers
56
Articles
183
Reach
239