ASML Lines Up TSMC and Samsung for High-NA EUV Tools

The Two Announcements ASML Made on September 8

ASML landed two related commitments from the industry's leading-edge chipmakers, and they operate on very different timescales. The near-term one is High-NA EUV adoption. The long-term one is a shared move from the semiconductor industry's decades-old 6-inch photomask to a 12-inch format.

• Samsung plans High-NA EUV in high-volume manufacturing by 2028; TSMC from 2030, joining Intel, which is already running the tools
• Intel has produced more than one million wafers on High-NA EUV for its 18A line
• 12-inch photomask pilot line targeted for 2031, with advanced-node production from 2033
• ASML's CTO says system productivity rises about 40% with the larger masks

Intel Foundry, TSMC and Samsung issued joint statements alongside ASML. SK Hynix is evaluating participation.

Why a Bigger Stencil Is a Real Engineering Fix

A photomask is the stencil that prints circuit patterns onto a wafer. High-NA EUV uses an anamorphic optical design, which let chipmakers keep their existing 6-inch masks — but at a cost. The exposure field is halved, so large dies have to be split and stitched together across separate exposures.

Stitching works. It is also slow, it constrains design, and it adds a failure mode. Moving to a 12-inch mask restores the full exposure field and removes the constraint outright. That is where the 40% productivity claim comes from, and it is why Intel signalled it wants near-term enablement on 6-inch masks before transitioning — the fix is real, but 2031 is a long way off when you have tools on the floor now.

What Does This Mean for Semiconductor Investors?

Two things, and they point in the same direction on a long horizon.

First, this is demand visibility for ASML stretching to 2033. High-NA systems reportedly run around $400 million each. Locking three leading-edge foundries onto a shared mask roadmap is about as durable a commitment as this industry produces, and it reduces the risk that the format fragments — which was the genuine danger for a transition this expensive.

Second, it is a capital intensity story for the foundries. TSMC's 2030 date is notably later than Samsung's 2028, which reads as a deliberate choice to extract more from existing tooling rather than a stumble. That fits a company posting the kind of numbers we covered in TSMC's record July revenue.

Standard caveat: none of this is investment advice, and roadmaps at seven-year range slip. But industry-wide format agreements are rare, and they tend to be the moments where long-term supply chains actually get decided — the same dynamic behind deals like Nvidia's $3.5 billion MediaTek investment.

The Bottom Line

The headline is AI chip demand pulling the whole lithography supply chain forward, and the substance holds up better than most headlines of that shape. A pilot line in 2031 and production in 2033 is not a trade — it is a planning assumption. What is tradeable is the confirmation that the three companies capable of leading-edge manufacturing have all now told the market they are buying, and roughly when.

For more on the chip and market side of the AI build-out, see our stock trading coverage.

Sources: The Register — September 8, 2026; Engadget — September 8, 2026.