Intel Ships the First High-Volume Chips Made on High-NA EUV
Key takeaways
- Intel is the first to run High-NA EUV, the roughly 400 million dollar next-generation lithography machine, in high-volume production
- It beat TSMC and Samsung to using the tool at scale on its new 18A node, with yields reportedly climbing
- High-NA prints finer chip features in a single exposure, which can simplify manufacturing and cut cost per wafer
- After years of falling behind, this is Intel's clearest sign yet that it is back in the leading-edge race
Intel has become the first chipmaker to ship high-volume chips made on a High-NA EUV machine, the roughly 400 million dollar scanner that represents the next step in how the most advanced processors get built. It beat both TSMC and Samsung to using the tool at scale, running it on its new 18A node with yields that Intel says are climbing. After years of watching from behind, that is a real marker in the chip race.
High-NA EUV is a mouthful, so here is the plain version. Chipmakers print circuit patterns onto silicon using extreme ultraviolet light. The "NA" is numerical aperture, a measure of how sharply that light can be focused. A higher numerical aperture means finer detail, the same way a better lens resolves smaller print. High-NA machines, built by the Dutch firm ASML, can carve features finer than anything in current mass production.
The catch has always been cost and complexity. Each scanner costs around 400 million dollars, weighs as much as a couple of buses, and ships in dozens of freight crates. Getting one to produce usable chips reliably, at volume, is a different problem from getting it to work at all. That is the bar Intel says it has now cleared.
Why getting there first matters
Printing a fine feature in one High-NA exposure can replace two or more passes on older machines. Fewer passes means fewer steps, fewer chances for error, and lower cost per wafer once yields are healthy. Whoever masters that first tends to set the pace for the next few process nodes, because the learning compounds. Every wafer teaches the engineers something the competition has not learned yet.
For Intel, the timing is loaded. The company has spent years being the cautionary tale of the industry, the former leader that missed a beat and let TSMC pull ahead. Landing High-NA first on 18A is the strongest counter-argument it has offered in a while. It also fits a wider pattern of enormous bets on chip capacity, from Samsung's spending spree to the scramble for new fabs on home soil.
The bigger picture
None of this means the race is won. TSMC and Samsung are not standing still, and yields at the leading edge can swing quickly. High-NA also does not fix everything; packaging, memory bandwidth, and power are still hard problems that no single machine solves.
But for the first time in a long while, Intel is the one with the newest tool running hottest. In an industry where being first to a process advantage can define a decade, that is worth paying attention to.
For more on the money flooding into chip capacity, see our piece on Samsung's 648 billion dollar chip bet, the fight to build fabs closer to home in the Texas mega-fab story, and how memory makers are cashing in through SK Hynix's blockbuster listing.