🧪 Atom Beam Mask

ASML’s 🇳🇱 lithography machine etches chips with light at 13.5 nanometres. Lace Lithography 🇳🇴 wants to replace it with a helium atom beam 0.1 nanometres wide. That number reads like a spec detail. It isn’t.

Every advanced chip made anywhere on earth passes through an ASML 🇳🇱 machine. One Dutch company controls the chokepoint, at the precise moment AI is pushing compute demand well past what the industry planned for. EUV scanners now sell for €380m each, the next generation is expected to clear €700m, and the supply chain has no redundancy. The industry has known for years this arrangement wouldn’t hold. It had no credible answer, and the list of failed alternatives is long.

From a nanophysics lab at the University of Bergen, Bodil Holst co-founded Lace in 2023. Helium atoms have no diffraction limit: they reach resolutions that light, by its own physics, cannot. The last remaining lock was mask design, widely assumed combinatorially intractable. Lace broke it with an AI algorithm that accelerates the calculation by fifteen orders of magnitude. Fifteen. The company is validating the approach with Imec 🇧🇪, operating from four sites spread between Bergen, Barcelona, the UK and the Netherlands.

On paper, the €34.5m Series A pulls together five investors with no obvious common ground: Atomico 🇬🇧, M12 (Microsoft’s venture arm), Linse Capital 🇳🇴, Nysnø 🇳🇴 and Spain’s SETT. A Norwegian climate sovereign does not decide on the same clock as an American corporate arm. Staggered closes in European deeptech read as governance weakness to North American LPs. They are the only coherent architecture when investors operate in different time registers, and pretending otherwise has cost this continent a decade of missed rounds.

Thirty years and the combined capex of TSMC 🇹🇼, Intel 🇺🇸 and Samsung 🇰🇷: that is what ASML 🇳🇱 spent building EUV. For two decades the programme looked physically dubious, economically preposterous, and impossible on the timelines announced. Roughly what people say about Lace now. Major industrial breakthroughs almost always look unrealistic for a very long time. That is not a reason to believe. It is a reason to look more closely.

A pilot fab tool in 2029, on €51.7m of total funding. Believing the physics and the algorithm is one claim. Believing the calendar on that budget is a separate one. Honestly ambitious. European deeptech raised €18.5bn in 2025, an all-time record. The capital is finally there. What is still missing is patience for bets nobody will know how to score for a very long time.