High-NA EUV: Inside ASML's $350M Machines Saving Moore's Law

Transistors are now so small that we are literally fighting the wavelength of light to draw them. When you want to pack 100 billion transistors onto a piece of silicon the size of a postage stamp, standard lasers just look like giant, blunt crayons.

Before you read: We recently covered how the industry is redesigning the physical foundation of chips in The Glass Substrate Revolution. But before you can mount a chip to glass, you have to actually print the sub-2nm transistors on the silicon, which requires an entirely different kind of physics-breaking machinery.

Table of contents

• The Blunt Crayon Problem
• The $350 Million Physics Experiment
• The Economic Breaking Point
• Fun Facts
• Sources

The Blunt Crayon Problem

Standard Extreme Ultraviolet (EUV) lithography machines have a numerical aperture (NA) of 0.33. That was perfectly fine for the 5nm and 3nm nodes. But as we aggressively push toward 2nm and beyond, that 0.33 NA lens simply cannot focus the light tightly enough. The lines blur. The circuits short out.

Think of it like trying to draw a highly detailed architectural blueprint using a thick Sharpie. You need a finer pen. ASML’s answer is High-NA EUV, which increases the numerical aperture to 0.55. This allows fabs to print features 1.7 times smaller and achieve transistor densities nearly three times higher than standard EUV.

The $350 Million Physics Experiment

Building a finer pen at the atomic scale requires engineering that borders on magic. A single High-NA EUV machine costs roughly $350 million, weighs 150,000 kilograms, and requires three Boeing 747s just to ship it to a fab.

To generate the necessary light, the machine fires a high-powered laser at microscopic droplets of molten tin falling through a vacuum chamber, hitting them 50,000 times a second. This creates a plasma that emits EUV light. That light is then bounced off a series of the flattest mirrors ever created by Zeiss. The mirrors inside are so perfectly smooth that if you scaled them up to the size of the Earth, the tallest mountain would be less than a millimeter high. Meanwhile, I can’t even get my 3D printer bed level enough to print a plastic tugboat.

The Economic Breaking Point

I honestly think the industry is ignoring the massive financial crater High-NA is creating. At $350 million a pop, the financial physics are breaking faster than the optical ones.

TSMC famously dragged its feet on ordering these machines because the cost per wafer is astronomical. Intel, desperate to regain its manufacturing crown, rushed to buy the first available units. But the technical trade-offs are brutal. High-NA cuts the depth of focus in half, forcing fabs to use incredibly thin chemical resists. If the resist is too thin, the etched lines fail. It is an engineering nightmare that proves Moore’s Law isn’t dead—it just got obscenely expensive.

Fun Facts

• Did you know that EUV light is so easily absorbed by matter that even ordinary air blocks it entirely? The entire optical path inside the ASML machine has to be maintained in a near-perfect vacuum.
• The carbon dioxide lasers used to blast the tin droplets consume so much power that a single High-NA machine requires about 1.5 megawatts to operate—enough to power a small neighborhood.

Sources

• ASML Newsroom — High-NA EUV Lithography
• SemiAnalysis — ASML High-NA EUV Economics
• IEEE Spectrum — Inside the High-NA EUV Machine