When people talk about the semiconductor supply chain, they fixate on fabs, foundries, and the geopolitics of Taiwan. What they rarely mention is glass. Specifically, the lenses and mirrors that make modern chip manufacturing possible in the first place.
The most advanced chips today are built using extreme ultraviolet lithography, a process that prints circuit patterns onto silicon wafers using light with a wavelength of just 13.5 nanometers. At that scale, conventional glass lenses are useless. EUV light gets absorbed by almost everything, so the entire optical system must be built from multilayer mirrors polished to atomic-level smoothness. A speck of dust, a surface irregularity measured in fractions of a nanometer, and the whole system fails.
Optics as the Rate-Limiting Step
This is why Carl Zeiss SMT, the semiconductor optics division of the German optics giant, is so central to the chip industry. Zeiss supplies the projection optics for ASML's EUV machines. Without Zeiss optics, there are no EUV machines. Without EUV machines, there are no leading-edge chips from TSMC, Samsung, or Intel.
The relationship between optics and chipmaking is not incidental. It is structural. The physics of lithography means that the quality of the lens system directly determines the smallest features you can print. Shrink the wavelength, tighten the tolerances, and you push the boundaries of what lenses can do. This is why the companies that dominate semiconductor equipment tend to have deep roots in precision optics.
Nikon and Canon, both originally camera and lens companies, built the lithography equipment that powered the semiconductor industry for decades. ASML, which now holds a near-monopoly on EUV, relies on Zeiss because building optics at this level requires institutional knowledge that takes generations to develop. You cannot simply spin up a new optics supplier.
The Convergence of Light and Silicon
There is a broader lesson here about how advanced manufacturing works. The frontier of chip technology is not just a matter of transistor counts or fab capacity. It is constrained by a web of dependencies, many of which trace back to seemingly unrelated industries.
Precision optics, metrology, chemical engineering, materials science. These fields converge in the lithography machine, and a bottleneck in any one of them ripples through the entire supply chain. The recent push to reshore semiconductor manufacturing in the United States and Europe has run headlong into this reality. Building a fab is hard. Building the machines that go inside it is harder. Building the optics that go inside those machines is a challenge that only a handful of organizations on Earth can meet.
This is also why export controls on semiconductor equipment have teeth. When the U.S. restricted sales of advanced lithography tools to China, it was not just blocking machines. It was blocking access to optical systems that cannot be replicated without decades of accumulated expertise.
The chip industry talks a lot about Moore's Law and the relentless march of miniaturization. What gets less attention is the parallel story of optical engineering, where the demands have scaled in lockstep. Every new process node requires optics that are cleaner, more precise, and more stable than the last generation.
In the end, the ability to manufacture advanced semiconductors rests on the ability to manufacture advanced lenses. The two disciplines are inseparable. If you want to understand where the chip industry is going, follow the light.
