
Insight
The next semiconductor race will be won on the factory floor.
By Peter Jenkins, President and CEO, Infinitesima.
Angstrom process control for the AI era.
When I started my career in the semiconductor industry in 1984, we were working on what was then an advanced one-micron process.
Today, leading-edge chips are being manufactured at the two-nanometre node, with tens of layers of metal interconnect and increasingly complex three-dimensional structures. We are no longer simply making transistors smaller. We are building chips by combining elements fabricated on different wafers, connecting chips together using Advanced Packaging techniques to deliver ever more powerful compute devices.
That matters because the semiconductor industry is entering one of the biggest manufacturing transitions in its history.
For decades, progress was driven primarily by shrinking features. That is becoming harder as we reach dimensions measured in atoms. The industry is responding through advanced packaging, chiplets, high-bandwidth memory and increasingly complex 3D device architectures.
The challenge is that manufacturing these structures is one thing. Measuring them accurately and quickly enough to manufacture millions of them economically is another.
A human hair is around 100 microns across. Divide that into 100 and you are roughly at a micron. Divide one of those pieces another thousand times and you are at a nanometre. A Silicon atom itself is around 0.3 nanometres in diameter.
Existing measurement technologies all have strengths, but also limitations. Optical techniques become constrained as structures get smaller. Electron-beam technologies can measure small features but have limitations in capturing the three-dimensional information manufacturers increasingly need. Atomic force microscopy can provide extremely precise 3D measurements, but conventional approaches have historically been too slow for widespread use in high-volume manufacturing.
This is what we describe as the Measurement Gap.
It is an economic problem as much as a technical one. Semiconductor manufacturing is ultimately about yield. If manufacturers cannot identify and control tiny variations quickly enough, yields fall, costs rise and new technologies take longer to reach volume production.
AI has brought semiconductors to public attention in a way I have never seen during my career. But AI does not simply require more chips. It requires increasingly sophisticated chips.
Advanced processors, high-bandwidth memory and new approaches to packaging and integration are increasing performance, while the huge energy requirements of AI mean efficiency is becoming equally important.
The result is greater complexity, and greater complexity makes process control more important.
You cannot deliver the economics of AI simply by designing a better chip. You have to be able to manufacture it repeatedly, at high yield and at enormous scale.
That is the challenge Infinitesima has spent the past 25 years working to solve. Our RPM™ (Rapid Probe Microscopy) technology is designed to bring the precision and 3-dimensional measurement capabilities of atomic force microscopy into high-volume semiconductor manufacturing.
Metron3D™ was released by SK hynix for volume DRAM manufacturing in 2025, and we have a three year development programme with imec, working with ASML, across areas including hybrid bonding, High-NA EUV lithography and future 3D transistor structures.
I spent 28 years at ASML and watched it grow from a scale up into a company occupying a critical position in the global semiconductor supply chain. Its story demonstrates that countries do not need to own every part of semiconductor manufacturing to possess strategically important capability.
The UK has a strong base to build from in this area. We have specialist semiconductor equipment companies, world-class engineering, materials expertise and intellectual property. These companies may be less familiar than the major chipmakers, but their technologies can become just as important to the manufacturing process.
That is why I believe semiconductor equipment is a relatively safe bet for the UK. Building leading-edge fabs means competing directly with countries that have spent decades and enormous sums establishing their manufacturing position. Equipment is different. It plays into a global market: develop a technology that the world's leading manufacturers need, and it can become part of their production lines wherever those fabs are located.
And that opportunity is growing. The next phase of semiconductor innovation will depend on advanced packaging, high-bandwidth memory, hybrid bonding and increasingly sophisticated 3D architectures. Each creates new manufacturing and measurement challenges, and therefore new demand for the equipment needed to solve them.
I saw at ASML how a specialist equipment company can become critical to the manufacture of the world’s most advanced chips. As the industry goes through its next manufacturing transition, new technologies will become just as critical. The UK already has real strength in semiconductor equipment, and that puts us in a good position to play an important part in what comes next.
The next semiconductor race will not just be about who can design the most advanced chips, but who can provide the technologies needed to manufacture them at scale.
Continue reading





