Thoughts from SEMICON Europa 2025
Treating the Fab Like a Wafer – Thoughts from SEMICON Europa 2025
I’m in Munich this week at SEMICON Europa 2025. Walking the halls and talking to engineers, project managers, fab owners and suppliers, one idea kept coming back in different conversations. It came up with Irish partners like Arcadis, with German contacts such as Drees & Sommer, Exyte, and other members of Silicon Saxony, with fab owners from different companies, and with suppliers like Daifuku who think a lot about how fabs actually run day to day. We kept circling around the same thought: fabs are no longer “just buildings”. They are so dense, so complex and so critical that we should start thinking about building and controlling them more like a wafer. The comparison seemed to click with people, so I wanted to write it down while I’m still here at the fair.
From hand-built circuits to patterned wafers
In the early days of microelectronics, circuits were built almost like small crafts: components and transistors were placed and wired more or less one by one. That worked up to a point, but it was clear it wouldn’t scale. The real shift came with the planar process and silicon wafers. Instead of assembling everything individually, engineers defined patterns and then used lithography to transfer whole layers onto a wafer, with metrology and inspection to check what actually happened. Once that loop was in place – pattern, lithography, inspection – densities went up, yields improved and investments in new nodes and fabs became less of a gamble because the process was under control.
How fabs are still often built today
If you now look at how many fabs are physically built, you still see a very “project-based” approach. Models and drawings don’t always match what is happening on site, different trades sometimes work with slightly different information, surveying appears at certain milestones but not as a continuous control mechanism, and as-built information can be incomplete or scattered. That might be acceptable on a warehouse, but it becomes dangerous when you think about what a modern fab really looks like: very high tool density, multiple layers of pipes, ducts, cable trays and steel in the same space, more modular and offsite elements like pipe racks and skids that need to fit first time, and tolerances where a few millimetres can really matter. In other words, the way we construct the facility is lagging behind the precision we expect from the processes inside it.
Wafer-grade fab construction
This is where the wafer comparison helps. On the wafer you have a clear coordinate system, a defined pattern, a lithography step that transfers it and metrology that keeps everything under tight control. For fabs, I think we are now seeing the beginnings of the same mindset. I call it wafer-grade fab construction. In simple terms that means: a robust geospatial control network across the site, so everything ties back to a stable reference, a coordinated BIM or 3D model that defines where things actually belong, digital setting-out that transfers positions from the model onto concrete and steel, and regular laser scanning to compare what has been installed with what was intended. It’s the same basic loop – pattern, “lithography”, “metrology” – just applied to the building instead of the wafer.
Where Murphy Geospatial fits
This is exactly where I see the role of Murphy Geospatial on fab and other high-tech projects. We are not the ones designing the process tools or defining the fab layout in detail. Our job is to provide the geospatial backbone that gives everyone else confidence that their design can really be built and that what is built is where it is supposed to be.
In practice, that means we establish and maintain the survey control on site, we set out the positions for tools, plinths, embeds, anchors, openings and penetrations, we carry out laser scanning of structures, modules, pipework and services, and we compare the point clouds against the model so that we can flag problems early rather than when it is too late. We also look after the geospatial data so that design, construction and operations share the same picture of reality. If you stay with the wafer analogy, we try to be the alignment and inspection part of the process for the fab.
Modular and offsite make this even more critical
A lot of discussions here in Munich have been about modular and offsite construction. Prefabricated pipe racks, process skids, ceiling modules and similar elements are becoming normal. They are assembled and tested elsewhere and then brought to the fab as large 3D components. The advantages are clear: more work done in controlled factory environments, faster installation on site, less disruption and fewer people squeezed into tight spaces. But all of that depends on geometry being right. The design of the module, the conditions in the fabrication workshop and the conditions on site in the fab need to match within tight tolerances. If they don’t, modular turns from a solution into a source of delay and rework.
Wafer-grade fab construction – with strong control, accurate setting-out and systematic scanning – is what makes modular work at the scale we are now talking about.
Construction yield
In semiconductor manufacturing, yield is a central concept. Everyone knows how many good die per wafer they are getting and how process variation eats into that. Construction has a kind of yield as well, but we usually don’t call it that. We talk about rework, clashes that were found too late, lost options for future tools, or slow ramp-up because the as-built reality is different from what was assumed. It is, however, the same idea: how many “good outcomes” you get from the effort and money you put into a project. If we apply a wafer-style mindset to fab construction, we can treat these issues more as process defects that can be reduced, rather than just “bad luck on this job”. Wafer-grade fab construction is about improving that construction yield: fewer surprises, smoother tool installation, faster ramp-up and better preparation for future densification in the same footprint because we really know what has been built and where it is.
A simple question from Munich – and beyond fabs
So, after a couple of days of conversations at SEMICON Europa, I keep coming back to one simple question: Do we still see a fab as just a very complex building – or do we start to see it more like a wafer? If we choose the second view, then geospatial data, survey control, digital setting-out and laser scanning are no longer nice extras. They become part of the core process of delivering the fab.
What I find interesting – and this is on my mind as I head to a tunnelling conference next week – is that the same pattern appears in other worlds too. In tunnels and underground works you also have restricted access, high risk, tight geometry, modular and offsite elements, and a strong need for monitoring and performance-based decisions. The language changes, but the underlying idea is similar: better digital control of what we build, and better information about how it performs. At Murphy Geospatial, this is the role we want to play across these sectors: helping project teams bring this kind of thinking into both cleanrooms and tunnels, and into the other complex assets that tie them together.
If you are involved in a new fab, an expansion or a major tool install – or in similarly demanding infrastructure – and want to discuss this further, I am happy to connect.
Matthias Gropp, Murphy Geospatial a Woolpert Company
Project News
