My core view on
$XFAB:
X-FAB should not be understood as a company trying to compete head-on with
$TSM or Samsung on leading-edge digital nodes.
That is not the point.
TSMC and Samsung dominate the world of advanced digital computing: GPUs, CPUs, AI accelerators, smartphone SoCs, etc.
X-FAB operates in a very different battlefield: the real-world interface layer.
Analog. Mixed-signal. MEMS. Sensor interfaces. High-voltage. Power management. SiC. Automotive-grade specialty processes.
These are the chips and technologies that allow machines to sense, measure, control and interact with the physical world.
Pressure. Temperature. Motion. Sound. Light. Force. Power. Safety. Reliability.
This is not “low-end” semiconductor manufacturing.
In many cases, these are qualification-heavy, sticky, long-lifecycle technologies where customers care less about the smallest transistor and more about reliability, process stability, regulatory certification, long-term supply and the cost of switching suppliers.
That is why X-FAB’s exposure to automotive, industrial, medical and energy markets matters.
These customers do not change suppliers casually. Once a chip is qualified into a braking system, medical device, factory automation platform, energy system or sensor module, the product lifecycle can last many years, sometimes decades.
This gives X-FAB a very different type of moat from leading-edge digital foundries.
Another important angle is supply-chain sovereignty.
As EU-China trade tensions rise and the West continues to push friend-shoring and local semiconductor capacity, European and US-based specialty foundries become strategically more important.
Automotive, industrial, medical and energy customers cannot afford fragile supply chains.
Governments also understand this.
That means companies like X-FAB may benefit not only from customer demand, but also from policy support, subsidies and long-term strategic contracts.
Then comes the AI angle.
Most people still think of AI as something inside data centers: GPUs, networking, memory and servers.
But the next stage of AI is physical.
Embodied AI. Industrial automation. Autonomous systems. Medical devices. Smart factories. Humanoid robots.
When AI moves from the cloud into the real world, it needs much more than compute.
It needs sensing, actuation, power control and physical-world interface chips.
Humanoid robots, for example, will not just need GPUs, motors and batteries. They will need large numbers of pressure, motion, inertial, tactile, force, temperature, sound and optical sensors to interact safely and intelligently with the environment.
That is where X-FAB’s MEMS, sensor-interface, analog and mixed-signal platforms could become increasingly relevant.
So to me, X-FAB is not just a “European chip sovereignty” story.
It is also a picks-and-shovels play on AI moving from the digital world into the physical world.
The market may still be focused on leading-edge compute.
But the physical layer of AI — sensing, power, safety, reliability and long-lifecycle specialty semiconductors — may become increasingly valuable.
That is where
$XFAB sits.