The Eindhoven Connection
How Dutch Photonics Could Shape the Future of AI Infrastructure
The global race to build artificial intelligence is often framed as a contest between American technology giants and Asian chip manufacturers. Silicon Valley dominates the headlines with massive investments in AI chips and data centers, while Taiwan and South Korea supply the manufacturing muscle that turns designs into silicon.
Yet behind this visible competition lies a quieter infrastructure race—one that determines how quickly data can move between the processors that power modern AI systems.
As AI models grow larger and more complex, the real bottleneck is shifting. The challenge is no longer simply building faster processors. It is ensuring that thousands of processors inside hyperscale data centers can exchange data quickly enough to function as a single machine.
Some of the key technologies addressing that challenge—especially those involving photonics and optical communication—have roots in a place that rarely appears in discussions about the AI race: the Dutch city of Eindhoven.
“The real breakthrough is not the laser alone, but the integration. We are moving from a world where we glue components together to a world where complete optical systems can be printed on a millimeter scale. The Netherlands holds the blueprint for this transition.”
Prof. Dr. Martijn Heck
Professor of Photonic Integration, Eindhoven University of Technology
For decades, researchers at the Eindhoven University of Technology—often referred to as TU/e—have been working on technologies that guide light through microscopic circuits on a chip. Known as integrated photonics, these systems promise to transmit data using photons instead of electrons, dramatically increasing bandwidth while reducing energy consumption.
That work is now becoming central to the infrastructure behind artificial intelligence.
From Compute Power to Data Movement
The first phase of the AI boom focused on raw computational power. Companies such as NVIDIA and AMD built specialized processors capable of training massive machine learning models. Hyperscale cloud providers—including Microsoft, Meta and Amazon—then assembled thousands of those chips into enormous AI clusters.
But scaling these systems revealed a new constraint. The processors must constantly exchange model parameters, training data and intermediate results. As clusters grow to tens of thousands of chips, the amount of data flowing between them becomes staggering.
Electrical interconnects based on copper cables struggle at these speeds. They consume increasing amounts of power and generate heat, making them inefficient for the massive scale of modern AI infrastructure.
This is why engineers are increasingly turning to photonics.
Instead of transmitting electrical signals, photonic systems move data using light traveling through optical waveguides. The approach offers higher bandwidth, lower energy consumption and significantly less heat generation—key advantages in massive data centers where power efficiency is critical.
Companies such as OpenLight in the United States are now pushing these technologies toward large-scale deployment, integrating lasers directly onto silicon photonics chips designed for high-speed data center networking.
Yet the intellectual foundations for many of these developments trace back to research ecosystems such as the one in Eindhoven.
Silicon and Indium Phosphide: Two Worlds Converging
One of the central technical challenges in photonics is the laser.
Silicon—the material used to manufacture most modern chips—is excellent for electronics but poor at generating light. Optical systems therefore require another material capable of efficiently producing laser signals.
That material is often Indium Phosphide, a compound semiconductor widely used in high-performance photonic devices.
Eindhoven’s photonics ecosystem has become one of the world’s leading centers of expertise in indium phosphide technology. Researchers and startups in the region have spent decades refining methods to integrate photonic components into compact, scalable systems.
The emerging solution for AI infrastructure may be a hybrid architecture that combines both materials: silicon for electronic circuitry and indium phosphide for laser generation.
“Silicon is an excellent material for electronics, but it does not naturally emit light. For that you need indium phosphide. The future of AI infrastructure will likely depend on hybrid platforms that combine the strengths of both.”
Prof. Dr. Martijn Heck
Photonic Integration Researcher, TU Eindhoven (interpretation of his published research perspectives)
This hybrid approach is precisely what many companies developing next-generation photonics platforms are pursuing today.
Brainport: Europe’s Photonics Ecosystem
The Eindhoven region—often referred to as Brainport—has evolved into one of Europe’s most concentrated clusters for high-tech innovation.
The ecosystem includes not only the university but also research institutes, startups and specialized manufacturing companies focused on photonic technologies.
One of the most notable is SMART Photonics, a semiconductor foundry dedicated to producing photonic integrated circuits. Unlike traditional semiconductor fabs that manufacture digital processors, SMART Photonics focuses on optical chips capable of guiding and manipulating light.
“We are on the verge of a massive rollout. Demand from the AI market for faster and more energy-efficient connections is the catalyst that is pushing photonic chips out of the lab and directly into the backbone of the internet.”
Luc Augustin
Chief Technology Officer, SMART Photonics
This shift reflects the growing realization that optical technologies are not just useful for telecommunications networks spanning continents. They are becoming essential inside the data centers that power cloud computing and artificial intelligence.
Supporting this ecosystem is PhotonDelta, a national initiative designed to strengthen the Netherlands’ position in integrated photonics. The program aims to build a complete value chain—from academic research to industrial manufacturing—capable of competing globally.
The Role of ASML
Any discussion of the Dutch semiconductor ecosystem inevitably leads to ASML, the world’s only manufacturer of extreme ultraviolet (EUV) lithography machines.
These machines are essential for producing the most advanced semiconductor chips and are used by manufacturers such as TSMC, Intel and Samsung Electronics.
While ASML is primarily associated with digital semiconductor manufacturing, the broader Brainport ecosystem it anchors is increasingly relevant to photonics as well. The same precision engineering, advanced materials science and specialized supply chains that enabled EUV lithography are helping drive innovation in photonic technologies.
In that sense, the Dutch semiconductor ecosystem is expanding from enabling the world’s fastest processors to influencing the technologies that allow those processors to communicate.
Photonic Sovereignty
For Europe, photonics represents more than just a technological opportunity. It may also offer a strategic position in the global technology landscape.
While the United States dominates AI software and chip design and Asia leads in semiconductor manufacturing, Europe has developed significant expertise in photonic hardware and optical systems.
“Photonics is an opportunity for Europe not just to consume technology, but to become an architect of the infrastructure on which the global AI ecosystem will run.”
Eelko Brinkhoff
CEO, PhotonDelta
If optical interconnects become as critical to AI systems as GPUs themselves, regions with strong photonics ecosystems could play a disproportionately important role in shaping the next generation of computing infrastructure.
The Quiet Backbone of the AI Era
The rapid rise of artificial intelligence has captured global attention. Governments are investing billions in semiconductor manufacturing, cloud infrastructure and AI research.
Yet the future of AI may depend on technologies that receive far less attention.
The processors inside AI systems may be the engines of the digital age, but engines alone do not build highways. Without the infrastructure that allows those processors to exchange data at extraordinary speeds, even the most powerful chips would be constrained by network bottlenecks.
In that sense, the AI revolution is not only about faster chips.
It is also about the optical pathways that connect them—and those pathways may lead, unexpectedly, through Eindhoven.
Photo credit
AI-generated illustration / OpenAI
Caption
Conceptual illustration of integrated photonics chips transmitting data with light. Optical interconnects are emerging as a critical technology for connecting AI processors in hyperscale data centers.
This shift—from electrons to photons—is part of a deeper transformation in how we understand intelligence: not just as software, but as something rooted in physics, energy and infrastructure.
I explore this idea further in my ebook The Age of Light — Meaning, Machines and the Physics of Intelligence, about how photonics and physical computing architectures are reshaping AI and global power.
Available worldwide on Amazon (Kindle):
https://www.amazon.com/dp/B0GMXLX56T
