For decades, digital power appeared to reside in software. Platforms, data and algorithms dominated economic narratives, while the physical foundations of computing faded into the background. Silicon Valley’s rise reinforced the idea that control over code — not factories — would determine the future.
Yet the AI era is quietly overturning that assumption. Training frontier models requires enormous amounts of hardware, energy and specialized chips. Cloud infrastructure has become less a virtual abstraction than a network of industrial facilities consuming electricity on the scale of small nations. The supposedly weightless digital economy turns out to be built on steel, silicon and rare materials.
This shift exposes a deeper vulnerability. Europe excels at research and regulation but depends heavily on external suppliers for advanced semiconductors and compute capacity. Without domestic infrastructure, even the most sophisticated European AI systems ultimately run on hardware designed, manufactured or operated elsewhere.
“The global race for leadership in chips is a reality and Europe must claim an active role. We have the talent and the research, but we lack the link to production and scaling.” Thierry Breton — Former European Commissioner for the Internal Market, European Commission
Breton’s warning reflects a growing recognition in Brussels that technological sovereignty cannot be achieved through software policy alone. It requires rebuilding the industrial base that underpins digital capabilities — a project now framed as essential to economic security, defense readiness and political autonomy.
Read More
For much of the past two decades, European telecom operators have occupied an awkward position in the digital economy. They build and maintain the networks, absorb the capital intensity, navigate regulation and public scrutiny — yet capture relatively little of the perceived value. In public discourse they are often reduced to “dumb pipes”, utilities competing on price in saturated markets while hyperscalers dominate software, platforms and data.
That framing made sense in a world where digital value was assumed to scale independently of physical constraints. If computation could be centralised in ever-larger data centers and delivered globally over high-speed networks, then proximity, geography and infrastructure ownership mattered less than software ecosystems.
That world is changing. As energy constraints tighten, latency becomes economically meaningful and architectural limits surface, the physical layer is reasserting itself. The supposedly “dumb” pipe turns out to control something Silicon Valley increasingly lacks: geographically distributed sites, grid connections, real estate and trusted operational presence close to end users.
“Whether it be digitalization, decarbonization, deglobalization and defense […] our infrastructure is the foundation for all of that to happen. […] If infrastructure is used, it can fuel real growth.”
Alison Kirkby
CEO, BT Group
Kirkby’s remark is notable not for its rhetoric but for its reframing. Infrastructure is no longer presented as a commodity input but as a precondition for economic transformation. In this view, telecom operators are not peripheral to the digital economy — they are the physical substrate upon which it increasingly depends.
Read More
Personnel announcements rarely signal structural change. Scientists retire, executives move, institutes reshuffle leadership — the quiet churn of a global research ecosystem. Yet occasionally, a transition reveals far more than a personal decision. It exposes the invisible currents of capital, power and technological competition shaping the future of infrastructure itself.
The decision by Martin Schell — long-time Executive Director at Germany’s Fraunhofer Heinrich Hertz Institute (HHI) — to leave after two decades and join Huawei’s Bragg Research Center in London appears, at first glance, to fit this familiar pattern. A senior researcher returns to industry. A prestigious institute loses a leader. Another multinational strengthens its R&D bench.
But this move lands at a moment when the physical foundations of the digital world are being rebuilt. Artificial intelligence, cloud computing, quantum research and next-generation networks are colliding into a single infrastructure challenge: moving unprecedented volumes of data at tolerable energy cost.
“Photonics is currently the core technology of telecom infrastructure. And new investments will be needed to cope with AI-generated traffic demand. American and Asian players dominate the data centre equipment market.”
— Photonics21, AI Desperately Needs Photonics Report (November 2025)
If AI is the visible revolution, photonics is its invisible engine. And that is precisely where this story begins.
Read More