The Geopolitics of Light: How Photonics is Rewiring the Global AI Economy

AMSTERDAM – October 13, 2026 – The artificial intelligence revolution is colliding with a formidable adversary: the physical limits of the global power grid. As hyperscalers and nation-states race to build unprecedented computing clusters, the electrical infrastructure required to sustain them is buckling. In regions like Western Europe, where energy grids are already strained, the question is no longer just how to build smarter algorithms, but how to power the underlying hardware without plunging municipalities into the dark.

This month, the mechanics of this looming crisis—and its potential solutions—were dissected in the Netherlands, where 226 delegates from 22 countries convened for the Innovative Optical and Wireless Network (IOWN) Global Forum. The gathering of representatives from the consortium's 180-plus member organizations at the FUTURES Amsterdam event was not merely a technical symposium. It was a strategic summit aimed at dismantling the architectural bottlenecks of the modern internet and redefining who controls the physical layers of global data.

At the core of the discussions was a stark reality: traditional electronic switching architectures are hitting a wall. The energy required to constantly convert data from optical signals (for long-haul transport) to electrical signals (for processing and routing) and back again is unsustainable in the era of generative AI. The proposed solution is a radical shift toward All-Photonics Networks (APNs), a framework designed to keep data in the optical domain from end to end, drastically slashing power consumption and latency.

The Physical Limits of the AI Boom

The urgency of this transition is palpable in Amsterdam, a major European data center hub that has previously faced moratoriums on new large power connections due to severe grid constraints. The International Energy Agency has repeatedly warned of surging electricity consumption driven by data centers, emphasizing that incremental efficiency gains are no longer sufficient.

During the forum's public session, Ian Redpath, Research Director at Omdia, outlined the global view of the AI infrastructure market, detailing the rapidly changing requirements for capacity, power, latency, and sustainability. As AI workloads scale exponentially, the sheer volume of data moving between compute nodes creates a massive energy drain.

Dr. Gonzalo Camarillo, Marketing Steering Committee Chair of the IOWN Global Forum and Head of Implementation Components at Ericsson, outlined how the consortium is developing architectures and deployment approaches for next-generation networks. By leveraging Open APNs, the industry aims to achieve power consumption levels that are a fraction of current standards, while simultaneously delivering massive bandwidth capabilities. This is not just an upgrade; it is a fundamental rewiring of how digital infrastructure operates, replacing heat-generating electrical routers with frictionless optical pathways.

Sovereign AI and the European Imperative

Beyond the physics of power and light, the Amsterdam summit highlighted a critical geopolitical shift: the rise of "Sovereign AI." For European nations, the push to develop indigenous AI capabilities is inextricably linked to the desire to break free from reliance on foreign-owned cloud infrastructure.

The European Union's Digital Decade policy and the landmark AI Act have set a regulatory backdrop that heavily favors data sovereignty. But as policymakers are quickly realizing, software regulations are toothless without control over the physical hardware and network layers.

Stijn Grove, Managing Director of the Dutch Data Center Association, explored this changing landscape, emphasizing the critical availability of power and network capacity needed to support AI growth on the continent. A subsequent panel, moderated by Ronald van Loon, Founder and CEO of Intelligent World, brought the theoretical into the practical. Grove, alongside Accenture’s Jeff Wang and Ciena’s Dejan Bilbija, debated the complex relationship between data centers, networks, and cloud services, parsing the high-stakes policy and investment decisions that will dictate deployment.

Dr. Katsuhiko Kawazoe, President and Chairperson of the IOWN Global Forum, crystallized the stakes. “Sovereign AI depends on having greater control over the infrastructure that supports it,” he noted. “That means giving organizations more choice over how their data, computing, and networks are connected while reducing dependencies that can limit their ability to adapt as their needs change. By working across industry boundaries, our members can help shape a more resilient and flexible foundation for sovereign AI and the next generation of digital services.”

Beyond Telecoms: The Industrial Migration to Optical Rails

Perhaps the most telling indicator of this infrastructure shift is the rapidly diversifying membership of the IOWN Global Forum. What began in 2020 as an initiative heavily driven by telecommunications giants has expanded into a cross-industrial coalition.

At the Amsterdam event, the forum welcomed several new organizations, including sponsor member FET, academic institutions like the Singapore University of Technology and Design, and general members such as DataDirect Networks (DDN). More notably, the inclusion of Airbus SAS and the Narita International Airport Corporation signals that ultra-reliable optical backbones are no longer just a telecom concern—they are an industrial necessity.

For aerospace and aviation, the ability to process massive amounts of telemetry data at the edge, or to facilitate real-time, low-latency communication for air traffic and ground operations, requires networks that traditional electronics cannot support. Similarly, the forum announced expanded cooperation with the 5G Automotive Association (5GAA) and the Automotive Edge Computing Consortium (AECC). As the automotive sector moves toward autonomous, connected mobility, vehicles are effectively becoming rolling data centers. The vehicle-to-everything (V2X) communication required to prevent collisions and manage traffic flow demands the ultra-low latency that only end-to-end photonics can guarantee.

Proving the Concept in the Field

While the geopolitical and industrial theories are sound, commercial adoption relies on proven, field-tested economics. To that end, the forum recognized the "Remote Media Production in Broadcast Industry" project as its Implementation of the Year.

Broadcast media has traditionally relied on deploying fleets of heavy production trucks and dozens of personnel to live event sites—a logistical nightmare with an immense carbon footprint. The award-winning project demonstrated how dynamic optical path provisioning through the Open APN can provide high-capacity, low-latency connections between remote production sites on demand.

By transmitting multiple uncompressed video streams and enabling the remote control of production equipment from centralized hubs hundreds of miles away, the implementation proved the economic feasibility and reliability of the architecture. It is a microcosm of the broader economic promise: replacing heavy, localized, power-hungry operations with centralized expertise connected by frictionless light.

As the global economy becomes increasingly tethered to AI, the entities that control the speed, efficiency, and sovereignty of the underlying networks will dictate the terms of the next industrial era. The gathering in Amsterdam makes one thing clear: the future of computing will not be measured merely in the sophistication of its code, but in the mastery of its physical infrastructure.