Photonics emerges as a crucial solution to Europe’s growing AI energy challenge

A new report underscores photonics as the key to reducing AI-induced electricity demand and carbon emissions in Europe, amid soaring data centre growth and infrastructure pressures.

A new report by Photonics21, authored by TEMATYS, underscores photonics, the use of light to transmit and process information, as the only feasible solution to address the escalating energy demands of artificial intelligence (AI) without exacerbating carbon emissions. As AI models and data centres rapidly expand across Europe, the electricity consumption driven by these technologies is surging at a pace that current power grids cannot sustainably support.

Global data centre electricity usage already stands at 415 terawatt hours (TWh) in 2024, according to the International Energy Agency (IEA), representing about 1.5% of global electricity consumption. This figure is projected to more than double by 2030 if present growth trends continue, largely fueled by increased AI workloads. Within Europe, data centres currently consume roughly 2–3% of electricity, depending on the region, but this share is set to rise sharply. Industry analyses suggest that data centre power demand in the EU, UK, Norway, and Switzerland alone could almost triple by 2030, escalating from an existing load of 10 gigawatts to approximately 35 gigawatts, which would push overall electricity consumption by data centres beyond 150 TWh, about 5% of Europe’s total electricity use. This surge coincides with a stagnation of overall electricity demand growth across the continent since 2007, intensifying pressure on existing power infrastructure and climate targets.

The report highlights that current silicon-based semiconductor technologies cannot meet these rising computational needs without unsustainable energy costs and environmental repercussions. Photonics is thus presented as an essential complement to electronic chips, not a replacement for CPUs or GPUs, but as a means to alleviate their workload through innovations like co-packaged optics. This approach can substantially reduce energy consumption and carbon emissions associated with AI computing, supporting Europe’s ambitions for cleaner and more competitive AI infrastructure.

Already integral to digital networks through fibre optic technology, photonic components are now progressing towards integration within chip packages, a critical developmental phase for translating laboratory advances into scalable industrial solutions. Notably, the Massachusetts Institute of Technology (MIT) showcased fully photonic neural-network chips in 2024, signalling promising breakthroughs in optical computing that could further enhance energy efficiency.

Europe boasts world-class photonics research and a growing cluster of pioneering start-ups. However, the report warns that without coordinated investment, scaled manufacturing capabilities, and workforce development, Europe risks falling behind and becoming dependent on overseas suppliers for key AI infrastructure components. Investment calls emphasize the need for increased funding in pilot manufacturing, scaling photonics firms, prioritizing photonics within major technology funding schemes, and building suitable skills for production at scale.

This concern is echoed in recent EU initiatives, including a €133 million investment in pilot photonic semiconductor production facilities in the Netherlands, part of a broader €380 million effort under the Chips Joint Undertaking. Spearheaded by leading universities and research institutes, this initiative aims to enhance Europe’s semiconductor competitiveness, secure supply chains, and boost photonics’ strategic importance. Facilities are planned to commence development by 2025.

Parallel industry developments illustrate the rising momentum for photonic chips in AI data centres. For example, STMicroelectronics, in collaboration with Amazon Web Services (AWS), is launching a photonics chip designed to improve speed and power efficiency in AI infrastructure. This chip will enter production in 2025 at ST’s French facility and targets the expanding market for optical transceivers, forecasted to grow significantly by 2030. Such advancements underscore the critical role photonics is poised to play in the future AI ecosystem.

Meanwhile, the scale of the challenge is apparent at the policy level. Belgium is contemplating imposing energy allocation limits on data centres to manage soaring electricity demands largely attributed to AI. The national grid operator, Elia, has proposed reserving fixed grid capacity for data centres to prevent them from monopolising electricity resources and crowding out other sectors. This issue is being incorporated into Belgium’s upcoming 10-year grid development plan, reflecting broader concerns about infrastructure constraints amidst rapid AI-driven growth.

At the EU level, policymakers acknowledge the urgent need for enhanced energy efficiency in data centres. The European Commission is preparing new measures aimed at improving energy use across these facilities as part of its broader sustainability goals. Data centres currently account for approximately 3% of the EU’s electricity usage, a share expected to rise sharply with AI’s adoption.

The commercial and regulatory landscapes converge on the recognition that without significant innovation, particularly in photonics, and strategic infrastructure planning, the rapidly expanding footprint of AI in Europe threatens to become both costlier and environmentally more damaging. According to Sébastien Bigo, Nokia Bell Labs Fellow and Photonics21 work group leader, Europe holds the research foundation to lead in clean AI infrastructure, but coordinated investment and industrial-scale efforts will be decisive in determining if this potential is realised.

As Europe’s AI-driven digital economy evolves, the interplay of cutting-edge technologies, ambitious industrial policies, and energy sustainability will shape its competitive position globally and its ability to meet climate commitments.

📌 Reference Map:

• ^[1] IT Brief – Paragraphs 1, 2, 3, 4, 5, 7, 8, 9
• ^[2] Reuters (Belgium energy limits) – Paragraph 6
• ^[3] Reuters (McKinsey data centre power demand) – Paragraphs 2, 6
• ^[4] Reuters (EU investment in photonics) – Paragraph 7
• ^[5] Reuters (EU energy saving measures) – Paragraph 6
• ^[6] Reuters (STMicroelectronics photonics chip) – Paragraph 7
• ^[7] European Parliament study (IEA data centre energy use) – Paragraph 2

Source: Fuse Wire