Europe's energy and digital infrastructure are converging at speed, and the investment decisions made today will directly shape the continent's competitive position for decades to come. Across four keynote sessions, one conclusion was consistent: the binding constraint on AI-driven digital growth is not power generation capacity, it is grid connection access. For those who recognise this early and position accordingly, the opportunity ahead is substantial.

​

• Setting the EU Policy & Climate Baseline: Paolo (European Commission) opened with the trajectory that frames everything else: global data centre electricity consumption is on course to nearly double by 2030, driven by AI workloads that have pushed GPU rack power density up 200 times over the past 15 years. EU regulatory action expected in Q2 2026 will mandate full reporting of data centre energy, water, and carbon metrics, and the difference between a regulated and unconstrained growth scenario amounts to 100TWh of annual demand. Paolo also stressed AI's dual value: well-targeted AI applications in grid management and building efficiency could reduce global carbon emissions by approximately 5% over the long term.

• Reframing Grid Strategy Around Flexibility: Marc (EDF) shifted the industry conversation from securing stable power supply to actively participating in grid balancing. With 76% of European data centre operators citing grid connection delays as their top operational challenge, Marc made the case for data centres to take on a proactive balancing role, through workload shifting, cooling modulation, and battery storage. Drawing on France's low-carbon, nuclear-anchored grid, EDF is positioned to offer long-term fixed-price power contracts that deliver the supply certainty operators need while contributing to greater system-wide flexibility.

• Framing Data Centre Demand as an Energy Upgrade Catalyst: Neal (NTT Global Data Center) reframed the central debate: large-scale data centre power demand is not a burden on European energy networks, it is a catalyst for accelerating investment in grid upgrades, renewables, and storage. He outlined three persistent industry pain points: rigid ramp-up limits and "use it or lose it" power clauses that strand contracted capacity; restrictive phased grid agreements that extend timelines and inflate backup generation costs; and the constant tension between energy cost control and decarbonisation targets in a competitive green PPA market. NTT's response is a hybrid integrated energy model, combining grid supply, transitional gas, and renewables, paired with flexible grid access, demand-side response, and waste heat recovery, as a scalable blueprint for next-generation AI infrastructure.

• Grounding Planning in Real-World Grid Data: Christina (Aurora Energy Research) closed with the market constraints that underpin all infrastructure decisions. Wide divergences in demand forecasting across research institutions are not a minor inconsistency, they directly distort long-term investment planning. Aurora's analysis is unambiguous: the scarce resource is nodal grid connection capacity, not power generation. Established digital hubs face connection queues of 3 to 10 years, with Dublin's moratorium running until 2028. Emerging locations such as Zaragoza are gaining traction not because of proximity to end users, but because of available nodal grid headroom, making granular grid modelling a non-negotiable tool for site selection, asset financing, and network investment planning.

Europe has the industrial ambition, policy momentum, and private investment appetite to lead in the global AI infrastructure race. The single most important priority now is deeper coordination, between grid operators, data centre developers, investors, and policymakers, to translate shared diagnosis into durable, well-planned infrastructure that can support the next decade of digital growth.