AI Data Centers Follow Energy: Power Decides Digital Growth — NRG-IA

AI data centers are starting to follow energy. Google's statement at the Eurelectric Power Summit 2026 highlights one of the most significant shifts for the European digital economy: new investments in data infrastructure increasingly depend on access to electricity, particularly renewable energy that is rapidly and predictably available. Diane Hodnett, Director of Data Center Investment and Development at Google, explained in Helsinki that energy has become a decisive criterion in selecting sites for new data centers. Economedia reports that the discussion took place against the backdrop of growing pressure exerted by artificial intelligence on electricity demand. This shift moves the center of gravity for investments. Under the old digital logic, proximity to users, fiber optics, land, the tax climate, and permits were what mattered. In the new logic of AI, energy becomes the primary filter: how many megawatts can be delivered, how quickly the connection can be made, how stable the grid is, how much clean energy can be contracted, and how flexible consumption can be. AI transforms electricity into a location criterion Data centers have long been perceived as invisible infrastructure: technical buildings, servers, cables, cooling, and connectivity. The rise of AI changes the scale of the issue. Large artificial intelligence models require permanent computing power, and computing power demands continuous, secure, and ever-increasing electricity. Eurelectric estimates that energy consumed by data centers in the EU and the UK could reach 149–287 TWh by 2030, up from approximately 100 TWh in 2022 . In the high-growth scenario, data center consumption approaches the annual electricity consumption of a large European economy. The scale of this growth changes the conversation. A large data center is not added to a grid like an ordinary consumer. It requires firm power, long-term contracts, connection capacity, redundancy, cooling, operational security, and, increasingly, clean energy certificates or contracts. In this context, energy becomes digital infrastructure. Those who cannot deliver electricity on time lose the investment, even if they have land, good internet, and technological ambitions. The grid becomes the new bottleneck for the European cloud Eurelectric data highlights the gap between digital speed and grid speed. A data center can typically be built in 18–24 months , but in European data center hubs, grid connection delays can reach 7–10 years . This is the imbalance that could redraw the investment map. Capital moves quickly. Servers can be ordered. Buildings can be erected. Commercial contracts can be signed. The power grid operates on a different timeline: studies, permits, substations, transformers, lines, reinforcements, balancing, and coordination with system operators. For Europe, the issue becomes one of competitiveness. The European Union wants its own AI infrastructure, digital sovereignty, and local data processing. These goals require an electrical system capable of rapidly connecting large consumers without increasing risks for other grid users. If connection takes too long, investments flow to faster markets. If clean energy is difficult to contract, companies look elsewhere. If the grid cannot support new consumption peaks, data centers remain mere projects on paper. Clean energy becomes a competitive advantage Google emphasizes renewable energy because major tech companies have climate targets, reputational pressure, and global contracts requiring low-emission electricity. For a data center investor, clean energy no longer serves merely as a sustainability message. It becomes a condition for financing, permitting, public acceptance, and commercial positioning. Europe has a potential advantage: growing renewable capacities, decarbonization policies, integrated electricity markets, and an industrial base that can support grid infrastructure. However, Eurelectric warns that nearly 28% of the growth in European electricity demand up to 2030 could come from data centers. This means that renewables alone do not solve the equation. Solar and wind energy bring clean volumes, but data centers require continuous availability. The system needs reinforced grids, storage, long-term contracts, flexible capacities, demand-side management, and clear rules for connecting large consumers. An AI data center does not just look for green energy on paper. It looks for deliverable energy, during the hours it needs it, within a grid that can sustain the consumption. Data centers can become flexible consumers The growth in consumption should not be treated solely as pressure on the system. Google is trying to present data centers as flexibility resources as well. In March 2026, the company announced demand response agreements totaling 1 GW , allowing the consumption of its data centers to be reduced during critical periods for the grid. The mechanism is significant. Instead of a data center operating…