Can we power AI-driven data centres without derailing Net Zero progress?

Lately, one of the major questions that is being asked is - can we accurately forecast tomorrow’s electricity demand? Especially in light of the increases driven by AI-powered data centres. But the more pressing question is - how do we meet that soaring demand without sacrificing the hard-won progress we've made on the road to Net Zero?

Data centres, particularly hyperscale and AI-led facilities, are fast becoming some of the most power-hungry components of the digital economy. According to the International Energy Agency (IEA), global data-centre electricity use is set to more than double by 2030, reaching around 945 TWh, roughly equivalent to Japan’s current total consumption.

In Europe, data-centre capacity is expected to grow from around 10 GW today to 35 GW by the end of the decade, according to McKinsey. In the UK alone, data centres currently account for 2.5% of national electricity use. That figure is expected to more than double within the same timeframe.

This rapid growth is already straining infrastructure. Roughly half of Europe’s transmission lines are more than 40 years old. Many parts of the grid are in urgent need of reinforcement to handle the rising peaks, not only from data centres but also from electric vehicles and other electrified sectors. 

Predicting future electricity needs has always been part science, part art, traditionally based on stable patterns like growing factory output, seasonal heating and cooling demands, and gradual population increases. But AI-powered data centres don’t follow those patterns.

Support for the digital economy

AI workloads can be unpredictable because one moment power use is steady, and the next an AI training run sends demand soaring by tens of megawatts. These sudden spikes—called “burst” loads—are hard to predict and put big pressure on the grid. If the grid can’t respond instantly, the risk of a blackout grows.

Adding to this challenge, data centres often cluster in the same places such as Amsterdam, Frankfurt, London or Slough, which puts extra strain on local networks. In many areas, grid operators have paused new connections until upgrades are made. That has forced developers to find quick fixes like on-site gas turbines, batteries, or diesel, just to keep running.

These stop-gap solutions may keep the lights on, but they are insufficient long term. To support the growing digital economy without compromising reliability or climate goals, Europe’s energy mix must evolve. Wind and solar will remain the backbone of the energy transition, but their inherent variability makes them unsuitable as the sole power source for critical infrastructure, especially for facilities that cannot tolerate even a millisecond of downtime.

Powerful, green solutions

This is where gas turbines, particularly those capable of running on gas blended with hydrogen, can play a key bridging role. They ramp up quickly, provide essential grid stability services, and can significantly reduce emissions. Blending hydrogen into existing turbines offers a glimpse of what’s possible while volumes of hydrogen increase to enable the full transition to 100% hydrogen combustion and a ramping up of renewables in the power mix.

The latest turbines, such as the ones we have at Mitsubishi Power, can already operate on 30% hydrogen and 70% gas, and are progressing towards 100% hydrogen firing capability. In the UK, for example, the Saltend H2H project on the Humber is getting ready to blend hydrogen with natural gas to produce low-carbon power for both industrial and digital users. In Scotland, Peterhead Power Station is preparing a combination of hydrogen-ready turbines with carbon capture to remove 1.5 million tonnes of CO₂ annually.

Of course, powering the future of AI and cloud computing isn’t just about generation. Cooling systems, backup power, and energy management all add complexity and further pressure on energy systems. That is why we are seeing developments in modular, end-to-end solutions tailored specifically for hyperscale and colocation data centres. These include integrated systems for power generation, liquid cooling, backup solutions, and digital energy management, designed to support both rapid deployment and long-term sustainability.

This comprehensive approach helps data-centre operators move faster, cut emissions, and prepare for a hydrogen-powered future, offering a robust pathway to meet both uptime and Net Zero goals.

Changes needed at policy level

More than just a trend, the rise of AI and digital infrastructure marks a structural shift in how we use electricity. That means energy policy must shift too. We need clear, long-term signals in capacity markets that reward flexibility, as well as incentives for plants that can ramp quickly and run on low-carbon or hydrogen-ready fuels.

Regulators should also prioritise grid access for projects that combine low-carbon generation, energy storage, and smart controls. This will ensure data centres can continue to scale without compromising the broader reliability of the grid.

Europe’s Green Industrial Plan has laid important groundwork, but more needs to be done. Streamlining interconnection procedures and prioritising critical digital infrastructure will be key to meeting both economic and environmental objectives.

If data centres are to meet the demands of AI and digitalisation, they must look beyond traditional power purchase agreements. By integrating on-site generation, battery storage, and intelligent energy management systems, they can hedge against grid delays, enhance resilience, and deliver the uptime their customers demand in our electrified future.