Photo: Justin Sullivan/Getty Images
The advent of AI technology has unleashed a global race that is both commercial and geopolitical in nature. Private firms are staking out positions in the AI value chain and seeking to secure market share. Governments must capture the economic opportunity of AI and secure an emerging domain of national security. Both perceive potentially existential stakes in victory or failure.
For all parties, the physical infrastructure that enables AI computation is a central vector of competition. Behind every AI model and application lies a vast network of data centers, densely packed with computing hardware and requiring vast volumes of electricity to operate.
The AI race is without a doubt centered on the United States and China, who host 45 percent and 25 percent of world data center capacity, respectively. They also host the leading AI labs, including firms like OpenAI, Anthropic, and DeepSeek. Still, the policy challenge of the AI-driven data center and electricity demand boom is global. The International Energy Agency (IEA) puts global data center electricity consumption at 415 terawatt hours (TWh) of electricity—roughly 1.5 percent of global electricity demand—which is anticipated to nearly triple by 2035 to 1,200 TWh. This commentary examines the strategies of four countries—France, the United Arab Emirates (UAE), Japan, and Ireland—for attracting and managing electricity demand in the age of AI. Taken together, this brief survey highlights the ways economic, security, and technology strategies are increasingly centered on the electric power sector.
In February 2025, the French government unveiled a €109 billion infrastructure investment plan to support AI development. The Make France an AI Powerhouse strategy emphasizes stable electricity prices, grid reliability, and streamlined permitting processes, which indicates a clear understanding of the innate link between AI and electric power sector scaling. The plan includes a long-term commitment to grow electricity generation by 2 percent annually. This provides a clear signal to data center developers and technology firms that, in France, power will not be a bottleneck on AI scaling.
France comes into the AI race with a strong position. It is already Europe’s largest electricity exporter, reaching a net export balance of +89 TWh in 2024. For context—a 1 GW data center running at a 70 percent annual utilization rate would consume about 6 TWh annually. This significant electricity generation surplus could therefore support significant data center expansion, though this creates new questions about how nations that rely on these export volumes will adapt their electric power systems.
France is also Europe’s nuclear powerhouse and has announced plans to build six new reactors. In 2024, 68 percent of France’s electricity was generated from nuclear energy, which offers additional advantages to data centers. Besides its zero-carbon profile, this base of nuclear power anchors a reliable, resilient grid, which is crucial for data center operators. The stability of the French power grid amid the recent Iberian blackout, which might have cascaded further into Europe via France, is testament to this.
France’s strong inherited position and proactive growth strategy are already drawing global investment. UK-based Fluidstack plans to build a €10 billion, 1 GW AI data center powered by nuclear energy, while a €50 billion Franco-Emirati initiative will establish a second 1 GW AI campus. Ultimately, France has a very strong position on the power supply portion of the AI stack. But power supply is likely only a necessary, not sufficient, condition to be a true AI powerhouse on or near par with the United States and China. France must think about securing high-end graphics processing units and unlocking a more dynamic, innovative digital technology ecosystem.
Like France, the UAE has implemented an ambitious strategy to secure an advantageous position in the global AI ecosystem. The UAE has secured commitments from Microsoft for 160 MW of data center capacity, and Emirati AI technology firm G42 has plans to attract investment and scale total UAE data center capacity to 5 GW. Access to abundant energy is a key part of the pitch.
Domestically, the UAE is pursuing an ambitious scaling of electricity generation capacity across a wide range of technologies. Initiatives include a 5 GW gas power expansion project and a solar-plus-storage project, which at 1 GW in size will be the world’s largest. The UAE hosts the region’s first civil nuclear program, which has already brought 5.6 GW of capacity online in recent years. To manage the data center sector’s rising energy footprint, a national task force is developing regulations to balance digital growth with sustainability.
But there are challenges to this growth. The UAE’s extreme heat drives increased energy usage and operational costs. One survey found that cooling alone accounts for over 40 percent of UAE data center energy consumption. Another central issue is geopolitics. The AI Diffusion Framework, established by the Biden administration and kept in place by the Trump administration, creates challenges for Emirati access to leading-edge AI chips.
Part of the UAE AI strategy is clearly to hedge bets with both domestic and international investment. The UAE is pouring tens of billions into AI infrastructure and innovation investments abroad, including the 1 GW, €50 billion investment in a data center in France and a $25 billion partnership in U.S. data centers via ADQ and Energy Capital Partners.
The UAE is seeking to convert today’s energy wealth into an advantageous position in the AI economy of tomorrow. Energy is an Emirati advantage; a strong track record of solar, nuclear, and gas power indicates an ability to meet any conceivable data center demands. With a power supply advantage in place, it is clearly the intersection of geopolitics and commercial strategy that will most determine the ultimate shape of the UAE’s AI position.
Japan is undergoing a strategic shift in its energy planning as it anticipates a surge in electricity demand driven by AI and data center expansion. Today, data centers account for less than 20 TWh of electricity consumption in Japan, but the IEA estimates that this could increase by 80 percent (+16 TWh) by 2030. The country faces a unique set of challenges: limited land for infrastructure, constrained domestic energy resources, and strong public sensitivities around nuclear power following the 2011 Fukushima incident.
Still, the AI boom is prompting a policy pivot. A key asset for Japan is the large fleet of idled nuclear reactors, which represent spare capacity available for data centers, if politics allow for it. In a dramatic reversal, Japan’s latest energy strategy drops language about reducing nuclear reliance and instead calls for maximum utilization of both nuclear energy and renewables. The Ministry of Economy, Trade and Industry now supports restarting idle reactors and constructing next-generation nuclear plants at existing sites.
Companies like Ubitus K.K. are planning to construct new data centers in regions utilizing existing nuclear power infrastructure, such as Kyoto, Shimane, or Kyushu. Meanwhile, Japan continues to expand offshore wind and solar capacity, with plans to double renewable capacity by 2040. Tokyo-based telecommunication giants like Nippon Telegraph and Telephone Corporation and KDDI Corporation are building data centers with dedicated renewable power purchase agreements.
Japan’s Green Transformation initiative includes some innovative policies at the intersection of digital and energy policy. These include subsidies for energy-efficient computing infrastructure and incentives for companies locating data centers in regions with surplus renewable power. Japan is also one of the few countries that has a reporting requirement for data center electricity consumption.
Japan shows how the AI race can radically shake up the policy status quo. While perhaps less rhetorically ambitious than France or the UAE, the Japanese policy response is nonetheless significant. The successful reversal of deeply established policy on nuclear is no easy feat. If well executed, this policy will not only boost Japan’s position in the AI economy but also dramatically improve prospects for its long-term energy security and decarbonization goals.
Ireland offers a crucial reminder of the risk of rapid data center expansion outpacing energy infrastructure. Data centers consumed 5 percent of Ireland’s electricity in 2015; that figure reached 21 percent in 2023 and could grow to 30 percent by 2030, according to Ireland’s Commission for Regulation of Utilities. Ireland’s rise as a hub for data centers started with the “Celtic Tiger” boom at the turn of the century and has since been bolstered by its strategic position alongside undersea fiber-optic networks, tax incentives, and EU membership. Good weather helps, too; in contrast to the UAE, the Irish climate keeps data center cooling costs low.
Instead of seeking to attract data center investment, Ireland’s strategy reacts to the challenges of rapid growth. Concerns over blackouts prompted Ireland’s grid operator to pause approvals for new data centers in the Dublin area until 2028. In response to this escalating problem, the Irish government proposed a new electricity connection policy for data centers. Under the proposed rules, a “bring your own generation” standard would require new data centers seeking connection to the national grid to provide backup generation or storage equal to their demand and participate in the electricity market to support grid adequacy. The policy allows for both fossil (likely gas) and renewable sources, but emissions must be reported to regulators.
Wind power is well placed to help Ireland meet the surging data center demand. Already, some technology firms have contracted directly with onshore wind projects. For example, Microsoft has a 336 MW power purchase agreement (PPA) with onshore wind developer Statkraft, while Amazon Web Services signed a 105 MW PPA with Bord na Móna to offtake from their Derrinlough Wind Farm. The Irish government has ambitious plans to develop offshore wind resources backed by a state-administered contracting and subsidy scheme. This approach precludes direct contracting between offshore wind and data center projects, a result which could help deliver on climate and power system reliability goals at the same time and reduce subsidy costs. On the other hand, it is not clear if commercial terms could be struck between data centers and offshore wind projects, given the latter’s high cost, pointing to the reality that some type of policy mechanism is likely necessary to link the two together.
The Irish policy response to the surge in data center demand in the country remains a work in progress. In this respect, Ireland looks a lot like northern Virginia, where the power sector is acting as a brake on an active data center boom and attractive economic opportunities. Ireland is putting forward new policy models, and further iterations are likely. Regulatory models in the power sector need to evolve to allow new contracting pathways and market instruments that enable data center investment to coordinate and contract with energy infrastructure projects.
The electric power sector is defined by investments with long lead times and very long operational lifetimes. This reality clashes with a global AI technology race where performance breakthroughs are achieved month by month.
Still, policymakers must take a long view. The trajectory of AI technology means that computational capacity will become increasingly central to economic growth. Electricity-intensive economic growth means that proactive long-term planning in the power sector is a strategic necessity. Whether the task is attracting investment or reacting to booming data center growth, the core objective must be investments that unlock sustainable, affordable, and reliable long-term scaling in the power sector.
Cy McGeady is a fellow with the Energy Security and Climate Change Program at the Center for Strategic and International Studies (CSIS) in Washington, D.C. Rebecca Riess is an intern with the Energy Security and Climate Change Program at CSIS.