At the Almaraz nuclear power station in western Spain, two reactors that have been generating electricity since the mid-1980s are scheduled to shut down by 2028. Across the country, five more reactors at sites including Ascó, Cofrentes, Trillo, and Vandellós face the same fate over the following years, with the last one going dark by 2035. Together, these seven reactors still produce roughly a fifth of Spain’s electricity and do so without burning fossil fuels. Closing them on schedule, according to peer-reviewed research published in the journal SERIEs earlier this year, would raise wholesale electricity prices and push carbon emissions higher, precisely the opposite of what Spanish climate policy is supposed to achieve.
That finding arrives at an awkward moment. Technology companies are scouting locations across Spain for AI-focused data centers that will need enormous amounts of steady, round-the-clock power. Grid planners are already grappling with how to keep the lights on as demand climbs. And a growing number of voices in Madrid are asking whether pulling the plug on working nuclear plants is a luxury the country can still afford.
The research behind the numbers
The SERIEs study, titled “The price and emissions effects of extending nuclear lifetimes: evidence from Spain,” models what happens to the Spanish electricity market when nuclear reactors retire on their current schedule versus when their operating lives are extended. The results are clear in direction, even if the precise figures depend on assumptions about future gas prices and carbon costs: removing nuclear capacity forces the grid to rely more heavily on gas-fired power plants during hours when wind and solar fall short. That reliance drives up wholesale electricity prices and increases CO₂ output.
Because the study was published in a peer-reviewed journal, its methodology and conclusions have been vetted by independent academics before reaching print. The research does not advocate for or against nuclear power; it quantifies the tradeoffs. Different assumptions about storage costs or the speed of renewable buildout could narrow or widen the gap, but across every scenario the authors tested, the phase-out led to both higher prices and higher emissions compared with keeping reactors running.
Spain’s official position, meanwhile, remains unchanged. The country’s final updated National Energy and Climate Plan (known domestically as the PNIEC), submitted to the European Commission in 2024, commits to a staggered nuclear shutdown. The plan assumes that renewables, battery storage, grid interconnections with France and Portugal, and demand management will fill the gap left by retiring reactors. It is a legally significant document: EU governance rules tie national energy plans to bloc-wide climate targets, meaning any major deviation would require a formal update and engagement with Brussels.
Why AI data centers change the equation
Spain’s sunny climate and competitive land costs have made it a target for technology firms looking to expand their European data center footprint. Facilities designed to train and run large AI models consume electricity at industrial scale, often drawing hundreds of megawatts around the clock. Unlike a factory that can shift production to off-peak hours, a data center training a language model needs uninterrupted power for weeks or months at a stretch.
Solar panels, Spain’s most abundant renewable resource, produce nothing after sunset. Wind generation fluctuates with weather patterns and can drop sharply for days during high-pressure systems. Nuclear plants, by contrast, run at near-full output regardless of the time of day or the weather. Removing that steady baseload while simultaneously plugging in large new loads from data centers creates a gap that must be filled by some combination of natural gas generation, massive battery banks, imported electricity, or demand curtailment. Each option carries costs, and several carry emissions.
European grid operators have already flagged data center growth as a source of strain. Ireland’s EirGrid, for example, has imposed connection moratoriums in parts of Dublin because data centers threatened to overwhelm local infrastructure. Spain has not reached that point, but the trajectory of AI-related demand suggests the pressure will intensify. The SERIEs research gives Spanish officials a framework for weighing whether existing nuclear capacity could serve as a low-carbon bridge during this transition, hedging against both volatile gas prices and tightening EU carbon permit costs under the Emissions Trading System.
Political signals, not policy reversals
It is important to distinguish between political debate and actual policy change. As of June 2026, no official Spanish government statement referenced in primary institutional documents confirms that policymakers are actively revising the nuclear phase-out timeline. Spanish media outlets including El País and Cinco Días have reported growing discussion within political circles, with opposition parties and some figures within the governing coalition questioning the wisdom of closing plants that still run safely and cheaply. But speculation in newspaper columns is not the same as a cabinet decision or a revised energy plan filed with the European Commission.
The PNIEC remains the current rulebook. Changing it would be a multi-step process: a political signal from the government, a public consultation, a revised draft submitted to Brussels, and a Commission review to ensure alignment with EU climate targets. Some member states, led by France, have pushed hard for nuclear energy to be recognized as part of the EU’s clean energy toolkit. Others, notably Germany and Austria, have resisted. Where Spain positions itself in that European debate could matter as much as domestic grid modeling.
What the numbers cannot tell us yet
Several pieces of the puzzle are still missing. No publicly available Spanish government projection isolates how much electricity AI data centers will demand over the next decade. European-level estimates point to significant growth, but country-specific figures for Spain have not appeared in the PNIEC or in published reports from Redeia, the operator of Spain’s transmission grid. Without those numbers, the scale of the demand challenge remains an informed estimate rather than a verified forecast.
The SERIEs study, for all its rigor, is a model, not a crystal ball. Its outputs depend on assumptions about future fuel prices, carbon permit costs, technology learning curves, and how quickly Spain can build out wind, solar, and storage capacity. Peer review validates the methodology and internal consistency of the work, not the certainty of any single number. If battery storage costs fall faster than expected or if Spain accelerates grid interconnections with France, the relative advantage of keeping nuclear plants online could shrink. If gas prices spike again, as they did after Russia’s invasion of Ukraine, or if renewable deployment hits permitting bottlenecks, that advantage could grow substantially.
There is also the question of reactor safety and economics. Extending the life of a nuclear plant is not simply a matter of political will. It requires investment in maintenance and upgrades, approval from Spain’s Nuclear Safety Council (CSN), and agreement from the plant operators, primarily Endesa, Iberdrola, and Naturgy, who must decide whether the economics justify continued operation. Some operators have publicly expressed reluctance to invest in aging facilities without long-term regulatory certainty, creating a chicken-and-egg problem: the government hesitates to change policy without operator commitment, and operators hesitate to commit without policy clarity.
A decision that cannot wait much longer
The clock is what makes this debate urgent. Almaraz’s two reactors are the first scheduled to close, and the preparatory steps for decommissioning take years. If Spain decides to extend nuclear lifetimes, the regulatory and engineering groundwork needs to begin well before the scheduled shutdown dates. Waiting until the last moment would leave the country with fewer options and higher costs, whether it ultimately keeps the plants running or replaces them with something else.
For now, the evidence points in a clear direction without dictating a single answer. The SERIEs research shows that closing nuclear plants under current conditions would make electricity more expensive and more carbon-intensive. Spain’s own energy plan assumes the gap can be filled, but that assumption has not been stress-tested against the kind of demand growth that AI data centers could bring. The political conversation is heating up, but it has not yet produced a policy shift. What happens next will depend on whether Spanish leaders treat the phase-out as a settled question or as a plan that deserves a second look in light of new evidence and new demand.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
