Spain’s $3.4B AI Campus Bets on On-Site Power

A proposed €3 billion ($3.4 billion) artificial intelligence campus in central Spain is advancing through regional permitting with an on-site power strategy, developers EdgeMode, BlackBerry AIF and the town of Mora said Monday. The 300 MW DC MALPICA AI project, comprising two 150,000 square meter AI data centers, is designed to generate its own electricity to reduce dependence on the conventional grid, reflecting a growing trend among AI infrastructure developers to build around power constraints. The project ranks among Europe's largest AI campuses outside France, according to Dell'Oro Group research director Alex Cordovil, and is expected to create up to 5,000 construction jobs and 1,000 permanent positions.

Spain’s $3.4B AI Campus Bets on On-Site Power The planned 300 MW AI campus in Spain would rank among Europe’s largest as developers pursue fast-track approval and an on-site power strategy. A proposed €3 billion $3.4 billion artificial intelligence campus in central Spain is advancing through regional permitting with an energy strategy that reflects a growing reality for AI infrastructure: developers are building around power constraints rather than waiting for the grid. EdgeMode, BlackBerry AIF and the town of Mora on Monday presented the DC MALPICA AI project to officials from the Castilla-La Mancha regional government, outlining plans for a 300 MW campus comprising two 150,000 square meter AI data centers. The developers said they are preparing an application for the project’s designation as a regional Priority Project, a status intended to accelerate permitting for investments considered strategically important. For data center operators, the project’s most notable feature is not its size but its power model. Developers said the campus is designed to generate its own electricity, reducing dependence on the conventional grid while using technologies intended to minimize water consumption. The announcement did not identify the generation technology, fuel source or expected mix of on-site and utility power. Bloom Energy’s Europe director, Reto Zueger, participated in the meeting alongside executives from EdgeMode, BlackBerry AIF and other project partners. The companies have not disclosed Bloom’s role in the project, and a Bloom Energy spokesperson declined to comment. “Grid interconnection has become the single biggest determinant of when, and whether, an AI campus comes online, so on-site generation is no longer a backup plan; it is a strategic asset,” Alex Cordovil, research director at Dell’Oro Group, told Data Center Knowledge. Cordovil said developers are designing AI campuses with dedicated generation even when utility service is planned, allowing projects to move forward if grid connections are delayed while preserving capacity for future expansion. As one example, he pointed to Nscale’s AI campus in Norway, where delays to a planned grid connection forced the developer to pursue an on-site fuel cell solution to keep the project on schedule. Europe’s AI Buildout The proposed campus would provide 300 MW of capacity across two facilities designed for high-performance AI computing. Developers estimate the investment at €3 billion $3.4 billion and forecast as many as 5,000 construction jobs, followed by roughly 1,000 permanent direct and indirect positions after the campus enters service. Cordovil said the project ranks among the largest AI campuses currently planned outside France. “A 300 MW Spanish campus is genuinely at the top end of what Europe is building today,” he said, noting that most announced European AI campuses fall between roughly 50 MW and 300 MW. France has emerged as the exception, with several announced gigawatt-scale AI projects. Officials from Castilla-La Mancha’s ministries overseeing economic development, energy transition and telecommunications attended Monday’s presentation as the project moves through the regional approval process. “Artificial intelligence needs a new generation of infrastructures, and Spain has the opportunity to become one of the great technological hubs of Europe,” BlackBerry AIF CEO José Mora said. He said the project is designed to combine AI computing capacity with energy independence and sustainability. A New Class of AI Data Centers Cordovil said the industry’s move toward large, power-intensive AI campuses is unlikely to reverse, even as AI models become more efficient. “What we are really watching is a bifurcation – purpose-built AI factories emerging as a distinct class of facility alongside general-purpose data centers,” he said. Whether DC MALPICA ultimately becomes one of those facilities will depend on details the developers have yet to disclose, including how much of the site’s planned 300 MW will come from on-site generation, what fuel will be used and how the campus will interconnect with Spain’s electric grid. If those plans come together, the project would show how access to power has become as critical to AI infrastructure as access to GPUs.