,

Granada lands an AI datacentre campus

Alto’s Granada project pushes Spain deeper into Europe’s AI infrastructure race.

Granada lands an AI datacentre campus
Summary
  • Alto Infrastructure has begun work on a 70MW IT capacity AI and HPC campus in Granada.
  • The project links regional development, renewable energy claims, liquid cooling, and Spain’s wider AI compute ambitions.
  • Spain’s datacentre market is expanding beyond Madrid, but grid access and customer commitments remain decisive constraints.

Alto Infrastructure has started construction on a 70MW AI and high performance computing campus in Granada, adding another major project to Spain’s expanding digital infrastructure pipeline.

The SP01 campus, located in Escúzar, is planned to reach 70MW of IT capacity by late 2029, supported by 100MW of gross electrical power. The first phase is expected to go live in summer 2027 with 10MW of IT capacity, rising to 25MW before the end of that year.

The technical design is aimed at AI, cloud, and high performance computing workloads. Alto plans direct to chip liquid cooling and rack densities of up to 250kW, a level associated with intensive AI training and inference environments rather than conventional enterprise hosting. The company also says the site will run on renewable energy and use low levels of water, which will be central to local acceptance as datacentre development faces closer scrutiny over power and environmental impact.

The project strengthens Spain’s position in Europe’s AI infrastructure map. Madrid has attracted much of the country’s datacentre attention, but the Granada campus suggests capacity growth is spreading into regions where land, power, and industrial policy may align more favourably. Spain has also approved major AI infrastructure spending as it seeks to reduce dependence on foreign technology and build capacity compatible with European regulation.

Demand alone no longer determines where datacentres are built. Grid connections, power pricing, cooling design, planning consent, and anchor customer commitments can decide whether a campus moves from announcement to operation. Secured power is becoming a commercial advantage and a planning flashpoint, particularly as AI workloads change the density and energy profile of new facilities.

Regional economic claims need careful treatment. Large AI campuses can create construction activity, high skilled operational roles, tax income, and demand for energy infrastructure, although they may produce fewer long-term jobs than political rhetoric suggests. The more credible projects are those that can demonstrate power availability, heat and water management, customer demand, and integration with local industry or research ecosystems.

Granada gives Alto a chance to present AI compute as regional industrial infrastructure rather than a property play. If the campus is to support European AI sovereignty, it will need to host workloads that create value beyond raw capacity: scientific computing, enterprise AI services, industrial modelling, public sector research, or cloud platforms serving European customers under European rules.

AI is also changing datacentre design. Higher rack densities and liquid cooling move new facilities away from generic server halls and toward specialised compute environments with different supply chains, maintenance skills, and energy profiles. Those requirements are now shaping where projects can be built and who can finance them.

Spain has no shortage of ambition in AI infrastructure. Granada’s test will be whether secured power, customer commitments, and regional execution can keep pace with the scale of the plan.