Latest News

Europe AI data centres are looking for cheaper and faster energy and land

Data analysis by shows that Europe's data centre developers for artificial intelligence seek out locations outside of major cities where they can take advantage of cheaper energy,?land, and faster connection times.

This trend could mean investing in "underprivileged" areas where governments try to stimulate growth and jobs, but developers risk facing opposition from residents who are concerned about the dwindling of natural habitats as well as competition for water and power.

JLL, a world-leading commercial real estate company, has shared information on the plans for hyperscale data centers that will be coming online in the next two to three years.

Centres that do not have to be so close to the end users will be three times farther away from major cities compared to those built in the previous three years.

JLL has not revealed the names of companies that are developing the centres. However, the largest companies in the world, including Meta, Google, and?Microsoft?, are investing worldwide.

Sites in Cities Become Rare

The average distance between 2026 and 2020 will be 175 km (109 miles), compared to 46 km for projects delivered in 2022-2025. This is because power and land are becoming scarcer in major cities like London and Frankfurt.

Greenfield projects make up 39% of Europe’s future pipeline compared to only 8% delivered?projects. The data showed that the pipeline projects located in inner-city areas will fall from 13% to 5%, with the remainder in industrial and edge-of-city sites.

This?shift' is primarily driven by AI training campus, which need a lot of electricity and water to cool.

"The determining factors are increasingly where enough power can be obtained, and not simply where the demand is," said Assad Nori, JLL’s head of data centers in Europe, Middle East, and Africa. "Data centres will be brought to the places where power is available, and not the opposite."

JLL estimates that the four world's largest hyperscale cloud service providers will spend $725 Billion in 2026. This is up 77% over $410 Billion in 2025. The majority of this money will be spent on AI computing, data centres, and infrastructure.

The report added that by 2030, AI-related workloads could make up around half of the global data centre capacity.

The trend JLL has shown is supported by other sources of data centre analysis.

According to DC Byte, which tracks early stage projects, only one of nine proposed data centres with a gigawatt or more capacity is located near a major European city. The rest are spread across rural Spain and northern Sweden.

The?core' markets of Frankfurt (London),?Amsterdam (Paris), and Dublin continue to be the most popular and in high demand. However, these markets are increasingly facing land shortages, restrictions on planning, and long waits to connect to the grid.

Martin Jensen is the president of JLL’s EMEA Data Centres division. He said that Europe's core markets would remain crucial because enterprise demand "isn't moving anywhere".

He added, "A hyperscale AI infrastructure needs a different scale of power and land."

According to JLL powered land costs on average EUR2,36 million per megawatt for IT loads in core markets. This cost is EUR978,000 for cities considered secondary such as Copenhagen, Warsaw, and Milan. The cost of EUR512,000 in tertiary locations like Bordeaux can be as low as EUR200,000.

London is the second most expensive market, at EUR2.6 millions per megawatt. Frankfurt comes in third at EUR2.5million.

Land prices are high in London because other asset classes are competing for space. Rupert Duckworth is the associate director of EMEA data centre advisory at Savills. (Reporting and editing by Barbara Lewis; Simon Jessop)

(source: Reuters)