Data center systems spending is expected to surge 42.4 percent this year, as organizations invest heavily in AI capabilities (Gartner). With data centers’ power demand expected to reach 4 percent of the global power demand by 2030 (McKinsey), software computation, or compute, is shifting to regions that offer cheap energy and permissive grid regulations.

NTT Global Data Centers is scaling nearly a gigawatt across Europe, “doubling down in Frankfurt, London, Amsterdam and Paris, and in tier two sites like Berlin,” says EMEA Managing Director Konstantin Hartmann. To address power, NTT is “exploring SMRs, fuel cells and green hydrogen.” Meanwhile, Oliver Schiebel, CEO of hscale, is targeting Milan, Madrid, and Oslo, and “actively investigating” opportunities in the Middle East. “We expect substantial growth in Norway,” he adds. The Nordic countries, after first serving the crypto mining boom, are now an AI data center hotspot with cooler climates and renewable energies cutting power costs. Here, NVIDIA’s GPU boom has fueled demand from a new class of providers: NeoClouds – GPU-as-a-service specialists building AI-native environments from the ground up. Hyperscalers, by contrast, still run broad, multi-purpose clouds and are only now densifying infrastructure for AI at scale. 

“In the coming years, NeoClouds are expected to account for a significant portion of demand,” says Dominic Ward, CEO of Verne, which operates centers in Iceland and Finland. Cologix observes the same outsourcing trend in Canada. “Instead of spending $75 million on GPUs and IT staff, [companies] can connect to GPU-as-a-service providers in our centers,” says President Sean Maskell. He compares the shift to cloud adoption, which saw hybrid deployments combining legacy equipment with cloud services. “Now we are seeing hybrid AI: businesses will control part of their IT while using AI-as-a-service from providers next door.” 

Like the Nordic countries, Canada is a climate and green-energy sweetspot and, in Alberta, a huge data infrastructure build-out is underway. “A couple of years ago, AWS invested in building a $4 billion data center here and told us: ‘we need to triple the number of data centers worldwide, and so do our competitors,’” recalls Rick Christiaanse, CEO of Invest Alberta. That project kickstarted the region’s infrastructure focus. Today, it is developing ‘Wonder Valley’, set to become the world’s largest AI data center park, with 49 projects worth $1 billion each.

Elsewhere, Malaysia is attracting new data infrastructure, offering proximity to tech-hub Singapore while providing more energy capacity. Indonesia, Morocco and South Africa are also gaining momentum. But when it comes to so-called AI factories, NVIDIA CSO Josh Parker points out that LLMs can be trained anywhere – client proximity isn’t an issue. “We say ‘AI doesn’t care where it goes to school’,” he says. “Having a fast connection to the internet doesn’t matter for AI training like it does for traditional data centers. Regions outside of today’s data center hubs could see new AI factories – if energy is available.”