NTT Global Data Centers: US AI boom surges as Europe builds momentum
Data‑center leader says US drives first wave of AI infrastructure while Europe accelerates investment in power‑hungry computing campuses
The artificial‑intelligence (AI) boom is creating unprecedented demand for computing infrastructure. The United States is absorbing most of the first wave of AI deployments, while Europe is beginning to scale investment to support the next phase of growth.
Executives at the world’s third‑largest data‑center provider say the US is currently moving fastest in building AI infrastructure. However, Europe is increasingly viewed as one of the most promising growth markets as governments, utilities, and technology firms accelerate new data‑center development.
“In the US, the AI demand is unquenchable. I’ve never seen anything like this in my 25 years of doing this. It is unbelievable,” Doug Adams, Chief Executive and President of NTT Global Data Centers, told TechJournal.uk in an interview during his recent trip to London.
“There is a very small actual installed AI base within Europe compared with the US. It’s shockingly small how little direct-liquid‑to‑chip (DLtC) infrastructure has been built across Europe,” Adams said, noting that the gap also highlights how much room there is for future expansion across the continent.
“Even the large AI companies like OpenAI and Anthropic are tenants inside other people’s data centers,” Adams said. “They rely on partners like us because the powered land and supply chains required to build these facilities are extremely difficult to replicate.”
AI infrastructure differs significantly from traditional enterprise or cloud computing environments.
“The big difference between AI and traditional cloud or enterprise infrastructure is that a large percentage of the infrastructure is directly tied to the chip,” Adams said. “The same facility can deliver both, but with some modifications depending on how the cooling and design are implemented.”
NTT Global Data Centers operates more than 150 data centers across over 20 countries and regions and supplies large‑scale computing facilities for cloud providers, enterprises, and AI developers. Its global footprint enables customers to deploy workloads near users while maintaining access to large pools of computing power.
The company is part of Japan‑based NTT Group, a global telecommunications and technology conglomerate with more than 300,000 employees worldwide. Within the group, the data‑center division focuses on designing, building, and operating large‑scale digital infrastructure used by cloud providers, multinational corporations, and AI developers.
Power bottleneck
One of the biggest factors shaping where new AI facilities are built is access to electricity.
“It’s difficult to get 100‑megawatt chunks of power here in Europe,” Adams said. “The biggest constraining issue for us is power.”
Large AI clusters consume enormous amounts of electricity. As a result, operators often need to secure energy capacity years before construction begins, and developers compete aggressively for sites with guaranteed power supply.
“We reserve the power ahead of time and secure powered land very early in the process,” Adams said. “If utilities cannot supply enough power immediately, we use bridging solutions such as gas turbines until the grid connection is ready.”
Bruno Berti, Senior Vice President of Global Product Management at NTT Global Data Centers, said these temporary solutions will remain critical as the AI industry expands.
“For the next three to five years, the key challenge will be bridging solutions,” Berti said. “Those help us operate until the grid connection arrives or until utilities can provide the full power capacity.”
The company is planning several large‑scale sites designed to support hundreds of megawatts of computing capacity. One planned campus in Germany could support roughly 500 megawatts of IT load, making it one of the largest data‑center developments in Europe.
Such projects remain relatively rare across Europe because grid capacity varies widely between countries. In many markets, data‑center developers compete with industrial users, renewable‑energy projects, and urban expansion for limited electricity capacity.
Europe’s AI race
While AI investment is rising across Europe, the continent has yet to establish a clear geographic hub for large‑scale AI infrastructure.
“We’re seeing most of the AI demand going into the Nordics right now, especially Sweden and Norway,” Berti said. “That’s largely because the cost of power is cheaper there compared with other parts of Europe.”
The Nordic region has become one of the most attractive destinations for AI data centers due to its abundant renewable energy, cool climates, and strong digital infrastructure. Lower electricity prices and access to hydroelectric power make countries such as Sweden, Norway, and Finland particularly appealing for power‑intensive AI workloads.
Electricity pricing, regulatory frameworks, and infrastructure planning will likely determine which European countries eventually become major AI infrastructure hubs.
Berti said the Nordic momentum does not mean other European markets are standing still. Several countries are seeing rising demand as cloud providers and AI developers expand their presence across the region.
“Power pricing and legislation across different EU countries will determine where the AI hubs eventually emerge,” Berti said. “At the moment, it hasn’t fully been determined which country will win.”
Large digital‑infrastructure markets such as Germany, the United Kingdom, the Netherlands, and Ireland continue to attract strong demand because of their connectivity and proximity to major enterprise and cloud customers. However, these mature markets are also facing tighter power constraints, forcing developers to search for new locations across the continent.
As a result, many operators expect the next wave of AI infrastructure investment to spread across multiple European regions rather than concentrating in a single hub.
Another factor shaping infrastructure distribution is the growing emphasis on data sovereignty.
Governments increasingly require sensitive data to remain within national borders, pushing technology companies to build local computing infrastructure.
“Data sovereignty and private data requirements are probably the biggest trend we’re seeing,” Adams said. “As regulations tighten, more infrastructure will need to be built in the same countries where the data is generated.”
“As sovereign data requirements increase, you’ll see AI infrastructure appear in those countries because the data has to stay within the country,” Berti said. “That will create smaller pockets of AI infrastructure around the world.”
Together, regulatory pressure, growing enterprise demand, and improving energy infrastructure are gradually accelerating AI data‑center development across Europe, even as the United States continues to dominate the current wave of large‑scale deployments.
Cooling innovation
The rapid expansion of AI workloads is also driving innovation in how data centers manage heat generated by advanced processors.
Modern AI chips consume far more power than conventional server processors, creating major cooling challenges for operators.
“Direct immersion is the most efficient way of cooling servers,” Adams said. “You literally place the servers in a dielectric fluid that dissipates the heat directly.”
However, immersion systems can be complex to deploy at scale, so many operators are adopting hybrid approaches.
“The more practical approach right now is direct-liquid‑to‑chip cooling,” Adams said. “A heat sink sits directly on the CPU or GPU and removes heat far more efficiently than traditional air cooling.”
Cooling systems themselves are also evolving rapidly as data‑center operators seek to reduce energy consumption.
“There is a huge amount of innovation happening in cooling technology right now,” Berti said. “It’s not just about removing heat from the chips, but also getting the heat out of the building more efficiently.”
Operators are also experimenting with ways to recycle waste heat from data centers to supply heating to nearby communities. In several European cities, district‑heating systems already use excess heat from data‑center facilities to warm residential buildings, helping reduce overall energy waste.
Supply chains and geopolitics
Beyond energy and cooling challenges, geopolitical tensions are increasingly influencing data‑center supply chains.
Long‑term procurement strategies have helped large infrastructure providers manage equipment price fluctuations and supply constraints.
“Because we plan out in three‑year increments, we haven’t been overly affected by tariffs or supply‑chain disruptions,” Adams said. “We purchase equipment years in advance and maintain strong strategic frameworks with our suppliers.”
Regional politics can still influence sourcing decisions.
“In the US, we purposely do not use Chinese equipment inside our data centers,” Berti said. “In other parts of Asia, it may make sense to use Chinese supply chains depending on the customers and market requirements.”
Scale has become a key competitive advantage as demand for AI infrastructure surges. Large operators can secure equipment, land, and power capacity years ahead of projects, creating barriers for smaller developers.
“We have hundreds and hundreds of megawatts of equipment pre‑ordered and ready to ship to our data centers,” Adams said. “It would be very difficult for large AI providers or smaller developers to replicate that kind of supply chain.”
NTT Global Data Centers is currently developing dozens of new facilities worldwide and expects annual growth to exceed 20% as demand for AI infrastructure continues to accelerate.
Adams said the rapid expansion of AI computing will likely reshape global digital infrastructure over the next decade as governments, technology firms, and cloud providers race to deploy increasingly powerful computing clusters to support new AI models and services.