Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
The QTS data center in Cambois, north-east England
When the UK announced its AI Opportunities Action Plan – a grand plan to deploy the technology across society – in January, Prime Minister Keir Starmer said the strategy would make the country an “AI superpower”.
A key pillar of this plan was the rapid construction of data centers capable of meeting the enormous computing needs needed to deploy AI. This would be driven by “AI growth zones” – designated areas with relaxed planning permission and improved access to electricity.
Almost a year later, and Nvidia, MicrosoftAnd Google have all committed billions of dollars to the country’s AI infrastructure. Four AI growth zones have been revealed and local startups like Nscale have emerged as key players in this area.
But critics point to severely restricted access to energy through the national grid and slow construction as signs the country risks falling further behind its global rivals in the AI race.
“The ambition and results are not yet aligned,” Ben Pritchard, CEO of data center energy provider AVK, told CNBC.
“Growth has been held back largely by constraints on electricity availability. Grid bottlenecks, in particular, have slowed the pace of development and mean that the UK is not yet deploying its infrastructure quickly enough to keep pace with its global competitors.”
Network connection delays
The development of AI infrastructure in the UK is still in its early stages, as AI growth areas are currently in their early stages of development.
A site in Oxfordshire, the first announced in February, has not yet started construction work and is still considering proposals from delivery partners. Ground preparation work has begun in the north-east of England, announced in September, and official construction will begin in early 2026.
Two further sites, in north and south Wales, were revealed in November. The first is seeking an investment partner, which the Department of Science, Technology and Innovation (DSIT) told CNBC it expects to be confirmed in the coming months. The latter is made up of a set of sites, some of which are already operational and additional construction work is to be carried out on others, said the DSIT.
The British government said in July that it was targeting a core group of AI growth areas meeting demand of at least 500 megawatts by 2030, with at least one expected to reach more than a gigawatt by then.
But the most serious challenge to realizing these ambitions is the limited capacity of the UK network, Pritchard said.
“Developers are expecting grid connection delays of eight to 10 years, and the volume of pending connection requests, particularly around London, is unprecedented,” he told CNBC.
AI workloads also “significantly increase energy demand” as businesses and consumers begin to use the technology, putting additional pressure on a strained energy system, Pritchard added. “These are no longer isolated risks; they are actively slowing or blocking developments across the country.”
The open call for applications for the AI growth zone initiative created a situation where landowners with power towers or cables crossing their land applied for the designation, said Spencer Lamb of Kao Data.
“This resulted in the national grid being flooded with power grid requests from speculative sources,” without any realistic chance of success, he told CNBC.
Lay the foundations
The National Energy System Operator (Neso) – the UK public body responsible for managing the national grid – has taken steps to remedy the situation.
Earlier this month, it announced plans to prioritize hundreds of projects for faster access to the network. Neso declined to say whether AI infrastructure projects were among the priority projects when asked by CNBC, but said a significant portion were data centers.
Tech giants have also made significant financial commitments, many of which were presented by the UK government in September.
Microsoft, Nvidia, Google, OpenAI, CoreWeave and others announced billions of dollars of investment in AI during the state visit of US President Donald Trump, who planned to deploy the latest chips in the country and open new data centers.
Local startup Nscale, which provides access to AI computing and builds data centers, also announced deals to deploy tens of thousands of Nvidia chips in an AI factory just outside London by early 2027.
The Nvidia GB10 Grace Blackwell superchip is demonstrated during the company’s GTC conference in San Jose, California on March 19, 2025.
Max A. Cherney | Reuters
“Investments from major private players have laid important foundations,” Puneet Gupta, managing director for the UK and Ireland at data infrastructure company NetApp, told CNBC. “Momentum is also building around national research supercomputers and projects to create new computing capabilities, with commitments to build AI ‘gigafactories’ in the UK.”
But the “real test” will be how quickly these plans translate into usable calculations for UK organisations, Gupta said.
Avoiding the “sugar rush” of AI infrastructure
The long-term success of developing the country’s AI infrastructure will require it to invest in the “full stack”, including data pipelines, storage, energy supply, security, talent and skills, Stuart Abbott, UK and Ireland managing director at AI infrastructure company VAST Data, told CNBC.
“If the UK wants this to be sustainable rather than a year-long sugar rush, it needs to treat AI infrastructure as economic infrastructure.”
Stuart Abbott
Managing Director for UK and Ireland at AI infrastructure company VAST Data
This means “developing an operational fabric that allows real institutions to safely deploy AI at scale,” he added. “If the UK wants this to be sustainable rather than a year-long sugar rush, it needs to treat AI infrastructure as economic infrastructure.”
The challenges are significant. The value of data center contracts in Europe is nothing in comparison to the sums injected into projects in the United States. The UK also currently has the most expensive energy in Europe, around 75% higher than before Russia’s invasion of Ukraine, and existing network infrastructure that can take many years to connect to new sites.
Microgrids are a potential solution for projects that fail to secure access to the national grid, AVK’s Pritchard said. Microgrids are self-sustaining electricity networks from sources such as motors, renewable energy and batteries.
AVK is currently designing two microgrids for partners building cloud computing, but not AI, in the UK. They can take about three years to build and currently cost about 10 percent more than grid power, according to Pritchard.
Collocating compute where power already exists, rather than “forcing everything to be entirely new” — the term for undeveloped sites — is also a way to get AI infrastructure up and running more quickly, VAST Data’s Abbot said.
The pace of implementation will be critical, Kao Data’s Lamb told CNBC. “Unless fundamental issues around energy availability and pricing, AI copyright and funding for AI development are addressed quickly, the UK will miss out on one of the most remarkable economic opportunities of our time and risks ultimately becoming an international AI backwater.”
