One year later from the UK’s ambitious AI strategy: has its infrastructure development been a triumph?

In January, when the U.K. revealed its AI Opportunities Action Plan — an ambitious framework to integrate the technology throughout society — Prime Minister Keir Starmer asserted that the strategy aims to transform the nation into an “AI superpower.”

A significant element of this initiative was the swift establishment of data centers equipped to handle the extensive computational demands associated with AI deployment. This would be facilitated by “AI growth zones” — specified areas where planning permissions are eased and power access is enhanced.

Almost a year later, Nvidia, Microsoft, and Google have pledged billions toward AI infrastructure in the nation. Four AI growth zones have been announced, and local startups like Nscale are rising as important figures in the field.

However, detractors highlight heavily restricted energy access through the national grid and sluggish development as indicators that the country may fall behind its global competitors in the AI competition.

“There’s a disconnect between aspiration and execution,” remarked Ben Pritchard, CEO of data center power provider AVK, in an interview with CNBC.

“Expansion has been stunted mainly due to limitations on power accessibility. Grid congestion, in particular, has decelerated development and means the U.K. isn’t rolling out infrastructure swiftly enough to match global contenders.”

The journey of developing AI infrastructure in the U.K. is still in its early stages, as AI growth zones are presently in initial development stages.

A location in Oxfordshire, the first to be disclosed in February, has not yet commenced construction and is still evaluating proposals from delivery partners. Preliminary site preparation has started in one area in the North East of England, revealed in September, with formal construction expected to kick off in early 2026.

Two additional sites in North and South Wales were introduced in November. The North Wales site is in search of an investment partner, which the Department of Science, Technology and Innovation (DSIT) indicated to CNBC is anticipated to be confirmed in the coming months. The South Wales site consists of several locations, some operational while others are slated for further development, according to DSIT.

The U.K. government stated in July that it aimed to establish a core set of AI growth zones catering to a minimum demand of 500 megawatts by 2030, with at least one growing to over one gigawatt during that period.

Yet the most significant obstacle in achieving those goals is the U.K.’s constrained grid capacity, Pritchard asserted.

“Developers anticipate grid connection delays between eight to ten years, and the number of pending connection applications, particularly around London, is unprecedented,” he informed CNBC.

AI workloads are also “significantly escalating energy requirements” as businesses and consumers increasingly utilize the technology, adding further strain to an already stretched energy framework, Pritchard noted. “These are no longer isolated threats; they are actively hindering or obstructing developments nationwide.”

The outreach for applications for the AI growth zone initiative created a scenario where landowners with transmission lines or power cables running through their properties sought designation, commented Spencer Lamb from Kao Data.

“This led to the national grid being overwhelmed by power grid applications from speculative entities,” with little chance of actual success, he stated to CNBC.

The National Energy System Operator (Neso) — the U.K.’s agency in charge of overseeing the national grid — has taken steps to address the issue.

This month, it disclosed intentions to fast-track numerous projects to gain quicker access to the grid. Neso refrained from commenting on whether AI infrastructure projects are included in those prioritized when questioned by CNBC but stated a significant portion involved data centers.

Considerable financial commitments have also come from technology giants, many of which were highlighted by the U.K. government in September.

Microsoft, Nvidia, Google, OpenAI, CoreWeave, and others proclaimed multi-billion dollar investments in AI during U.S. President Donald Trump’s state visit, including plans to deploy the newest chips in the nation and set up new data centers.

The homegrown startup Nscale, which facilitates access to AI computing and is constructing data centers, also revealed agreements to implement tens of thousands of Nvidia chips at an AI facility just outside London by early 2027.

“Investment from leading private entities has established crucial groundwork,” stated Puneet Gupta, general manager for the U.K. and Ireland at data infrastructure firm NetApp, in an interview with CNBC. “There is also growing momentum regarding national research supercomputers and plans for additional computing capacity, with pledges to create AI ‘gigafactories’ in the U.K.”

However, the “real challenge” will be the speed at which these plans convert into usable computing resources for U.K. businesses, Gupta remarked.

The long-term success of the nation’s AI infrastructure development will necessitate investing in the “full stack,” encompassing data pipelines, storage, energy procurement, security, talent, and skills, according to Stuart Abbott, U.K. and Ireland’s managing director at AI infrastructure company VAST Data, in his comments to CNBC.

This requires “creating an operational network that allows genuine institutions to implement AI securely on a large scale,” he continued. “If the UK aspires for sustainability over a fleeting surge, it must regard AI infrastructure as akin to economic infrastructure.”

The hurdles are considerable. The financial figures for data center deals in Europe pale in contrast to the amounts funneled into U.S. initiatives. The U.K. also currently endures the highest energy costs in Europe, which are approximately 75% higher than pre-Russia invasion of Ukraine levels, alongside an aging grid infrastructure that may take many years to connect to new sites.

A potential answer for projects unable to secure access to the national grid involves microgrids, according to AVK’s Pritchard. Microgrids are self-sufficient power networks drawing energy from sources such as engines, renewables, and batteries.

AVK is in the process of designing two microgrids for collaborators constructing cloud computing facilities, although not for AI, within the U.K. These can take around three years to establish and typically cost about 10% more than energy supplied from the grid at this point, Pritchard explained.

Locating computing resources where power is already available, instead of “relying solely on undeveloped sites,” is another method to expedite the establishment of AI infrastructure, remarked VAST Data’s Abbott.

The speed of execution will be essential, Lamb from Kao Data warned CNBC. “Unless fundamental issues around energy accessibility and pricing, AI copyright, and funding for AI initiatives are addressed swiftly, the U.K. will forfeit one of the most extraordinary economic chances of our era and ultimately risk becoming a global AI backwater.”