One year after the UK unveiled its ambitious AI Opportunities Action Plan, the country’s progress in building AI infrastructure remains a mix of promise and persistent challenges. Prime Minister Keir Starmer had pledged to transform the UK into an "AI superpower," with a central focus on rapidly expanding data center capacity through designated AI growth zones—areas offering streamlined planning permissions and better access to electricity. Since then, major tech players including Nvidia, Microsoft, and Google have committed billions to AI infrastructure in the UK. Four AI growth zones have been officially announced, and homegrown companies like Nscale have emerged as key contributors. However, despite these developments, significant obstacles continue to hinder the pace of deployment. A major bottleneck is the UK’s constrained national grid. According to Ben Pritchard, CEO of data center power supplier AVK, developers are facing grid connection delays of eight to ten years, with demand far outpacing available capacity. "Growth has been held back largely by constraints around power availability," Pritchard told CNBC. "Grid bottlenecks are actively slowing down or blocking developments across the country." The government’s target to have at least 500 megawatts of AI infrastructure in core growth zones by 2030, with one zone reaching over one gigawatt, now seems increasingly difficult to meet. The surge in connection requests—especially around London—has overwhelmed the system, and many applications have come from landowners with existing power infrastructure, rather than serious developers, according to Spencer Lamb of Kao Data. This has created a flood of speculative applications, further delaying real progress. Experts stress that long-term success requires more than just data centers. Stuart Abbott, managing director at VAST Data for the UK and Ireland, emphasized the need to invest in the full AI infrastructure stack—data pipelines, storage, energy sourcing, cybersecurity, and workforce development. He noted that co-locating compute facilities where power is already available, rather than building on undeveloped "greenfield" sites, could accelerate deployment. Microgrids—self-contained power systems using renewables, batteries, and backup generators—offer a potential workaround. AVK is currently designing two such systems for cloud compute projects, though not yet for AI. While microgrids can be built in about three years and cost around 10% more than grid power, they provide a viable alternative when grid access is blocked. Still, the pace of implementation is critical. Lamb warned that without swift solutions to energy access, pricing, AI copyright issues, and funding, the UK risks missing a historic economic opportunity. "Unless these fundamental issues are addressed quickly, the U.K. will miss out on one of the most remarkable economic opportunities of our time and ultimately risks becoming an international AI backwater."