{
  "id": 18289561,
  "title": "The UK's AI strategy is missing one crucial piece: compute",
  "url": "https://urgent.news/2026/11/05/the-uks-ai-strategy-is-missing-one-crucial-piece-compute",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-05T09:58:10.000Z",
  "source": {
    "name": "TechRadar",
    "slug": "techradar",
    "url": "https://www.techradar.com/pro/the-uks-ai-strategy-is-missing-one-crucial-piece-compute"
  },
  "original_language": "en",
  "account": "As Britain embraces AI as both a national security and economic priority, a crucial aspect of its AI strategy has been largely overlooked: the availability of computing power. While the UK boasts strong universities, a robust startup ecosystem, and a growing number of companies developing AI products, their success heavily relies on reliable access to computing resources. However, the largest cloud providers and the biggest AI companies are committing significant resources to AI infrastructure, leaving smaller companies and startups struggling to secure the compute they need. This is due to limited competition for available capacity, with only 32 countries hosting specialized AI data centers, and American tech giants controlling over 90% of them. The consequences of this imbalance are significant. Smaller British AI companies, even those with sizable funding, face a different economic landscape compared to those able to secure multi-billion-pound infrastructure deals. They lack the negotiating power and long-term commitments required to access compute on favorable terms. Rental prices for Nvidia's latest GPUs soared by nearly half in two months, while CoreWeave, a leading specialist provider, increased prices and extended minimum contracts. This creates a paradox: while there is a shortage of GPUs, much of the existing capacity remains idle, with an average GPU utilization of just 5%. Some analysis of around 23,000 cloud clusters revealed that organizations pay for up to twenty times more capacity than their workloads actually use. To address this issue, pooling GPU capacity from data centers, infrastructure providers, and enterprise environments into a common marketplace would provide British AI companies with an alternative source of compute, particularly when they are unable to negotiate directly with the largest providers or do not wish to lock themselves into a single cloud. This approach would also align with the UK's sovereignty concerns, as the pooled capacity would be under UK jurisdiction, ensuring that AI infrastructure benefits the entire nation, not just the largest players. Currently, much of the country's computing power is scattered across university clusters, company server rooms, regional data centers, and specialist providers, insufficient in scale to support AI workloads independently. Streamlining this capacity would enable British AI companies to access the computing power they need under local control, fostering regional growth and maintaining national independence in AI development.",
  "summary": "The UK needs compute sovereignty, not just AI partnerships, to build competitive companies.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}