The UK government spends less than £1 in every £6 with its largest technology suppliers on British firms, according to a report from the Institute for Public Policy Research (IPPR) published this week. While officials talk about buying British, the research exposes how deeply Whitehall depends on foreign technology. Once researchers looked past where companies are headquartered to trace the actual origins of products and services, American firms emerged as the overwhelming winners.
The headline figure tells only part of the story. At first glance, 42.5 percent of spending with the public sector's 150 largest technology suppliers appears to flow to UK companies, according to the IPPR analysis. But that number plummets to just 15.4 percent when researchers accounted for overseas ownership, foreign staffing, and the reality that many UK-headquartered firms simply resell American technology. In fact, 43 percent of spending credited to UK-headquartered suppliers goes to resellers, with Bytes and Softcat alone capturing roughly £2 billion and ranking among the three largest recipients of public sector technology spending. When contracts were traced back to the actual product makers, the United States accounted for £9.2 billion of the £19.2 billion spent with the largest suppliers. The concentration is stark: just 150 companies receive 83 percent of public sector technology spending, and the largest 15 firms collect nearly half. Smaller suppliers fare poorly, receiving only 12 percent of government technology spending compared with 27 percent of overall government spending. Competition data raises additional concerns, with 89.9 percent of government IT contracts over the past 12 months lacking any published record of how many suppliers bid for the work.
The report warns that Britain's dependence on American technology could become especially problematic if Washington restricts access to advanced AI systems, pointing to recent US restrictions on Anthropic models and limits on early access for Britain's AI Security Institute as evidence the threat isn't hypothetical. According to IPPR senior research fellow Roa Powell, "we cannot do it alone, and building domestic capability can't be our only bet." Darren Jones MP, who wrote the report's foreword, made a similar case, arguing Britain should collaborate with international allies rather than attempting to build everything domestically.
The challenge deepens when AI enters the equation. Government spending records show relatively little money flowing directly to companies like OpenAI and Anthropic, but that doesn't mean Whitehall isn't purchasing their technology—much of it is bundled into existing cloud agreements, enterprise software licenses, and partnerships. The report cites NHS England's rollout of Microsoft 365 Copilot to 505,000 staff, which was carried out under a national licensing agreement signed in June 2023, before Copilot was even generally available. That means AI tools can reach hundreds of thousands of public sector workers through deals negotiated before those products existed, potentially blocking rival suppliers from competing. The government has already committed £500 million to a Sovereign AI Fund, but researchers argue throwing money at homegrown technology companies won't be sufficient to loosen Silicon Valley's grip. Instead, they recommend Whitehall negotiate technology contracts as a single customer, make it easier to switch suppliers, and work with countries including France, Germany, Japan, and South Korea to secure access to advanced AI. Whitehall may have grand plans for digital sovereignty, but it's facing an uphill battle when so much of its technology comes from across the Atlantic—and with AI already bundled into software contracts worth billions, US tech giants aren't making it easy for the government to shop elsewhere. The procurement patterns revealed suggest that sovereignty ambitions will require structural changes to how technology is purchased, not just funding announcements. For decision-makers in allied governments and enterprise buyers, the underlying tension between operational convenience and strategic autonomy will only intensify as AI becomes infrastructure rather than product.