Civo plans to deploy 40 edge data centers across Britain to power sovereign AI, with the first site in Hertfordshire (March 2027) housing NVIDIA Vera Rubin systems and a longer-term ambition to scale to 1 GW capacity.

UK cloud provider Civo plans to build 40 edge datacenters across Britain to deliver sovereign AI services closer to customers, with the first facility scheduled to open in Hertfordshire by March 2027. The company aims for a longer-term capacity target of 1 GW across the country, powered by Nvidia Vera Rubin NVL72 systems. The Hertfordshire site will launch with 8 MW of capacity and eventually expand to 38 MW.

Civo has already secured five additional sites in Lancashire, Greater Manchester, London, and Oxfordshire. The company expects these locations to bring total capacity to 150 MW within three years.

At its recent Civo Navigate event in London, chief executive Mark Boost highlighted growing demand for data sovereignty among UK businesses. "For the past two years, we've surveyed 1,000 UK IT leaders about AI sovereignty," he said. "Last year, 61 percent said it was a strategic priority, and this year that's risen to 73 percent. That's really not a trend. That's the tipping point."

The pitch addresses growing anxiety about reliance on US-controlled cloud providers. The US CLOUD Act allows American authorities to seek data from US companies even when it is stored overseas. Currently, at least 80 percent of UK businesses using cloud services depend on AWS, Microsoft Azure, or Google Cloud.

Sovereignty, Boost argued, extends beyond physical location. "Your data sits in a UK building, but the company is owned and controlled from overseas, so your data is still subject to foreign laws," he said. "Sovereignty isn't just where your data sits; it's about who has it and who they answer to. And in these uncertain geopolitical times, there's a growing concern around who holds the digital kill switch."

The "kill switch" risk is not theoretical. In June, a US export control directive prompted Anthropic to suspend access to two of its advanced models, though the restrictions were lifted later that month. The incident demonstrated how abruptly access to US-hosted AI services could be restricted.

Civo also contends that edge capacity reduces latency between AI requests and responses. Nvidia estimates that inference—running a trained model—currently accounts for about 35 percent of AI workloads, with potential growth to 60–65 percent by 2030.

Civo claims it will be among the first UK cloud providers to offer the Vera Rubin platform once the Hertfordshire site goes live. The company is taking orders now and also offers existing capacity on Nvidia's B300, B200, H100, and H200 instances. Early access pricing for Vera Rubin starts at $10 per hour.