As Britain struggles to establish its AI credentials in an increasingly crowded field, London-based Valarian has secured a major £37m funding boost. The cash injection, led by Silicon Valley's New Enterprise Associates (NEA), is a vote of confidence in the company's mission to give governments and defence organisations control over their own data and security infrastructure.
Valarian CEO Max Buchan warns that countries should focus on controlling the environment where AI operates – not just developing cutting-edge models. He argues that while AI models are interchangeable, the operating environment is a constant advantage, giving the UK an edge in retaining control over its data and security infrastructure.
The timing of this investment couldn't be more crucial, as MPs have issued a stark warning: Britain risks falling behind in sovereign AI capabilities without a clear strategy to reduce dependence on overseas technology. The Science, Innovation and Technology Committee urged ministers to outline a comprehensive plan, highlighting the risk that foreign partners could restrict access to critical technologies.
AI Minister Kanishka Narayan welcomed the investment, stressing that Britain must build its own sovereign AI capabilities as AI becomes increasingly vital for both economic growth and national security. Defence Procurement Minister Luke Pollard described it as a significant vote of confidence in Britain's defence tech sector – with Valarian's platform allowing government departments to deploy multiple AI models while keeping sensitive data within their secure infrastructure.
The push for sovereign AI is gathering pace in Westminster, driven by geopolitical tensions and concerns over US tech giants' dominance. Danni Hewson, head of financial analysis at AJ Bell, notes that these factors have elevated sovereign AI on the government's agenda – particularly regarding sensitive data tied to critical infrastructure and defence. Yet, Forrester analyst Dario Maisto cautions that true sovereignty goes beyond software, ultimately relying on the underlying platform.