British technology company Valarian closed a $50 million Series A led by New Enterprise Associates (NEA), bringing total investment to $70 million. The round marks NEA’s first defence and dual-use investment in Europe, according to Valarian’s announcement and AI Business.
Lightbank, XTX Ventures, Sequel, and Litquidity VC also participated, alongside angel investors Gokul Rajaram and Nikesh Arora.
What Valarian Builds
Valarian’s core product is ACRA, a control architecture that wraps AI workloads in policy-enforced isolation on Kubernetes. Organizations deploy AI models and agents on any cloud provider or on-premises hardware. ACRA enforces identity, access control, network isolation, and audit logging per workload. Each agent or model runs inside its own enclave with a dedicated policy stack, and encryption keys stay with the customer.
The pitch is straightforward: use AWS or Azure for compute and scalability, but run Valarian’s control layer on top so no hyperscaler can access your data or override your policies. “The critical question of the AI era isn’t which model wins,” NEA Partner Mustafa Neemuchwala told Valarian. “It’s who controls the environment intelligence operates inside.”
UK Government Backing
UK AI Minister Kanishka Narayan praised the investment, calling Valarian a company that understands “the challenge that’s in front of us” and is “building the solutions that will help us deliver a safer and stronger Britain,” according to Valarian’s announcement. Defence Minister Luke Pollard separately described the round as “a strong vote of confidence in the UK’s world-leading defence and dual-use technology sector.”
That language positions Valarian alongside Britain’s broader Sovereign AI Fund and AI Hardware Plan, both designed to reduce dependency on US and Chinese technology infrastructure.
The Sovereign AI Market Is Growing
Valarian is not alone in this space, but few startups combine defence-grade architecture with commercial enterprise positioning. The company targets two deployment surfaces: government and military systems running mission-critical agentic workloads with strict audit requirements, and enterprises in regulated industries (financial services, healthcare, legal) that need operational control over AI without building from scratch.
The timing aligns with a broader shift. As enterprises deploy autonomous agents at production scale, the question of who controls those agents, their data, and their permissions has moved from theoretical to contractual. Citrix released its NetScaler MCP Gateway for centralized agent authentication and rate limiting. Ant Group open-sourced SingGuard-NSFA for guardrailing agent actions before execution. Valarian operates one layer beneath both: the infrastructure substrate that determines whether an organization or a cloud vendor holds the keys.
A Geopolitical Infrastructure Bet
CEO Max Buchan framed the raise as an infrastructure play for national sovereignty. “We need to build the sovereign digital infrastructure that allows government, defence and enterprise to adopt AI securely while remaining in control of their own data,” he said in the announcement.
For agent builders outside the US, the question is practical: if your agents run on US-hosted infrastructure, US regulations and CLOUD Act subpoenas apply to your data. Valarian’s control plane is designed to decouple the compute layer from the governance layer, giving non-US organizations a credible path to frontier AI adoption without sovereignty trade-offs. Whether that architecture holds under real geopolitical pressure remains to be tested.