Valarian Raises $50 Million Series A Led by NEA to Build a Sovereign Control Layer for AI Workloads
Valarian has raised $50 million in Series A funding led by New Enterprise Associates, bringing the London-based company’s total raised to $70 million. The round is NEA’s first defence and dual-use investment in Europe. Lightbank, XTX Markets, Sequel and LitVC participated, along with angel investors Gokul Rajaram and Nikesh Arora.
The product is a control layer. Valarian enforces workload-level governance across the environments where critical applications, AI systems and operational workloads run — dictating how those systems communicate, access data and operate. The governance lives at the infrastructure layer rather than in each application, and every deployed capability inherits it automatically.
Why the Control Plane, Not the Model
The pitch rests on a distinction worth taking seriously. Most AI security spending to date has gone into the model and the application surrounding it — guardrails, prompt filtering, output monitoring, access controls bolted onto individual deployments. Valarian’s argument is that this is the wrong altitude. If governance is implemented per-application, it fragments, drifts and eventually fails at the seams. Enforced beneath the workload, it holds regardless of what runs on top.
CEO and Co-Founder Max Buchan framed the problem as consolidation: the intelligence layer of Western institutions moving, contract by contract and department by department, into systems those institutions do not control. His position is that sovereignty cannot be retrofitted — it is an architectural property that has to exist from the ground up.
NEA partner Mustafa Neemuchwala’s version of the thesis is that the defining question of the AI era is not which model wins but who controls the environment intelligence operates inside.
Two Deployment Tracks
The capital funds two lines:
- Valarian Enterprise — organisations running AI and other high-consequence workloads that need workload-level governance, compartmentalisation and operational control.
- Valarian Defence — sovereign nations and defence programmes where control over mission-critical workloads is non-negotiable.
The dual-use structure is deliberate. The technical requirements that defence organisations have spent decades building bespoke infrastructure to satisfy — compartmentalisation, provable control over data flows, isolation between classification tiers — are now becoming ordinary enterprise requirements as AI moves into operationally critical positions. A platform built to defence tolerances can be sold down-market into regulated enterprise. The reverse rarely works.
The Political Layer
The round arrived with unusual political weight. UK Minister for AI and Online Safety Kanishka Narayan supplied a supporting quote, positioning Valarian alongside the government’s Sovereign AI Fund and AI Hardware Plan, and framing sovereign AI capability as a prerequisite for Britain shaping its own destiny. The release cites European defence spending of €392 billion in 2025 as the backdrop.
A sitting minister appearing in a Series A announcement is a signal about where this category sits. Sovereign infrastructure is being treated as industrial policy, not merely as a venture bet — which means procurement tailwinds, but also political exposure.
What to Watch
The sovereignty argument has been climbing the stack for several years: first silicon, then cloud compute, then models, and now the governance and control plane above all of it. Valarian is betting that the control plane is where the durable position sits, because it is the layer that survives model churn.
The open questions are the usual ones for infrastructure at this stage. Governance enforced beneath the workload is a strong claim that has to survive contact with heterogeneous environments — legacy systems, multiple clouds, air-gapped enclaves, and the accumulated exceptions that every real organisation runs on. And the defence sales cycle, however favourable the political climate, remains slow. Valarian now has $70 million and a well-timed thesis to work with.