Summary

• London based Valarian has raised a $50m Series A led by NEA, bringing total funding to $70m.
• The company is building sovereign digital infrastructure for government, defence, and enterprise AI use cases.
• The raise reflects growing demand for AI systems that preserve control over data, operating environments, auditability, and supplier dependency.

Valarian has raised a $50 million Series A led by New Enterprise Associates, as the London based company expands sovereign digital infrastructure for government, defence, and enterprise AI systems.

The round brings Valarian’s total funding to $70 million and marks NEA’s first defence and dual use investment in Europe. Lightbank, XTX Ventures, Sequel, Litquidity VC, Gokul Rajaram, and Nikesh Arora also participated, according to the company.

Valarian says the funding will accelerate the rollout of its UK sovereign digital architecture across organisations that need to use advanced AI while retaining control over sensitive data. Its ACRA platform is designed to let organisations deploy AI securely while keeping data inside their own environment, with the company describing its wider portfolio as infrastructure for governing, securing, and scaling AI across government, defence, and enterprise.

Sovereign AI has become a heavily used phrase, sometimes covering everything from domestic model development to cloud procurement preferences. Valarian’s pitch is narrower and more infrastructure focused. It is not only about whether a model was trained in the UK or whether a cloud region is nearby. It is about who controls the environment in which AI operates, which systems it touches, how data is governed, and how decisions are logged.

That distinction is becoming more important as AI moves into high consequence workflows. Government departments, defence organisations, regulated industries, and large enterprises are exploring AI for analysis, automation, knowledge work, decision support, and operational coordination. The more deeply those systems connect to internal data and workflows, the harder it becomes to treat AI as a simple software subscription.

Data sovereignty is only one part of the problem. Organisations also need to understand model access, supplier dependency, audit trails, fallback options, user permissions, prompt and output handling, and the governance of agents that may be able to trigger actions across enterprise systems. A system that can summarise documents is one risk profile. A system that can access operational data, recommend actions, and interact with critical workflows is another.

Valarian’s focus on government and defence gives the company a sharper route into that market, but it also raises the delivery bar. Buyers in those sectors will expect security assurance, policy controls, deployment flexibility, and evidence that systems can operate inside constrained environments. They will also want to avoid new forms of dependency where “sovereign” software merely wraps global infrastructure without changing control in practice.

The investor context is notable. NEA is a major US venture firm, and its first European defence and dual use investment reflects how sovereign infrastructure has become an investable category rather than only a government procurement theme. That brings capital into UK defence technology and AI infrastructure, although it also creates a tension: companies promising national or regional control may still depend on global investors, suppliers, cloud ecosystems, and component markets.

The UK government has been emphasising sovereign AI capability through its Sovereign AI Fund, AI hardware planning, and wider industrial strategy. Valarian’s raise sits alongside that policy direction, but private funding will not by itself settle whether the UK can build durable AI infrastructure. Compute capacity, procurement reform, cyber assurance, security cleared engineering talent, and buyer willingness to move beyond familiar suppliers will all influence whether domestic companies can scale.

Market discipline will still apply. Sovereignty will carry Valarian only so far if the product is harder to adopt than conventional cloud AI services or if it adds cost without clear control gains. The most credible sovereign infrastructure companies will need to prove that stronger governance and operational control can coexist with usable systems, integration speed, and measurable productivity.

Valarian’s funding round shows the UK sovereign AI debate moving from policy language into company building. The next question is not whether organisations want more control over AI. It is whether new infrastructure providers can make that control practical inside the messy, regulated, and supplier dependent environments where sensitive work happens.