The UK government has formally launched Sovereign AI, a new €573 million (£500 million) initiative designed to back British AI startups working in areas seen as strategically important for both economic growth and national security.
Positioned as a more agile, state-backed effort, the programme is intended to support homegrown AI companies with not only capital, but also access to compute, talent and government support, as ministers look to ensure that critical AI capability is developed and retained in the UK.
Technology Secretary Liz Kendall said: “We believe in Britain and we are betting on Britain. We are backing our brilliant innovators and entrepreneurs so we seize the benefits of AI to reshape Britain for the benefit of all. Sovereign AI is unlike anything government has ever done before. Its unique approach will help break down the barriers that have too often held back British enterprise and innovation. This is how we ensure Britain’s economic prosperity and national security in the modern age.
“My message to British founders and innovators is clear – we will ensure you never have to choose between your ambition and your home, because Britain will give you both.”
EU-Startups coverage suggests that UK AI funding in 2025–2026 has been characterised less by consumer-facing applications and more by infrastructure, agentic enterprise software, and tools designed for regulated or strategically sensitive environments.
Recent UK rounds span physical AI data infrastructure, healthcare workforce systems, AI-agent security, defence and robotics, industrial materials science, drug safety, finance automation and payments for AI agents, with London featuring prominently across the sample.
That pattern makes the launch of Sovereign AI look consistent with an existing market direction: public backing is arriving in a national ecosystem where private capital has already been flowing towards AI capability that is operational, infrastructure-heavy, and often relevant to sovereignty, resilience or regulated deployment.
EU-Startups’ reporting also indicates that the sums vary sharply, from low-single-digit Seed rounds to very large infrastructure financings such as Nscale (€936 million), suggesting a market split between early applied-AI builders and a smaller set of companies raising at much larger scale to provide the compute and deployment backbone.
Chair of the Sovereign AI Unit, James Wise, adds: “AI as a technology could be transformational for both our wealth and security. Britain has the foundations be a global AI leader in many fields, with a unique and enviable mix of talent, capital, and infrastructure which make this country the natural home for world-leading innovation. Now, through Sovereign AI, we can use the state’s unique capabilities to double down on these strengths, backing Britain’s founders to scale here in the UK and globally.”
Sovereign AI is being presented as a government vehicle that will operate more like a venture capital fund, investing directly in promising AI startups and helping them scale at speed. Its stated focus is on companies building technologies that could have a broad societal impact, from advanced healthcare applications to core AI infrastructure and next-generation chips.
The first equity investment from Sovereign AI will go to Callosum, a startup building a new class of AI infrastructure. Alongside that investment, six further startups have been selected to receive access to the UK’s AI Research Resource supercomputer network.
Those companies are Prima Mente, Cosine, Cursive, Doubleword, Twig Bio and Odyssey. EU-Startups had already covered Cosine in August 2024, when they raised €2.2 million to build autonomous AI software developers. We also covered Doubleword in May 2025, reporting a €10.6 million round for its self-hosted AI inference platform aimed at enterprises.
For these startups, one of the key benefits is access to the kind of computing power that is often prohibitively expensive for young companies trying to train advanced models, run simulations or test new architectures.
According to the announcement, startups backed by Sovereign AI will also receive support that includes fast-tracked visa decisions for global talent, as well as government help in navigating data access, procurement opportunities, product validation and regulation.
The broader ambition is to stop promising British AI companies from relocating abroad as they grow, while keeping expertise, decision-making and value creation within the UK.
Rachel Reeves, Chancellor of the Exchequer, tied the announcement directly to the government’s wider economic plans, arguing that AI is one of the sectors where Britain should aim to build globally competitive companies from the outset rather than simply adopting technologies developed elsewhere.
The initiative was announced at an event hosted by Wayve, itself one of the UK’s best-known AI success stories.
Among the selected startups, Cosine adds a useful example of the kind of company Sovereign AI is seeking to support. The London-based startup, which builds sovereign AI coding models for defence, national security and other regulated sectors, has been granted 500,000 GPU hours on Isambard-AI through the AI Research Resource.
According to the company, that allocation is worth millions of pounds in infrastructure value and will allow it to train a fully sovereign AI model entirely on British soil. The Sovereign AI venture arm has also secured the option to participate in Cosine’s next funding round.
That selection fits neatly with Sovereign AI’s positioning. Cosine is focused on customers operating in highly sensitive environments where sending code to foreign-managed servers may be legally or operationally impossible. Its platform is designed to run within a customer’s own infrastructure, without internet access or external model dependencies, and supports more than 38 programming languages including legacy systems such as Fortran, COBOL, Ada and Verilog.
In other words, it is exactly the sort of startup that speaks to the government’s twin priorities of AI capability and national resilience.
“For two years we’ve been telling defence primes and critical infrastructure operators that we can do what no one else can: air-gapped, on-premise, trained on the legacy code that runs Britain’s most sensitive systems,” says Yang Li, COO and co-founder. “The one thing we couldn’t say was that the model itself was trained on sovereign infrastructure. The AIRR grant completes that picture. The UK should be an AI maker, not an AI taker – and this is what that looks like in practice.”
More broadly, the launch of Sovereign AI marks a notable shift in tone from the UK government. Rather than limiting itself to creating favourable conditions for the private sector, it is taking a more active role in shaping which AI companies get the infrastructure and backing needed to compete.
With around 30 firms reportedly already in discussions over potential AI Research Resource access, and a separate funding call also being launched for datasets and related assets, this first cohort appears to be only the beginning for the UK as it looks to promoting British AI innovation.
Read the orginal article: https://www.eu-startups.com/2026/04/britain-puts-e573-million-on-the-table-as-sovereign-ai-names-its-first-startup-cohort/
