Sovereign AI Launch Speech

We launched the Sovereign AI fund today, after four months of work at the heart of Government turning an idea into a one-of-a-kind venture fund. The event was opened by the Secretary of State for Science and Technology, and I closed the event with more details on our offer and goals.

Wayve HQ Kings Cross, 16th April

Thank you to our panel. And thank you to Wayve for hosting us here today.

I‘ve told this story before but the first time I met Alex I had assumed he was running a company and expected to meet a huge team of engineers.

Instead, I found myself in his student digs, and we sat between his kitchen sink and bedside table, as he explained why billion-dollar autonomy projects were getting it wrong, in his impeccable Kiwi manner.

Which obviously left me intrigued. A year or so later, my colleague Suranga and I went back up to Cambridge and by now Alex had raised some funding, so I expected to be driven around in some gleaming car covered with servers, lidars and cameras. Instead, we were presented with the Twizzy on my right, which as the Secretary of State rightly pointed out, is essentially a go-kart with an iPhone strapped to the top.

But amazingly it worked.

Years later, Wayve's cars have driven autonomously around hundreds of the world's biggest cities, built by a global team — but anchored here, in King's Cross. And as Alex said, they are still just getting started.

Which is a reminder that the next world-changing company might not look like much at first.

Sometimes we in the UK can be cynical about the student, by the kitchen sink with a controversial idea.

But despite that, the team at Wayve chose to be here. That choice, I believe, has been a huge advantage for them, and for all of us in Britain. And if there is one message I hope you take away from this evening, it is this: Sovereign AI is here to make that choice, to scale here, easier.

Now it's my job to share more of the specifics of how we will do that. And how our
efforts complement the many other things Government does, and has done, to
support AI.

First and foremost, we are venture investors. Not grant-makers. Not a policy unit. Not a lobbying body. Investors, with a mandate to back the best AI companies in Britain, in areas of national priority, on commercial terms.

Venture investing is embedded in the DNA of this team. All our investing partners,
including our soon to be announced Managing Partner, come from the venture industry.

We have a unique structure, with our investment committee having full independence to make investment decisions unlike any other body in central Government.

That means we will move at the pace founders should expect of their investors.
To put that in context: last week, a company the team has been following for a while told us they had a funding offer on the table. We met them on Monday, had completed diligence by Tuesday, got approval by Wednesday, and took them out for a drink on Thursday - executing at a speed that would make Craig David blush.

And the Sovereign AI team have been exceptional.

In 4 months, we have gone from idea to a fully operational venture fund, introducing the civil service to vibe-coded websites, AI driven CRMs and even a new financial instrument.

It's an astonishing achievement, and a huge effort from Josephine and Will who many of you will know, and Thomas and Konstantin and many others who you won’t.

So thank you.

We have put in these long hours, and we will continue to do so, because we recognise how incredibly important it is for this to succeed.

Because we are venture investors - we know some of these investments will fail.

And we will be held to account for those failures - perhaps by some of the wonderful publications in this room.

But let’s remember that Charles Babbage worked on his Difference Engine for twenty years without ever seeing it succeed. The Prime Minister of the time declared his ‘wooden counting machine’ a delusion.

And yet he went on to spark a computing revolution.

So, the founders we work with should be assured: we know nothing great was ever
achieved without taking risk, and we will take risk with you.

As a venture fund, we will invest capital in return for equity. Explicitly, we will invest an average of five to ten million pounds. With that we expect to back around forty companies within our current mandate.

We will lead rounds, and we will be a follower - investing alongside the best venture firms, here and internationally. We will invest at Seed and Series A – if that means anything anymore - but we will also support a handful of our breakout AI companies at a later stage when a collaboration makes sense.

Now some people have asked, fairly, what good is that amount of capital in a world
where AI labs are raising billions, even tens of billions of pounds?

And I say, we are just the tip of the spear.

When the time comes for our founders to raise those larger rounds, we will introduce them to world-leading investors.

And they will have a hot line to the British Business Bank, who we will work with hand in hand, to unlock hundreds of millions of pounds of further investment into the sector.

But the real reason I believe we can be transformational for founders is what we bring beyond capital.

We can announce today our expanded scheme to provide computing grants of up to one million GPU hours directly to the companies we work with.

We have already allocated over three million GPU hours to six companies, that’s roughly the amount of compute required to train a competitive open-source model like Mistral Large of Deepseek V3. That’s more British AI companies less reliant on foreign technology platforms - running here, domestically, on sovereign cloud built in Bristol, Cambridge and Edinburgh.

We can help founders unlock invaluable data. Which we have already done with our work on OpenBind.

Today we can also announce a further multi-million-pound project to build out essential data through new automated labs, providing AI companies in life sciences more of the infrastructure they need here, on these islands.

We will directly procure solutions on behalf of Government, giving the best teams a guaranteed customer if they can build to a world-class standard.

And we can support you as you build your team. The ability to attract global talent is a competitive weapon. With our unique, rapid and cost free visa offer, we can fast track the best AI researchers and engineers in the world to come and build here, to contribute to Britain’s success.

Sovereign compute. Automated Labs. Unique Datasets. Government as a customer. Cost-free visas. Not promises, support we have already begun to deliver.
Focus

And this is just the start of the support we will bring to the table.

We want to do so much more. But we must do this in a focused way. Despite the
immense support of the Secretary of State we have limited resources, and we must
deploy them to maximum effect in the interests of our investor - the taxpayer.

So that means we will have to make some hard choices, and we will still have to say
‘No’, often.

The companies we back must meet two criteria. First, they must be building a product that could be a critical building block, or GDP moving application of artificial intelligence. And secondly, they must have the potential to be a huge commercial success.

Let me be clear on the latter. We are focused on commercial success.

Not because we hope to make a quick buck for the Chancellor — although I am sure that would be appreciated. But because a company must be commercially successful, to have a meaningful impact.

Our initial areas of focus will be on novel computing and chips; new model
architectures; AI for life sciences; AI for scientific discovery; and agentic security and safety.

Our first investments and compute allocations reflect this focus. With Callosum, we are investing in ways to give AI researchers more resilient and efficient computing. With Prima Mente, we are supporting a team developing models with the NHS to tackle disease. With Doubleword, we hope to help make enterprise adoption of AI dramatically cheaper. With Odyssey, we are working with a team pushing the frontiers of world models and with Cursive we are making generative AI run faster and more securely. We have even backed a new AI lab, but we can’t talk about that yet – because every venture fund needs to have at least one investment ‘in stealth’.

We have set all of this out on our website.

But our focus is not fixed in stone. We will continue to take input from experts within Government and across industry to test and refine our thinking. So, if you are working on other critical areas, such as sovereign supply chains for
robotics, for instance - please stay in touch.

I also want to be clear about what we are not doing, because there is a version of
"sovereign AI" that I think would be a mistake.

Sovereignty is not protectionism. We are not here to replace foreign investment or ban foreign technologies. Countries have tried that in technology before. They did not build sovereign capability. They missed out on global progress.

No — we invest in British companies because we believe they, you, can be the best at what you do globally.

And when you win, you will make Britain indispensable on the global stage, contributing to our wealth and security.

Now - some people don’t believe we should do that.

Some people argue we should be insular in our approach, and modest in our ambitions.

But just look where we are today.

In the first 3 months of 2026, British AI companies raised six billion pounds. That's more than all of continental Europe, and three times what we raised this time last year.

In the last 24 months, every major AI lab in the world has opened or expanded offices in the UK. For 90% of them, we are the first country they expand too.

They join the many world leading frontier labs that were already founded here, like
Eleven Labs in voice, Isomorphic in life sciences or Wayve in embodied AI.

Below the model layer we are seeing exceptional innovation in UK AI infrastructure with companies like Fractile in inference, or Olix in photonics, or Fluidstack and Nscale in compute, all building on our rich heritage with ARM and Imagination Technologies.

Our country, and the world, faces many challenges in making the adoption of AI work for everyone. We are not blind to the disruption such changes will bring.

But ask yourself, in the software revolution of the 1990s, did Britain benefit to the scale we could have?

And in the internet and cloud revolution of the 2000s and 2010s, do you feel we took advantage of all our strengths?

I don’t.

This time, with AI, things must be different. And so we must act differently.

So how will you know if we have succeeded?

Like the founders we back, we have set ourselves some ambitious goals.

We want our portfolio to create over ten billion pounds of enterprise value in Britain. And you will be able to measure our progress on that.

We want to create a positive return for the taxpayer. And you will be able to measure our progress on that.

And perhaps most importantly, we want Britain to be the home of ambition once again. And if anyone has worked out how to measure those vibes, please let us know.

So if that appeals to you, you can find us everywhere online.

We’re Sovereign AI, and we’re open for business.

Thank you.