Until last November, Joséphine Kant’s career path followed a textbook trajectory for a high-flying techie: eight years rising through the ranks at Google, followed by a stint investing at Y Combinator in Silicon Valley before helping launch US-based early-stage fund, Dogwood Ventures, in 2024.
“Being at YC, when you’re surrounded by the most obsessed people in the world, it gets really hard not to want to start your own thing,” Kant tells me over a video call.
But today she isn’t dialling in from the other side of the Atlantic. Instead, she’s calling from a new office in Whitehall — the heart of the UK government — where, seven weeks into her role as head of venture at the newly created Sovereign AI Unit, she is helping steer a £500m initiative tasked with building and scaling Britain’s AI capabilities.
“I had never considered any career in government,” Kant says, adding that when she first saw the role advertised it looked “impossibly hard to get right.”
So, how did a job in public office manage to draw this hotshot venture capitalist to the UK?
After being offered the role, Kant consulted a dozen or so people for advice. “Overwhelmingly most of them said: ‘you would be absolutely crazy to take this — how would the government ever be able to operate like a private VC and have a meaningful contribution?’” she says. “So I accepted.”
“I think if lots of people are telling you something is extremely crazy, that means you’re on to something interesting,” she chuckles.
Inside the Sovereign AI Unit
The Sovereign AI Unit was announced last year as part of the AI Opportunities Action Plan authored by Entrepreneurs First’s Matt Clifford in his role as AI advisor to the prime minister. The unit was established, and backed by up to £500m of public money, to invest in early-stage startups across four years.
Deployment will begin in April 2026, with the fund writing cheques of up to £10m to UK-based companies working in “priority areas,” Kant says, adding the unit’s “really focused” on domains where the UK has “structural advantages,” such as biotech, cyber, fintech and quantum.
“Britain is a small country,” she says. “We need to pick a few things that we’re really excellent at, and execute on those really well [...] That’s how small countries win ultimately: they are world-class at a few things, which then compound.”
But why would a top-tier startup select the UK government as an investor in the first place? “We are actually going to be in a really fortunate position,” Kant says. “We can offer founders something that private VC cannot always offer: access to the government.”
That access could translate into practical advantages, she argues — from opening doors to public procurement opportunities and helping companies navigate regulatory processes, to providing compute resources and supporting talent initiatives.
“Ultimately our hope is to earn the right to invest in the very best companies and [...] to anchor them to the UK and help them navigate the crazy world of the government.”
“I have been really pleasantly surprised by how the ecosystem has really welcomed us,” Kant says. “I think a lot of people are realising that the government is in some way a platform for companies. The government isn’t really just about regulating AI anymore — it can actually be a buyer and a customer.”
Recent policy moves confirm that shift. In November, for example, the UK government pledged to reform the public procurement system to make it easier for AI companies to win public-sector contracts.
We're in a unique moment, Kant says, where the tech world is building the AI that will come to underpin industries going forward, be it healthcare or defence. The window where that underpinning architecture is being decided is narrow, she adds.
But the unit needs to do more than being a gateway to further government support, according to Kant: “We need to offer the same terms [that] private investors can.
“We are here to invest like private VC and we have to do that by mimicking the speed and the processes of private venture capital, and also act with the discipline of the taxpayer, who ultimately should be rewarded for the risks that we’re able to take.”
How the government can attract tech talent
For this government venture arm to operate like a private VC, its team needs to hire like one, Kant says. “We can make the government more agile by bringing in more people that can and have built things,” adding she “didn’t come into the government to make announcements, but to build.”
This is increasingly the mantra of the UK’s technology department. The two AI advisors to the UK government, formerly Clifford and now Jade Leung, come from venture and Big Tech respectively, while the Sovereign AI Unit is chaired by Balderton partner James Wise, who, Kant says, “has been incredibly instrumental in helping us get set up.” The UK’s AI minister, Kanishka Narayan, also worked in venture before entering politics.
“Kanishka gets it, he moves really fast and talks our language,” Kant says, adding she “could not have dreamt up a better minister than someone who worked at a VC and understands AI.”
Despite bringing in that kind of talent, the UK government has at times faced criticism for offering salaries that fall short of what candidates could earn in comparable roles across the wider tech ecosystem — including for Kant’s own role, which was advertised with a salary of up to £80k.
Kant believes the government would receive criticism either way, regardless of how much pay it offers. “If salaries are low, the argument is that we won’t attract top talent [...] If they’re too high, the argument is that we’re overpaying with public money.”
The government shouldn’t be looking to outbid the compensation packages offered by frontier AI companies, Kant says. Instead, she thinks it should be advertising itself as a place for talented people to work on nationally consequential and high impact projects.
Ultimately, she says, the unit hopes to attract “people who do not just criticise, but [are willing to] do a tour of duty, and bring private sector speed and agility into the public sector.”
Why representation matters
There is one aspect of private venture capital Kant does not want to recreate: a set of portfolio founders and investors who do not represent the UK population at large.
With only 12% of Europe’s VC funding going to women-founded businesses in 2024, Kant believes describing funding inequity as “a pipeline issue” is “a cop-out answer.”
“It’s not actually a pipeline problem, it’s a compounding disadvantage problem: women face many more barriers at many stages of the journey, from education to hiring to funding to promotion to decision-making.”
If the UK wants to get serious about growth and building competitive AI companies, then representation isn’t a ‘nice to have’, Kant says, it’s an economic imperative.
“If we want to build these sovereign AI capabilities, they have to represent and benefit the whole country.”
In order to do this, Kant wants a diverse set of investors writing the unit’s cheques.“Representation in decision-making is really, really important because it actually changes outcomes,” she says.
Entering government without knowing a single person in it, Kant admits the adjustment was steep. “There are a million acronyms!” she laughs. But she now sees an opportunity to make the machinery of government more transparent for founders and venture folk.
“I actually really don’t like social media [...] It’s very against my core, but I’m doing it because I really want people to see what we’re building — because it’s pretty f**cking cool!”
“We can bring the speed of venture and the might of a nation to the very best and brightest of Britain. That is such a privilege, and I can’t not share that with anyone who’ll listen to me.”
