Balderton Capital has backed "something like 30 or 40 unicorns," and Suranga Chandratillake says the reason Britain does not produce more of them has nothing to do with how much the government is willing to invest directly.
The Chancellor wants to double the number of British unicorns, and one of the ideas on the table is for the state to invest in companies to create them. Chandratillake's answer is that this has been tried and it does not work, and that the binding constraint is somewhere else entirely: UK pension funds measure themselves against other UK pension funds.
"A lot of pension funds continue to track themselves against other local pension funds. So they're looking really at sort of a UK league table. And I think that's a huge mistake."
Chandratillake backed Wayve before 2020 — the autonomous-driving company that, a week or two before this interview, put robotaxis on London streets with Uber — which is the basis for his view on writing safety into an AI product at the seed stage rather than after it ships.
The full interview is covered here so you can skip it.
Here are the 8 arguments that matter.
👤 Guest: Suranga Chandratillake, General Partner at Balderton Capital, a venture firm whose European investments include Revolut and Wayve
🎙️ Hosts: Caroline Hepker and Stephen Carroll, presenters of Bloomberg Daybreak Europe
📰 Published: 14 September 2026 on YouTube (Bloomberg Podcasts)
🔴 YouTube | 🟣 Apple Podcasts | ⏱️ 10 min
Key Takeaways
The people with the most to gain from racing ahead in AI are the ones saying slow down
The AI slowdown will not change daily life in Britain, because AI has not changed it yet
"Not even in Silicon Valley, really, let alone the rest of the world"
Safety has to be engineered in at the seed stage, and partly by choosing the founder
Wayve was an AI company driving "a one or two ton piece of metal" from the start
The exit fee founders are supposedly worried about is barely discussed by founders
Governments have tried investing directly to create unicorns and it has not worked
What works, he says, is funding more venture firms so they compete with each other
The two fixes he wants are pension capital and procurement, in that order
British procurement rules, he says, favor large US technology companies over local ones
UK pension funds have underperformed because they took too little risk, not too much
1. The AI Slowdown Story
The programme opened on the morning's big story — American AI companies talking about slowing the pace of frontier development — and asked Chandratillake what he made of it.
He read the source of the warning as the point: "I mean, it's a huge story and it's really interesting to see, obviously, those who have most really to gain by running as fast as possible in this area are themselves nervous about the progress that's being made."
His framing of the problem is coordination, not conviction. "The key in this sort of thing always will be how can you actually compel all the different players with all the different incentives to actually work together?"
He was not confident about the odds: "Historically, humanity hasn't always done a great job of that, but we have at times. So let's see how that unfolds."
The host noted that China has called the slowdown warnings fear mongering, and asked what embedding oversight inside big US AI firms would mean for the UK and Europe.
2. Little Direct Impact Yet
Chandratillake's answer on the practical effect for British and European startups was that there is not much of one, and that the timing is the good news.
"Yeah, I don't think it will have a huge direct impact on people today because the reality is that, for all the staggering technical gains that AI is making, we haven't seen it completely change the nature of our economy," he said. "Not even in Silicon Valley, really, let alone the rest of the world."
That, in his view, is why now is the moment to set the rules — before the technology reshapes how people live and work.
On the specific proposal: "I think, the idea of including embedded sort of advisors who can monitor what the labs are doing is an interesting one."
The reason he thinks it is still hard is capability plus coordination: he said "we're building machines that will be able to, work certainly as effectively as us and, of course, much, much more rapidly than us and coordinate in groups, which is some of the stuff that we've seen over the last six months or so, which has really panicked people."
He does not read the failures as malice: "It's not that anyone was intent on evil or anything like that, but more just that, sometimes you set a certain set of incentives and rules and they can be interpreted in a way that ends up being negative." "And that's a really difficult problem to solve for."
3. Wayve and Safety by Design
Asked whether he now questions portfolio companies about their own rules for applying AI, Chandratillake said the firm has been living with the question for years, and gave the case that taught it.
"So, we were one of the earliest investors in the autonomous vehicle company Wayve, which just a week or two ago launched autonomous taxis with Uber across London," he said, noting Balderton invested before 2020.
The stakes were legible from the beginning: "And even then, it was obvious that this was an AI company that could have huge implications because it's AI that drives a one or two ton piece of metal around our streets."
"So it's critical to engineer things like safety and ethics and behavior into the heart of these systems," he said — and then immediately conceded the limit: "But knowing that doesn't make it easy."
His second answer is not technical at all. "It's also about, which entrepreneurs do you back?" — how they think about the world, and about the risks as well as the benefits of what they are building.
4. The Exit Fee Is a Non-Issue
The programme raised a British policy idea that has been floated but not adopted: an exit fee on founders, driven by a government trying to keep the economic gains of scale-ups at home.
"Yeah, it hasn't been a huge topic, I think, in startup world, if I'm really honest," Chandratillake said.
The reason is where founders' attention actually sits. "I think the majority of people starting a new company are focused on the much more sort of initial challenges of hiring the right people, getting the first customers, etc." "They aren't necessarily worrying about how and when they're going to exit and whether they're going to pay tax at that point."
Where it does register is with people who have already sold, who can see what such a rule would have cost them. He allowed that it would have an effect eventually, but said it is not a live problem today.
The problems he says are live are more mundane: funding for the right companies, and access to talent locally and globally. On Britain's standing: "I think we're probably number three globally in this arena behind the US and China, both of which are obviously much larger than us, so difficult to compete with on a fundamental level."
5. Doubling the Unicorn Count
Asked directly whether any Balderton company will reach unicorn status this year, Chandratillake said yes and then argued the target itself is aimed at the wrong lever.
"I'm sure some will," he said. "We've been fortunate to have had something like 30 or 40 unicorns over the period that we've been active, and a number of them are decacorns and centacorns as well."
His own portfolio is not the issue: "So I'm not at all concerned about our own portfolio. But again, this is about doing it across the entire country, and that's what we need to see."
The objection to the policy is empirical: "I think the challenge with part of what the chancellor has suggested is that the government would directly invest to try and create these unicorns. History shows that hasn't generally worked."
What he would do instead is fund the funders: "What we could do and what I think we should do is stimulate yet more venture capital locally in this ecosystem." He wants "many, many firms competing with people like me," on the theory that the pool gets the calls right even where an individual firm does not.
6. Unlock the Pension Money
Asked what the government needs to say at the October budget to convince people like him it is serious about growing British businesses, Chandratillake named two things and put money first.
Deep tech needs a lot of capital early, he said, because the technology has to be built before there is anything to sell. "You've seen that in the US with some of the AI labs, which spent billions of dollars before they turned even a dollar of revenue or turnover."
"And the best way for us to do that is to unlock pension capital," he said.
The comparison is direct: "In the US, pension funds are massively invested in venture capital and early stage technology. Here in the UK, we have almost the opposite trend and we need to be able to do that."
7. Procurement Favors US Firms
His second ask was about who buys, not who funds — and it is the one with a named target.
"The second thing, I think, is really all about purchasing power," he said. "Again, in the US, one of the key trends you see is that both government and corporations are happy to take a risk and take a bet on local technology companies. Here in the UK, we see a lot less of that."
"Procurement rules are backward and often actually favor large US technology companies rather than local UK technology companies. And we need to fix that," he said.
The companies losing out are real ones: "There are brilliant companies here who are building software that could compete with the likes of Oracle or Microsoft or Palantir or many others."
"And we need to give those companies the oxygen that comes with sales and contracts," he said, arguing that government buying British would set the example for British corporates.
8. The UK League Table Trap
The host noted that pension money had already come up with an earlier guest that morning, on public-private partnerships, and asked how a venture firm convinces pension funds its investments are safe enough for savers.
Chandratillake's answer was to change the comparison set. "A lot of pension funds continue to track themselves against other local pension funds. So they're looking really at sort of a UK league table. And I think that's a huge mistake."
The benchmarks he would use are the large US state funds — he named the ones that run teachers' pensions in California — plus the Norwegian sovereign wealth fund and Australian funds. He was explicit that he is not comparing against higher-risk private vehicles.
On the result of that comparison, he said "you will see that we've fallen behind and we've fallen behind because we haven't taken the right kinds of risk."
His model of a pension book is a barbell rather than a safe default: "The running of these funds requires a wide variety of investments, some which will be incredibly safe." Others should take higher risk and deliver higher gain, he said.
"And the sooner we do that, the better it will be for the innovation economy, but also, frankly, for the pensioners themselves," he said.
Bonus Insights
The programme's own framing set up the interview: Britain is home to three of the world's top 10 universities, spin-outs from the Cambridge, Oxford and London triangle have risen notably, and yet, the host said, experts point to a persistent technical literacy gap and to large mature spin-outs increasingly leaving Britain altogether.
Chandratillake was speaking from Cambridge Tech Week, and said the city is the proof that it can work: "Cambridge is doing incredibly well in this area, but we need that kind of ecosystem to hopefully develop in other parts of the country as well."
He also flagged the ceiling on the Europe comparison when it comes to any of this: the four freedoms and the scale of the US and China are structural, and Britain's answer has to be to compete on funding and on purchasing rather than on size.
Chandratillake's bottom line is that Britain's startup problem is a capital-allocation problem wearing a policy costume: the state cannot buy its way to more unicorns, but it can change what pension funds are measured against and what the government itself buys, and those two levers are the ones he would pull.
Products, Companies & Tools Mentioned
Balderton Capital (Chandratillake's firm; he says it has had something like 30 or 40 unicorns, with several decacorns and centacorns among them)
Wayve and Uber (Balderton invested in the autonomous-driving company before 2020; he says it launched autonomous taxis with Uber across London a week or two before the interview)
Oracle, Microsoft and Palantir (Named as the large US vendors he says British procurement rules favor over local companies that could compete with them)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: