Pricing the next cycle: valuations and exits in the age of AI

A third of all the venture capital raised in the UK last year went to AI startups. That single fact is why the old rules for pricing early-stage companies no longer hold, and why capital comes back out differently too.

In 2025, UK startups raised $23.6 billion in venture capital, up 35% on the year before and the first annual growth in four years [1]. AI accounted for a record $7.9 billion, around a third of the total, up from a quarter the year before [2].

This is impacting the benchmarks that investors have previously used to value startups. At an Accelerate UK panel talk on valuations and exits that I attended earlier this year, the discussion around AI valuations and exits highlighted that nobody has fully settled on how this should be tackled.

The split isn’t AI versus everything else

The easy framing is winners and losers: companies with AI and companies without. The investors discussing this described something more subtle. It isn’t that every business needs to be an AI company. But every business now needs an AI element, even if that simply means using best-in-class tools to serve customers better or more cheaply. The companies struggling to raise aren’t the ones without proprietary models. They’re the ones with no credible answer to the question at all.

That can put founders in an awkward spot. An AI story has become a baseline expectation for most, which means it no longer differentiates anyone. Investors have started discounting the narrative and looking harder at what sits behind it.

The old benchmarks have gone

For years, early-stage software was priced against a familiar rule of thumb: roughly two to three times revenue growth year on year, with a path to breakeven in sight. Investors could look at a growth curve and infer most of what they needed to know: the size of the opportunity, how efficiently the company was capturing it, and roughly when it would fund itself. AI has snapped that frame.

The clearest example is ElevenLabs, the voice AI company, which crossed $330 million in annual recurring revenue by the end of 2025 and roughly $500 million by April 2026, having been founded only in 2022 [3]. Its February 2026 round valued it at $11 billion [4]. Growth at that speed makes the old multiples almost meaningless. When a company can go from nothing to nine figures of revenue inside a couple of years, the growth curve stops being informative on its own: it says a company found a market fast, not whether it keeps it. The question investors actually need answered stops being “how fast is it growing” and becomes “how long will the growth last.”

That second question is where pricing now lives, and it’s much harder to answer.

Why defensibility matters more, not less

If anyone can build a product quickly with AI, then building the product is no longer the moat. The investors on the panel kept returning to four things that still separate a durable company from a fast-rising one.

• Proprietary or verticalised data. The kind a competitor can’t simply scrape or regenerate.
• Distribution. Because if product is cheap to build, getting it in front of customers is where the real advantage now sits.
• Trust and brand. The reputation that gets a company the benefit of the doubt with customers and investors when everyone’s making similar AI claims.
• Quality of revenue. Low churn and strong net revenue retention. This is the one they flagged as most often missing, and it’s usually what separates the companies that go up fast and stay up from the ones that go up fast and come back down. Impressive top-line growth can mask revenue that doesn’t stick.

For founders, the implication is direct. An AI narrative gets the meeting. Quality of revenue gets the term sheet.

The exit maths has changed too

Pricing a company at entry only makes sense against some view of how capital comes back out, and for years that view assumed the initial public offering (IPO) window would eventually reopen. That picture has shifted as sharply as valuations.

The dominant route is now acquisition, not flotation. Across Europe, M&A has become the overwhelming majority of startup exits while IPOs have fallen to a decade low, and the UK public market in particular has been largely shut to growth companies [5]. With the front door to liquidity narrowed, secondaries have grown as a release valve, letting earlier investors and founders take some money off the table while a company stays private for longer.

That matters for valuations because it compresses the range of plausible outcomes. Fewer blockbuster IPOs means investors are pricing more cautiously against a likely trade sale, even as the hottest AI rounds price as if a public listing is inevitable. Those two beliefs can’t both be right for long, and the gap between them is where a lot of the current pricing tension sits: a company can look expensive against a trade-sale outcome and reasonable against an IPO outcome, with no way of knowing in advance which one it will actually get.

The capital behind the cycle is a UK problem

There’s a structural backdrop the UK still hasn’t solved. Domestic pension money barely reaches the asset class. On average, UK defined contribution schemes hold well under 1% in unlisted companies, and even the largest schemes allocate only around 1.4% to growth capital, against roughly 13% for comparable US funds [6]. A retired teacher in the US is, in effect, far more invested in the next generation of companies than one in Britain.

Public capital is trying to close some of the gap. The British Business Bank has set out a plan to unlock around £26 billion of private capital alongside £13 billion of its own funding over five years, including the ability to write cheques of £100 million or more into growth-stage funds [7]. The government’s £500 million Sovereign AI Fund, launched in April 2026, is aimed squarely at keeping homegrown AI companies building in the UK [8]. The open question is whether that’s enough to stop the best companies pricing, raising, and eventually exiting elsewhere.

What it means for founders and the people backing them

For founders, the practical takeaway is to hold valuation expectations loosely. A number set in a frothy AI round is a liability if the next round has to clear it on fundamentals. Building genuine defensibility, particularly revenue that stays, is worth more than a headline valuation that the market may reprice.

For investors, the discipline is the same one it always was, just harder to apply in a fast market: separate the narrative from the durable business underneath it. The funds that price well through this cycle will be the ones putting capital behind companies whose growth has something solid under it, not just the ones with the cleanest AI story.

Finding investor fit means putting founders in front of the investors whose thesis genuinely suits what they’re building, so both sides spend less time pricing a story and more time on the business behind it.

References

1. 2025 UK Innovation Review (January 2026) | HSBC Innovation Banking / Dealroom — https://www.hsbcinnovationbanking.com/gb/en/resources/2025-uk-innovation-review
2. UK innovation – 2025 review (15 January 2026), AI investment of $7.9bn, up 80% on 2024 | Dealroom / HSBC Innovation Banking — https://dealroom.co/reports/uk-innovation-2025-review
3. ElevenLabs crossed $330M ARR in 2025, ~$500M by April 2026 | TechCrunch — https://techcrunch.com/2026/01/13/elevenlabs-ceo-says-the-voice-ai-startup-crossed-330-million-arr-last-year/
4. ElevenLabs raises $500M Series D at $11B valuation | ElevenLabs — https://elevenlabs.io/blog/series-d
5. M&A dominates EMEA startup exits as IPOs hit decade low | J.P. Morgan — https://www.jpmorgan.com/insights/business-planning/m-a-dominates-emea-startup-exits-as-ipos-hit-decade-low
6. UK plans to unlock pension capital for private markets; UK DC vs US private-asset allocation | PitchBook / GOV.UK — https://pitchbook.com/news/articles/uk-pension-funds-private-market-allocations
7. British Business Bank five-year plan to unlock ~£26bn of private capital | British Business Bank — https://www.british-business-bank.co.uk/news-and-events/news/british-business-bank-publishes-five-year-strategic-plan-deliver-step-change-smaller-business
8. UK government launches £500m Sovereign AI Fund | GOV.UK — https://www.gov.uk/government/news/ai-firms-pioneering-drug-discovery-cheaper-supercomputing-and-more-get-first-backing-through-uks-sovereign-ai