AI Has Repriced the Entire Venture Market

The headline everyone’s repeating: AI is distorting the UK venture market, pushing valuations into the stratosphere for anyone with “AI” in their pitch deck.

What the data actually shows: that headline is true for a handful of companies, and misleading for almost everyone else.

Where this myth comes from

Every headline number in H1 2026 points the same direction. Average deal size across the entire UK market hit a record £5.4m, up from a previous peak of £4.5m during the 2021–H1 2022 boom. Three AI companies alone accounted for 29% of all UK equity capital raised this half. Put those two stats side by side and the story writes itself: AI is repricing the market, and everyone in it.

It’s a compelling narrative, and one we’ve spoken about before. But it’s also built almost entirely on averages. And averages, as we’re about to see, are exactly the wrong tool for understanding what’s happening to AI valuations specifically.

In H1 2026, the average pre-money valuation for UK AI companies raising equity hit £70.2m. That’s the figure that ends up in press coverage and LP updates, and it’s the figure fuelling the narrative that AI has repriced the entire market. 

The number that tells the real story

The median pre-money valuation for those same AI companies was just £3.4m, roughly 20 times lower than the average.

When a mean and a median diverge this violently, it’s usually to do with distribution. A small number of very large numbers are dragging the average somewhere the majority of the data never goes.

This isn’t unique to AI as a category. Most sectors with a few outsized rounds show some divergence between mean and median. What makes AI’s gap notable is the scale of it. A 20x difference between average and median isn’t a mild skew, it’s a signal that the average has stopped describing anything real. 

It’s composition, not pricing

The average is being dragged upward by a small number of very large rounds. The same concentration effect is visible elsewhere in H1 2026’s data, where those three AI companies accounted for 29% of all UK equity capital raised this half. 

A handful of mega-deals can move an average dramatically without moving the typical deal at all. Three companies is a small enough number that any one of them exiting the dataset would materially change the average pre-money figure.

Isomorphic’s round landed in May, NScale’s in March, Wayve’s in February, spread across four months rather than clustered in one moment. That pattern holds at the quarterly level too: Q1 saw £4.57b deployed, rising to £5.15b in Q2 for AI companies. 

If the concentration effect were really just one outsized quarter, you’d expect to see it flatten out in the split. Instead, the elevated activity carried across both quarters, which suggests the AI-driven capital concentration is a sustained feature of H1 2026 rather than a one-off moment skewing the half-year average. 

Meanwhile, Seed-stage deals now make up 70% of all AI rounds. Most AI companies raising money in the UK right now are early-stage businesses raising modest rounds, not the frontier labs pulling in nine-figure cheques. The AI label has become the default framing for a huge share of ordinary seed activity, which is exactly why the average gets contaminated by outliers sitting at the opposite end of the market.

So when someone says “AI valuations are through the roof,” the honest answer is: for three companies, yes. For the other 70%+ of the market, not especially.

But the other question is: is this new? When we look at the ten-year average for AI companies, just 48% of funding went to Seed-stage companies, with Venture-stage companies taking up 23%.

Seed’s share of AI funding has risen from 48% over the past decade to 70% today, a jump of over 20 percentage points. So this isn’t a permanent structural feature of how AI gets funded, it’s a real, recent tilt toward the earliest stage. 

Combined with Venture-stage’s share falling from 23% historically, the picture is of a market where AI capital is increasingly chasing new, unproven companies rather than backing them through to later rounds. That’s worth sitting with: the “average AI valuation” isn’t just being distorted by three outliers, it’s being calculated across a seed-heavy pool that looks structurally different from AI funding a few years ago.

The one place AI does carry a premium

This isn’t a “there’s no AI premium at all” story, there is one, just a smaller and more precise one than the headline suggests. Seed-stage AI companies price around 25% higher than the wider seed median. That’s a real, measurable uplift, just nowhere near the 20x gap implied by comparing averages.

AI founders raising seed rounds do get a premium. AI founders being valued like the sector’s biggest outliers do not, unless they are one of those outliers. It’s worth sitting with why this particular premium exists, too. 

A 25% seed uplift is consistent with investors pricing in differentiation or scarcity at the earliest stage, the kind of premium you’d expect for a hot but not yet overheated category. A 20x gap between mean and median, by contrast, isn’t a pricing signal at all; it’s an artefact of a handful of outsized rounds sitting in the same dataset as thousands of ordinary ones.

AI is winning at exits too

If AI is dominating fundraising, it’s natural to assume it’s dominating the next stage as well. It isn’t, at least not yet.

Across the market, exit conversion rises steeply with stage, from 2.8% of Seed rounds to 11.8% of Growth rounds: a reminder that most of the exit story is really a stage story, not a sector story. But despite AI’s outsized and growing share of capital raised, AI companies convert to exit in line with the broader market rather than ahead of it.

Exits are the mechanism by which a valuation premium gets validated,  it’s where the market finds out whether the price investors paid was justified. Right now, AI hasn’t cleared that bar any more convincingly than the rest of the market has. That’s not necessarily a red flag, it’s early data, and exits lag funding by a median of four years, with only 36% of exits happening within three. But it does mean the “AI outperforms” narrative currently rests entirely on the fundraising side of the ledger, not the outcomes side.

In other words: AI is winning the fundraising conversation decisively. It hasn’t yet proven it wins the outcomes conversation, the part investors are actually underwriting for.

See past the average yourself

The gap between a £70.2m average and a £3.4m median isn’t a quirk of this dataset — it’s what happens whenever a market gets summarised into a single number. The only way to catch it is to have access to the underlying deals, not just the headline stat.

That’s the level Beauhurst data operates at. Every figure in this piece (the mean and median pre-money splits, the seed-stage share of AI rounds, the exit conversion rates by stage) comes from deal-by-deal records on UK private companies, not modelled estimates. If you’re benchmarking a raise, screening a deal, or just trying to write about “the AI market” without getting misled by it, that’s the difference between quoting an average and knowing what’s actually behind it.

Explore Beauhurst Insights to dig into the full H1 2026 dataset, or book a demo to see what it looks like with your own benchmarks.