What the Multiple on an Ordinary Deck Is Really Pricing
A deck lands before nine. Ordinary company, ordinary founder, the sort you actually understand. Then the comparables slide, and the multiple on it is a size too big for the business you have been reading.
You have seen that number before. It belonged to a different company, in a market you would never have funded. You cannot tell yet whether it is the founder's, or the market's.
In the first quarter of this year, seventy-four pence in every venture pound raised in Britain went to a company that calls itself AI. HSBC Innovation Banking and Dealroom counted it.
Almost all of that went to a handful of very large, very late rounds. The seed angel was nowhere near the money.
This week alone, a London startup raised more than sixteen million pounds for an agent that replaces the CV with a short voice chat, while good, dull companies went to the wire for a fraction of it.
And the market that money made is now sitting in your inbox, in a deck that has nothing to do with artificial intelligence.
What travelled was the price.
A valuation is a relative number before it is anything else, built from comparables, from what the last few companies "like this one" were said to be worth.
When one cohort re-rates hard enough, it drags the reference points up for everyone standing near it.
So the software founder in front of you, who has never trained a model and never intends to, opens on a comparables slide a size too big for the business underneath.
The ruler moved before the founder ever picked it up.
Carta's first-quarter figures read the same way from the other side: AI seed rounds priced well clear of everyone else, and more than half of the quarter's venture money chasing them.
The premium was never going to stay where it was minted.
The premium has leaked off the AI decks and settled in everyone else's comparables.
Which leaves you pricing a workaday company off a market you would never write a cheque into.
The way out is to ask what the multiple is actually paying for.
Two questions do most of the sorting.
First: is this an AI company at all, or an AI label on a business that would exist perfectly well without the word?
Cross it off the deck and read what is left. If the company still stands, price the company.
If it falls over, you are being asked to pay an AI price for AI risk, which is the worst trade going this year.
Second, where the technology is real: does the edge survive the morning the model provider ships the same feature by default, at no extra charge?
A great many "AI-native" products are a thin coat of paint on somebody else's model, and the paint is not the moat.
None of which means the premium is never earned.
A real data advantage, or a workflow a customer cannot leave: that carries a genuine number, and pretending otherwise is its own laziness.
Earned is a verdict, and verdicts come after the evidence.
But this quarter the starting assumption is doing far too much of the work.
So when the ordinary deck lands and the multiple looks a size too large, slow the meeting down.
Ask what that number is really pricing, the moat or the label, and whether it survives the day the model becomes a commodity.
The word on the slide came from a market you never joined. The company underneath reads the way it always did.
Since April, a venture capital trust gives its backers 20 percent in income-tax relief, where the enterprise scheme still gives 30, and the company limits on both have doubled.
Watch whether relief-led money now drifts from VCTs toward direct EIS rounds, and whether that steadies seed pricing in the corners the AI premium never reached.
Quantum Motion closed a $160m Series C to rack-mount its quantum chips
The Oxford and UCL spin-out raised about £120m from new backers including DCVC, Kembara and the British Business Bank, per UKTN, to commercialise its silicon-based qubits.
Deep tech still gets funded here, patiently and in size, with the state helping to write the cheque. This is the other market, the one the AI headline is not about.
Happl closed an oversubscribed $11m Series A for its benefits platform
Happl, a London employee-benefits company, raised about £8m led by Portage, with Y Combinator, Haatch and 6 Degrees Capital also in, per UKTN.
Oversubscribed, with Y Combinator and Haatch on the line at Series A. When the angels on a cap table are sharper than the sector, read the names before you read the label on the deck.
Lifted Ventures and the British Business Bank renewed their angel-diversity partnership
The pair extended for another year to grow women angels and regional networks; only about 14 percent of UK angels are women, per UKBAA.
The same British Business Bank backed Quantum Motion's Series C this week: state money on a unicorn's cap table, and state money trying to build the angels beneath it. It underwrites both ends of the ladder, which is either prudent or a tell about how thin the private middle has grown.
When your buyer also leads the round
This week a Cambridge battery firm reached a billion-dollar valuation on a round led by Symbotic, the listed robotics company that is also one of its largest customers. A customer leading your round is validation and a quiet warning at once.
The buyer who most wants the technology now is also best placed to buy it cheaply later, and its seat on the cap table keeps rival acquirers at a polite distance.
The question for any strategic-led round: if this investor never buys the company, who else realistically would?
Each Wednesday we put one question to the room. Here is this week's.
When a non-AI founder prices off AI comps, what's your move?
Subscribers answer this one every Wednesday. Subscribe free →
Ever paid an AI price for what turned out to be a database with a chatbot? Reply and tell us what happened.
Replies reach the editor directly. We read and respond to every reply, whatever the question.
Data21 May 2026 HMRC EIS and SEIS annual statistics, first estimates for 2024-25.
Rates18 Jun 2026 Bank of England Bank Rate decision, currently 3.75 percent.
Tax31 Jul 2026 Self Assessment second payment on account for 2025-26.
FiscalTBC Autumn Budget 2026, date to be confirmed.
The Carry · thecarry.co.ukWorth a screenshot for your diary.
Two prices are loose in one market now, and only one belongs to the company across the table. Ask what the multiple is really buying before it anchors you. Then get back to the founders who priced their own business and can tell you why. They make the week.
Until the next round.
Know an investor who nods along to every "AI-native" on a slide, and has never once asked which model it rents? Share this issue with them.
Share via WhatsApp Email LinkedIn X Copy link