The Class of 2023
What the official register really says about Britain's first AI-era startup generation.
894,494 companies were incorporated in the UK in 2023, the first full year after ChatGPT. Tero read the complete Companies House record of all of them: every profile, officer and capital filing. Every headline number is verified back to source documents. Nine findings follow, each backed by a chart; the methodology and its numbered assumptions are at the end, and any cut can be re-run under your own assumptions through the data edition.
- 1One company in 534 becomes venture-financed. 894,494 companies incorporated in 2023 boil down to a venture universe of 1,674, and 39% of it has already raised more than once.
- 2The verified value is £12.7bn, and 107 companies hold more than half of it. The median company prices at £2.5m; the 6% priced above £25m hold 55% of the value.
- 3Fintech prices 2.7x the class median and has the best funding odds. £6.8m median valuation, 56% raise a second round and 25% a third; its 1.00x valuation step-up rests on just 14 verified pairs.
- 4AI is the complete package: marked up 1.83x with only 10% down rounds. £3.8m median valuation and 47% second rounds make it the only sector strong on every dial.
- 5Food & beverage and health/bio pay the most per unit of risk: 0.43 and 0.26. Media is the worst trade, carrying the widest spread (8.5x) for a flat 1.01x valuation step-up.
- 6Under-30 founders mark up 1.63x, against 1.00x for founders 45 and older. Entry price barely moves with age (£1.6m to £2.6m medians); what happens next moves a lot.
- 7Second-round odds are 44% in London and 33% outside the hubs. Valuations are similar everywhere; continuation is the register's real geography effect.
- 897% of the venture universe discloses no revenue at all. Implied revenue multiples exist for just 63 companies and run from 0.3x to 13.9x.
- 9Exactly two nine-figure startups survive document verification. 28 companies claiming £45.8bn between them have no external evidence and are excluded from every number.
There are 1,674 venture-financed companies from the class of 2023, and 39% have already raised more than once.
Number of companies at each stage, from incorporation to venture financing. Each bar zooms into the shaded slice of the one before it.
- ▪One in 534 becomes venture-financed: of 894,494 companies incorporated in 2023, 30,675 (3.4%) filed a share allotment and 1,674 of those (5.5%) priced a round at market terms.
- ▪39% have already raised again: 651 companies raised a second round and 233 (14%) a third.
- ▪1,023 companies (61%) have raised exactly once so far; their next filing decides whether they compound or stall.
Venture-style: at least £25,000 raised at a market price (a premium over nominal value, or a venture share class), implying at least £250,000 of company value. Excluded: holding and investment vehicles, private-equity deal entities and infrastructure project SPVs, 366 entities in all (assumptions A3, A9). At-par allotments are founders or parents moving capital, not investors pricing a business.
For investors: the 1,674 are the real universe and the real odds. One company in 534 raises venture-style at all; once inside the universe, two in five raise again.
The verified value is £12.7bn, and 107 companies hold more than half of it.
Number of companies in each implied-valuation band (bars), with the average valuation inside each band (line), and each band's share of companies and of total value below.
- ▪The median company prices at £2.5m, the average at £7.7m: the 3x gap between the two is concentration, not typical size.
- ▪50% of the class holds 16% of the value. Half sits in the £1m to £5m band; the 6% priced above £25m hold 55%.
- ▪Average valuation climbs 248x across the bands, from £578k at the bottom to £143.6m in the £100m to £250m band, where 19 companies sit.
Implied post-money = latest market-round share price × all shares in issue after it; one number per company; statistics on the verified range (≤£250m plus independently verified companies: 1,648 of 1,674). Twenty-eight companies claim a further £45.8bn that no external evidence supports; they are excluded everywhere and dissected at the end of the report.
For investors: half the class prices in the £1m–£5m band, holding 16% of the value; the value you are hunting sits in a group of ~107 companies. Steep concentration is the mathematical case for breadth and for follow-on reserves.
AI is the complete package. Fintech has the odds but no marks. Health/bio is the safe lane.
Median valuation, funding odds, verified valuation step-ups and down rounds by sector, ranked by median valuation. One chart backs findings 3 and 4. Values marked * rest on fewer than 15 verified pairs.
- ▪£6.8m median valuation against £2.5m for the class, the highest of any sector with 30 or more companies.
- ▪56% reach a second round and 25% a third, against class averages of 39% and 14%.
- ▪The valuation step-up is 1.00x* on 14 verified pairs, with 29%* down rounds: the register has not yet paid the price it charges.
- ▪1.83x median verified valuation step-up on 42 pairs, the highest of any sector above the 15-pair bar; the class-wide base rate is 1.17x.
- ▪10% down rounds against 14% for the class: when AI re-prices, it usually re-prices upward.
- ▪£3.8m median valuation, 47% second rounds, 16% third rounds: priced and funded above the class average across the board.
Median valuation. the middle company's implied post-money within the sector, on the verified range
Reached 2nd / 3rd round. share of the sector's companies that raised again at a market price, rounds separated by more than 90 days
Verified valuation step-up. the multiple on your stake if you buy at a company's first round and mark at its latest verified round price; the middle company's multiple is shown
Down rounds. share of verified re-pricing events that came in below the earlier price, counted by number of events, not weighted by value
Sectors ranked by median valuation. Verified-range statistics; across all sectors the median verified valuation step-up is 1.17x and 14% of re-raisings price down. Valuation step-ups and down-round rates marked * rest on fewer than 15 verified pairs after the >10x quarantine (fintech 14, hospitality 10, property 9, energy 4) and are indicative only. Energy's pairs exclude one gold-mining project vehicle whose 8.5x re-pricing was natural-resources project financing, not venture pricing.
For investors: read it as two dials: odds (2nd and 3rd round) and payoff (valuation step-up, down rounds). AI is the closest this vintage comes to strong on both; the priced-highest sector is strong on exactly one.
Risk and return by sector (a Sharpe-ratio equivalent): food & beverage and health/bio pay the most per unit of risk.
Return (vertical: median verified valuation step-up, the latest round price divided by the first round price) against risk (horizontal: spread of outcomes) for every sector in findings 3–4; bubble size is average valuation, colour is months between rounds. Dashed bubbles and * mark sectors below 15 verified pairs.
- ▪Food & beverage pays 0.43 per unit of risk (2.00x on a 4.7x spread), the best ratio in the class; health/bio is the steady version at 0.26 on the narrowest spread (4.4x).
- ▪AI pays the most in absolute terms (1.83x) on a wide 7.7x spread, so its ratio (0.24) sits mid-table: high risk, well paid.
- ▪Media/creative is the worst trade: an 8.5x spread for a 1.01x valuation step-up (0.12), the widest outcomes on the register with no measured reward.
- ▪Fintech sits on the flat line at 1.00x* on a 6.6x spread (0.15*): the sector priced highest with the best odds has not yet re-priced upward on 14 verified pairs, which is why it is not near the top of this ranking.
- ▪Energy is quietly attractive: 1.51x* on a 4.8x spread puts its ratio (0.31*) second in the class. It rests on 4 verified pairs; verification removed a gold-mining project vehicle whose 8.5x was not venture pricing.
x: valuation spread (the 90th-percentile company's value divided by the median company's; robust where variance and standard deviation are owned by single outliers). y: median verified valuation step-up, the multiple between a company's first and latest verified round price. Valuation step-up divided by spread is the Sharpe-style ratio, reward per unit of outcome dispersion. It rhymes with Sharpe rather than replicating it: the denominator is cross-company spread, not return volatility. All findings 3–4 sectors shown; * marks valuation step-ups resting on fewer than 15 verified pairs.
For investors: above the diagonal, dispersion is paid for (food & bev, health/bio, AI); below it, you carry outcome risk without measured reward (media). Colour is the funding tempo: darker sectors make you wait longer between rounds.
Founder age doesn't move the entry price. It moves what happens next, and the under-30s own the mark-ups.
Funding odds, median valuation, verified valuation step-up and months between rounds, by the age of the youngest founder at incorporation.
- ▪Entry prices sit within £1.0m of each other across every age bracket (£1.6m to £2.6m medians): age does not price the ticket.
- ▪Under-30 founders mark up 1.63x, against 1.01x for 45 to 54 and 1.00x for 55+: the strongest age effect anywhere on the register.
- ▪The 30 to 34 bracket has the best second-round odds (42%); the 55+ bracket re-raises fastest (8.0 months) but does not get re-priced upward.
Youngest founder on the ticket: directors appointed within 60 days of incorporation with birth data on the register. Verified valuation step-ups only; the under-30 bracket's 1.63x rests on a full 303-company cohort.
For investors: if your sourcing filters for young teams you are screening out the modal funded founder (43, plausibly solo). If you are underwriting mark-ups, the under-30 premium is the strongest age effect on the register.
Prices are similar everywhere. The odds of a second round are not: 44% in London, 33% outside the hubs.
Company counts, verified £25m+ companies, median valuation and second-round odds, by registered location.
- ▪London holds 58 of the 107 verified £25m+ companies on 768 companies, against 773 spread across the rest of the UK.
- ▪Cambridge prices highest of any location (£5.1m median), with London at £3.1m and the rest of the UK at £1.9m.
- ▪Second-round odds fall from 50% (Cambridge, Edinburgh) through 44% (London) to 33% outside the hubs: the geography effect is continuation, not price.
Registered locations; hub cities separated, everything else pooled. Cambridge's £5.1m median is the highest of any location; verified valuation step-ups by city are withheld below 15 pairs.
For investors: sourcing outside the hubs is not a pricing edge; entry is barely cheaper. It is a continuation risk to underwrite: the register's geography effect is whether the second round happens.
Revenue multiples barely exist: 97% of the venture universe discloses no revenue at all.
Median implied valuation divided by disclosed revenue, by sector, for the companies where both numbers are public. Dashed bars are two-company samples.
- ▪Only 63 of 1,674 companies (3.7%) disclose revenue; UK small-company filing exemptions keep every other P&L private.
- ▪Multiples run from 0.3x (hardware) to 13.9x (e-commerce) where both numbers exist, with software at 10.7x.
- ▪Fintech's 112.5x and AI's 36.7x rest on two companies each; they are anecdotes, not benchmarks.
Median implied post-money divided by disclosed revenue, for the 63 companies where both numbers are public. Dashed bars are two-company samples, printed as anecdotes for that reason alone. UK small-company filing exemptions keep every other P&L private.
For investors: revenue multiples quoted for UK startups are built on a sliver of self-selected disclosers. Treat any typical-multiple claim, including ours, as directional at best.
There are exactly two verified nine-figure startups in the class. We deleted every finding that failed its documents.
- ▪Every company filing £100m+ was re-read at document level: most claims were parser errors, subsidiaries capitalised at par, or self-filed values with no external footprint.
- ▪Two companies survive: C2X (~£239m) and Sedex Information Exchange (~£145m), each with independently evidenced institutional backing ($100m from ENEOS and A.P. Moller; investment from LDC).
- ▪28 companies claim £45.8bn between them that no external evidence supports; they are excluded from every number in this report.
Companies House records what companies declare and verifies none of it. That is why this report treats the register as testimony, not truth: every value statistic runs on the verified range, and anything extraordinary had to survive its own paperwork. The billion-pound entities of 2023 are private-equity deal vehicles, not startups, and sit outside every number here.
The same rule applies in both directions: nothing was added back on instinct either. If a company's documents could not carry its claim, the claim went, however good the story.
Verification deleted our two most spectacular findings: an energy lottery that traced to infrastructure project vehicles and parser misreads, and a hardware dispersion trap built on fantasy filings. 366 vehicles were excluded, 158 extreme claims quarantined, and the universe shrank with each pass. Every number that remains has cleared the bar that killed them.
Sources: the complete Companies House record of the 2023 cohort: 894,494 profiles, 102,787 officer records, 135,710 capital filings, 77,293 parsed allotments, 43,731 first accounts. Filter chain: arithmetic QC with reality caps, strict 2023 incorporation, vehicle and project-SPV exclusion, the premium-over-nominal rule, the venture floor, verified-range value statistics, and document-level re-reading of everything at £100m+. The full ten-assumption register, the open-sourced QC code, and query access to re-cut any table under your own assumptions ship with the data edition. Figures dated August 2026; the class keeps filing.
© 2026 TRCPL Ltd, trading as Tero. Compiled from public registers at the dates shown; register name matches are not identity confirmations; verify against the live registers before acting. Tero is software: we are not a fund, we do not hold client money, and we do not provide investment advice or personal recommendations. Information, not investment advice. Capital at risk.