FundingValuations

AI company valuation multiples: figures, dates and sources

AI company valuation multiples: which figures carry a date and source, why private multiples resist verification, and how to cite one defensibly.

, AI Researcher4 min read
In this article6
  1. What multiple is being quoted, and on what
  2. Revenue, ARR and forward multiples distinguished
  3. Where dated figures are published
  4. Why private multiples are rarely verifiable
  5. Using a multiple you can defend
  6. Sources
Sober research-report illustration of three valuation-multiple bases (revenue, ARR, forward revenue) as separate measurement columns, each with a date and source label, on a neutral background.
Sober research-report illustration of three valuation-multiple bases (revenue, ARR, forward revenue) as separate measurement columns, each with a date and source label, on a neutral background.

What multiple is being quoted, and on what

A valuation multiple is a ratio of a company's value to a named base, and the base decides what the number means. For AI companies three bases dominate: trailing revenue, ARR and forward revenue. Quoting a multiple without naming its base, its date and its source is the recurring error on this topic, and it is the first thing a reader checking the figure should establish.

Revenue, ARR and forward multiples distinguished

  • Trailing revenue multiple: company value divided by revenue actually booked in a past period, usually the last twelve months.
  • ARR multiple: value divided by annualised recurring revenue, the contracted subscription revenue of one month multiplied by twelve. ARR has no standard definition, so two companies quoting the same ARR may not be measuring the same thing.
  • Forward revenue multiple: value divided by projected revenue for a future period. A forward multiple is a forecast, not a measurement, and it should be labelled as one.

A private round price is one instrument at one date: it is set by a single lead investor in a negotiated transaction, it applies to preferred shares with terms the headline does not show, and it is not a market price that a second buyer has paid. Treating it as a market-wide multiple imports both the date and the negotiation into a plan without saying so.

The scale of the funding market, by contrast, is well documented. US private AI investment reached $285.9 billion in 2025, more than 23 times the $12.4 billion invested in China, and 1,953 newly funded AI companies were counted in the United States in 2025, figures published in the Stanford HAI 2026 AI Index Report.[1] Those numbers carry a report, a year and a publisher. Most multiples in circulation carry none of the three.

Where dated figures are published

The comparison that matters for defensibility is between where round prices get quoted and the instruments that publish dated, checkable figures. Investor websites are not where multiples are published: a private round price reaches the public through press reports and databases, and a listed company's figures appear in filings a reader can open.

Public instruments behave differently. The European Innovation Council's 2024 work programme opened funding opportunities worth over €1.2 billion, including EIC Accelerator equity investments of €0.5 to €15 million per company.[2] The EIC describes itself on its home page as one of the largest deep tech investors in Europe, offering €6.5 billion of support to startups, research and tech transfer.[3] Bpifrance publishes its activity as the French state-backed investment bank on its own site.[4] Private-market databases such as Crunchbase[5] and Dealroom[6] aggregate round data, and Dealroom's global venture capital guide lists the biggest disclosed rounds with their date, amount and lead investors. What none of these publish is a verified private valuation multiple, which is why a multiple carried into a plan needs its own provenance check. How headline valuations inflate under preference stacks is examined in the Groq funding valuation analysis, and round-size dispersion across the segment in AI infrastructure funding round data.

Why private multiples are rarely verifiable

Most private AI multiples fail an audit for three reasons. The valuation is a round price set by one lead investor, not a transaction a second party has validated. The revenue base is usually self-reported ARR with no standard definition and no audited statement behind it. And neither figure appears in a statutory register or filing a reader can open: private European companies file limited accounts, and a preferred-share round price does not have to be disclosed at all. The verifiable money in the AI buildout sits elsewhere. Epoch AI's open database puts investment spending on AI data centers, hardware and networking at 1.4 percent of US GDP in Q1 2026, up from 0.7 percent, with methods published alongside the data.[7]

Deal structure changes what a headline figure measures, as the acquihire AI team analysis shows for licensing-fee arrangements that avoid an equity transfer altogether.

Using a multiple you can defend

  1. State the unit: which multiple, on which base, in which currency.
  2. State the date: the day the figure was set or published, not the day it was quoted.
  3. State the source page: a URL a reader can open, not a report of a report.
  4. State the instrument: a disclosed round size and lead investor, a filing, or a published grant. If any of the four is missing, describe the development qualitatively rather than quoting a number.

All sources cited in this article were retrieved on 23 September 2026 from the pages linked beside each figure. The article was last verified on 23 September 2026, and its next scheduled review is 23 March 2027.

Sources

  1. hai.stanford.edu
  2. eic.ec.europa.eu
  3. eic.ec.europa.eu
  4. bpifrance.fr
  5. crunchbase.com
  6. dealroom.co
  7. epoch.ai

Rankings and scores on Optinest cannot be bought — no commercial relationship changes how a provider is assessed. How we score

Related reading

All Funding →