Scott Stevenson, CEO of legal AI startup Spellbook, accused AI startups last month of systematically inflating their annual recurring revenue figures by reporting contracted ARR as actual ARR, a practice he says is enabled by major venture firms seeking to crown category winners. His X post drew over 200 reshares and comments from investors and founders, exposing what multiple sources confirm is a pervasive issue across the AI startup ecosystem.
“The reason many AI startups are crushing revenue records is because they are using a dishonest metric. The biggest funds in the world are supporting this and misleading journalists for PR coverage.”— Scott Stevenson, Spellbook CEO
The core tactic is substituting contracted ARR, sometimes called committed ARR or CARR, for traditional ARR in public announcements. CARR counts revenue from signed customers that have not yet been onboarded or deployed, meaning the startup has not collected the money and may never collect it if implementation fails or the customer cancels during a trial period. One investor told TechCrunch that CARR can run 70% higher than actual ARR at some companies.
At least one high-profile enterprise AI startup reported surpassing $100 million in ARR when only a fraction of that revenue came from currently paying customers, according to investors familiar with the company. The rest was from contracts that had not been deployed and in some cases would take months or years to implement. Another startup marketed $50 million in ARR while its actual figure was $42 million, a gap of $8 million that investors had access to in the company's books.
Key facts
- 01Some AI startups report contracted ARR (CARR) that is 70% higher than actual annual recurring revenue from deployed customers.
- 02At least one enterprise AI startup reported surpassing $100 million in ARR when only a fraction came from currently paying customers.
- 03Scott Stevenson's X post exposing the practice drew over 200 reshares and comments from investors and founders.
- 04One startup marketed $50 million in ARR while actual revenue was $42 million, an $8 million gap investors knew about.
- 05Bessemer Venture Partners defines CARR as ARR plus committed but not yet live contract values, adjusted for expected churn.
Bessemer Venture Partners defined CARR in a 2021 blog post as ARR plus committed but not yet live contract values, adjusted for expected customer churn and downsell. The metric was intended to track growth momentum, but it is far more susceptible to gaming than traditional ARR because startups can inflate it by failing to account realistically for churn or by counting free pilots and discounted early contract years at full value.
One former employee at a startup that routinely reported CARR as ARR said the company counted at least one substantial yearlong free pilot as ARR, with the board and a VC from a large fund aware that the revenue from the eventual paying part of the contract had been counted during the pilot and that the customer could cancel before paying the full amount. Ross McNairn, CEO of legal AI startup Wordsmith, said he speaks to VCs regularly who acknowledge choppy standards around ARR reporting.
The pressure to show hypergrowth has intensified in the AI era. Hemant Taneja, CEO of General Catalyst, said on the 20VC podcast last September that going from 1 to 3 to 9 to 27 million in ARR is not interesting and that startups need to go from 1 to 20 to 100 million instead. Michael Marks, a founding managing partner at Celesta Capital, said the higher valuations in AI have made the incentives to inflate ARR stronger.
Stevenson and Jack Newton, CEO of legal startup Clio, both said VCs are often aware of the inflated figures but stay silent because overstating ARR helps kingmake their portfolio companies. When a startup publicly reports high revenue, it attracts better talent and customers who believe the company is the undisputed category winner. One VC said investors cannot call out the practice because nearly every firm has a portfolio company monetizing CARR as ARR.
Alex Cohen, CEO of health AI startup Hello Patient, said the practice is well known inside the industry and that insiders read the headlines with skepticism. Another layer of confusion comes from startups using a different ARR metric with the same acronym: annualized run-rate revenue, which extrapolates current revenue over the next 12 months based on a given period. This method is particularly misleading for AI companies that charge based on usage rather than locked-in contracts.
Not all founders are willing to play along. McNairn said he remembers the struggle startups faced justifying high valuations after the 2022 market correction and does not want to create an even higher hurdle by exaggerating his startup's revenue. He called the practice short-sighted and said it overinflates already crazy high multiples with poor hygiene that will eventually backfire.
Newton, whose Clio was valued at $5 billion last fall, said investors looking the other way when their own companies inflate numbers makes them look good from the outside. Stevenson said VCs are incentivized to create a narrative that they have runaway winners and to get press coverage for their companies, which aligns their interests with founders willing to misrepresent revenue.
The ARR inflation is hardly a novel phenomenon, but sources say startups have become far more aggressive amid the AI hype. The lack of formal audits for ARR, since generally accepted accounting principles focus on historical revenue rather than future revenue, leaves room for interpretation that some founders and investors are exploiting.
The practice creates a credibility gap between public pronouncements and private reality. Startups that report clean ARR figures are competing for talent and customers against peers whose inflated metrics suggest they are winning by wider margins than they actually are. That dynamic pressures even conservative founders to consider fudging their own numbers just to keep up, creating a race to the bottom that multiple sources said will eventually hurt the entire AI sector when the gap between reported and actual revenue becomes impossible to ignore.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
