
The AI Startup Revenue Numbers You’ve Been Reading Are Fake.
The metric driving billion-dollar AI valuations is being inflated by 3 to 5 times by some startups. The biggest funds in the world are going along with it. And there’s a silent pact between founders and investors not to tell journalists the difference.
On April 17, 2026, Scott Stevenson posted something on X that nobody in the AI startup ecosystem wanted to see in public.
“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.” Fox Business
Stevenson is the co-founder and CEO of Spellbook, a legal AI startup. He is not a disgruntled outsider or a short seller with a financial interest in tearing down competitors. He is a founder operating inside the same ecosystem he just described as running a scam.
The post drew over 200 reshares and comments from high-profile investors, many founders, and generated several headlines. Fox Business
Chamath Palihapitiya commented. YC’s Garry Tan posted about correct revenue metrics. Jack Newton, co-founder and CEO of legal startup Clio, told TechCrunch that Stevenson had done a great job highlighting “nefarious practices.”
When prominent VCs, Y Combinator leadership, and successful founders all respond to a fraud accusation by essentially saying “yes, this is happening” — that response is the story.
The AI startup revenue boom may be one of the most sophisticated metric manipulation operations in the history of venture capital. And it has been running in plain sight for over a year.
The Metric That Built the AI Unicorn Era — And What It Actually Means
To understand the scam, you need to understand the metric being manipulated.
Annual Recurring Revenue — ARR — is the number that appears in every AI startup headline. “$50M ARR.” “$200M ARR.” “Fastest startup to $100M ARR in history.” It is the primary signal that journalists report, investors benchmark against, and employees use to understand how their company is performing.
ARR is meant to show the annualized value of recurring subscription contracts. It’s typically calculated by projecting the current month’s subscription revenue over a full year. So if a startup invoices $1 million in January, its ARR for the current year would be $12 million, on the assumption that the same monthly revenue will continue. Fox Business
The key word in that definition is “invoices.” ARR is supposed to represent money that has actually been billed to customers under active contracts — real revenue that is predictably recurring.
Stevenson said some AI startups have begun basing ARR figures on future revenue that is far from certain, noting that they do this by blurring ARR with something called CARR — contracted annually recurring revenue — which can include future revenue that may never materialize. Washington Times
CARR counts contracts that have been signed but not yet invoiced. It counts pilots that haven’t converted. It counts commitments from customers who can still cancel. It counts revenue that might arrive — not revenue that has.
When a startup reports CARR as ARR — and labels it simply as “ARR” in their press releases, investor decks, and media interviews — they are presenting a future revenue possibility as a current revenue reality.
Stevenson said: “The gap between these metrics has grown massively. I know 100% of confirmed cases where the gap is as much as 3 to 5 times.” Washington Times
Three to five times. A startup reporting $100M ARR may have $20M to $33M in actual invoiced revenue.
The headline number driving the valuation, the media coverage, and the competitive benchmark that other startups feel pressure to match is potentially three to five times larger than the real number.
The Specific Techniques Being Used
The manipulation is not a single clean fraud. It is a collection of practices that individually sound defensible and collectively produce a number that bears little relationship to actual business performance.
A startup might count a free three-month pilot as three months of real revenue — even though the customer hasn’t paid anything and has no obligation to continue after the trial.
A startup might count a full year of contracted value even when the contract includes monthly opt-out provisions that allow the customer to cancel after 30 days. The contract says “annual” but the customer’s actual commitment is one month.
A startup might take their highest revenue month — perhaps inflated by a promotional offer, a one-time enterprise deal, or seasonal demand — and annualize it as though that month represents their steady-state run rate.
One month of revenue multiplied by twelve, and many companies update ARR every month. When a company announces their ARR in August of 2026, they are not saying that it will have those revenues in 2026 but actually between July 2026 and June 2027. Tom’s Guide
Projecting forward from a single elevated month and calling the result “current ARR.” Counting pilots as revenue. Treating cancellable monthly contracts as annual commitments. Each practice sounds like an accounting judgment call. The combination produces a number that is systematically and significantly higher than what the company is actually earning.
The Case Study That Made It Impossible to Ignore
The abstract argument about metric manipulation became concrete in March 2026 when the CEO of Cluely — a fast-growing AI startup backed by Andreessen Horowitz — publicly admitted to inflating the company’s revenue figures.
In the admission posted on X in early March 2026, the CEO revealed the actual picture: consumer ARR of $2.7M with a run rate of $3.8M, and enterprise ARR of $2.5M with a run rate of $2.5M, totaling roughly $5.2M in actual ARR at the time. That’s not a rounding error. It’s a $1.8M gap — approximately 35% inflation on a headline metric that investors, press, and potential customers use as a primary indicator of business health.
35% inflation. On a metric reported to journalists who covered the company’s growth. By a startup funded by one of the most prestigious venture firms in the world.
And in the same post where the CEO admitted the lie, he also misrepresented the circumstances under which he gave the original interview, claiming it was a cold call rather than an interview his own PR representative had proactively arranged. Two lies in one correction. The math on the honesty wasn’t great either.
Two separate misrepresentations in the same public correction. The inflation of the revenue metric was not an isolated mistake in a culture of honest reporting. It was the visible expression of a broader relationship with factual accuracy.
Cluely was not a failed startup. It had real revenue, real customers, a real product, and real backing from A16Z. It inflated its numbers anyway — because the culture around it made inflation normal and the incentives made honesty feel like a competitive disadvantage.
The Part Where the VCs Knew
This is where the story moves from startup misconduct to something more structurally alarming.
One investor told TechCrunch: “For sure they are reporting CARR as ARR. When one startup does it in a category, it is hard not to do it yourself just to keep up.” Geeky Gadgets
An investor — speaking anonymously but candidly — confirming that the metric inflation is happening, acknowledging they know it’s happening, and explaining why the incentive structure makes it spread: when one startup inflates, every other startup in the same category faces competitive pressure to match the inflated number or appear to be losing.
“The valuations have gotten higher, and so the incentives are stronger to do it,” Michael Marks, a founding managing partner at Celesta Capital, told TechCrunch. Fox Business
The incentive structure is explicit. Higher valuations require higher revenue multiples. Higher revenue multiples require higher reported revenue. The path of least resistance to a higher headline valuation runs directly through an inflated ARR number that sophisticated investors privately understand is inflated while publicly treating as real.
Stevenson described a “silent pact” between founders and VCs not to discuss the difference with press, and to often use the bigger number for more coverage. “I feel like there is a bit of a silent pact between founders and VCs not to discuss the difference with press, and to often use the bigger number for more coverage,” he said. CNBC
A silent pact. Between founders and the investors funding them. To present inflated numbers to journalists who will report those numbers to the public — retail investors, potential employees, enterprise customers — who make real decisions based on the coverage.
The founders know the number is inflated. The VCs know the number is inflated. The journalists don’t know. The employees don’t know. The customers don’t know. The retail investors don’t know.
The people with the most information are collectively maintaining a fiction for the benefit of those with the least.
The Second-Order Effects Nobody Is Modeling
Stevenson warned of insidious second-order effects: “These illusions can create mania, cause companies to chase each other’s ghosts, and to do risky things that they shouldn’t — also very bad for employees who may not understand real ARR numbers, and for customers trying to understand the landscape.” CNBC
The damage from inflated ARR numbers is not limited to the companies doing the inflating.
When a startup in a given category reports $100M ARR that is actually $25M ARR, every other startup in that category faces a benchmark they need to compete against. They see a competitor apparently generating four times their actual revenue. They make hiring decisions, product decisions, and fundraising decisions based on a competitive landscape that doesn’t exist as described.
Employees join companies based on growth narratives built on inflated metrics. They negotiate compensation, make career decisions, and build professional identities around working at a “rocketship” — without understanding that the rocket’s fuel gauge has been tampered with.
Enterprise customers select vendors partly based on scale signals. A startup appearing to have $100M in revenue signals stability, product-market fit, and the likelihood of still existing in three years. A startup with $25M in real revenue signals something different. Customers are making vendor decisions based on false signals about vendor health.
And the employees at established companies being disrupted by AI startups are having their livelihoods threatened by competitors whose actual competitive position is significantly weaker than the public narrative suggests.
The founder who inflates ARR to journalists is building a company where salespeople inflate close probability to managers. When leadership demonstrates that metrics are negotiable, it establishes a cultural permission structure where the same logic migrates into sales forecasting, customer success reporting, and pipeline management.
Metric fraud at the top of an organization does not stay at the top. It becomes the cultural standard — the implicit message to every person in the company that numbers are instruments of narrative rather than instruments of truth.
Why This Is Harder to Stop Than It Looks
The obvious solution seems simple: use real ARR. Report actual invoiced revenue. Don’t count pilots. Don’t count opt-out contracts as annual commitments.
The actual incentive structure makes this solution individually irrational even for founders who understand why it’s collectively necessary.
If you are a founder reporting real ARR of $25M in a category where the apparent market leader is reporting $100M — and you know their $100M is actually $25M to $33M, but the press doesn’t — your honest reporting makes you look like you’re losing a race you’re actually running even with.
Most people interviewed for this story said that ARR overstatements of all kinds are hardly a novel phenomenon, but startups have become far more aggressive amid the AI hype. Fox Business
The AI hype cycle created conditions where the incentives for metric inflation are stronger than they have ever been. Valuations are higher. The pressure to demonstrate hypergrowth is more intense. The competitive signaling function of ARR numbers is more important. And the willingness of major VC funds to accept and amplify inflated metrics provides institutional legitimacy that makes the practice feel normalized.
The funds that have the most power to stop it — by requiring audited ARR definitions as a condition of investment, by refusing to amplify CARR numbers in their portfolio announcements — are the ones with the strongest incentive to perpetuate it, because their own portfolio valuations depend on the same inflated metrics.
What You Should Do With This Information
If you are an investor evaluating AI startups: the ARR number in the headline of a press release or pitch deck is probably not what you think it is. Ask specifically for invoiced ARR — money that has actually been billed under contracts with no near-term opt-out. Ask for the gap between that number and any CARR figures. Ask what percentage of reported ARR came from free pilots, single-month opt-out contracts, or non-renewed enterprise deals. The answers will tell you more about the real business than any benchmark comparison to competitors whose numbers are equally inflated.
If you are an employee at an AI startup: the ARR your company’s leadership mentions in all-hands meetings may not reflect what you think it reflects. Ask your finance team to explain the calculation methodology. Understand whether your equity and compensation decisions are being made against a real business performance picture or an inflated one.
If you are a journalist or analyst covering AI startups: the coverage has been consistently failing to ask the question that changes the entire story. Not “what is your ARR?” but “how do you define ARR, and what is the gap between that number and your invoiced recurring revenue from contracts with no near-term cancellation rights?” The unwillingness to ask that follow-up question is part of what the silent pact depends on.
If you are a founder under pressure to inflate your metrics to compete: Stevenson’s public post generated 200 reshares from prominent investors and got him covered in TechCrunch, Fast Company, and Yahoo Finance. The founders who build reputations for honest metrics in an environment of systematic inflation are building something with compounding value — trust from sophisticated investors who are exhausted by the manipulation and are actively looking for signals they can believe.
The choice is not between inflated metrics and boring anonymity. There is a third path — one that uses real operational transparency as a differentiated competitive signal, particularly in an environment where sophisticated audiences are increasingly capable of detecting AI startup ARR manipulation.
The Question This Entire Story Keeps Circling
The AI startup boom of 2025 and 2026 will be remembered as one of the most extraordinary periods of capital formation in technology history. The genuine capability advances are real. The market opportunity is real. The companies solving genuinely hard problems with AI are building real value.
But if the revenue numbers that have defined the narrative of that boom — the $100M ARR milestones, the record-breaking growth curves, the valuation multiples that make previous technology companies look conservative — are systematically overstated by three to five times, then the narrative of the boom is built on a foundation that does not match the underlying reality.
The dot-com crash was not caused by the internet being fake. The internet was real. It was caused by the valuations being built on projections that the underlying businesses could not support when the hype cycle turned.
The AI boom may not be the dot-com crash. The technology is more capable, the enterprise adoption is more real, and the revenue — even the inflated revenue — represents genuine market traction.
But a boom where the metrics driving the valuations are systematically and knowingly inflated by the people closest to the data, with the active participation of the funds whose returns depend on maintaining the fiction, and the passive participation of journalists who don’t know to ask the right follow-up question — that boom has a structural vulnerability that compounds quietly until it doesn’t.
Scott Stevenson just described it publicly.
The biggest funds in the world already knew.
The question is who finds out next.
If this changed how you’re reading AI startup revenue announcements — share it with someone still taking the headline ARR numbers at face value. Drop your take in the comments.
