AI Startups Inflating Revenue with Misleading ARR Metrics – The Truth Behind the Hype

Thousands of AI startups are fighting for the venture capital funding needed to win a slice of the enterprise market, but many are inflating their real revenue to gain an advantage. According to Scott Stevenson, cofounder and CEO of the legal AI startup Spellbook, these companies are perpetuating a “huge scam” in their metric reporting by misusing annual recurring revenue (ARR) figures.

In a viral tweet on April 17, 2026, Stevenson called out fledgling AI companies for blurring the lines between ARR and contracted annually recurring revenue (CARR), a practice he says is being enabled by major venture capital firms and used to mislead journalists for public relations coverage. He stated that the reason many AI startups are reporting record revenue growth is because they are using a dishonest metric that presents projected or uncertain income as actual, recurring income.

ARR is meant to reflect the annualized value of recurring subscription contracts that are already being invoiced to customers. It is typically calculated by multiplying the current month’s recurring revenue by twelve, under the assumption that the same revenue will continue steadily over the year. However, Stevenson explained that some startups are now incorporating future revenue—such as income from features not yet built or payments contingent on future milestones—into their ARR figures by labeling them as CARR and then presenting the combined figure as ARR in press materials and pitch decks.

While CARR can be a legitimate metric for long-term contracts where revenue accrues gradually—such as in healthcare AI or energy optimization projects—Stevenson said the distinction between CARR and actual ARR has been deliberately obscured. “Often in decks CARR and ARR are reported as separate metrics, but when companies go to press they are actually reporting CARR and calling it ARR in order to have the biggest number possible,” he told Fast Company in an email exchange.

He emphasized that the gap between legitimate ARR and inflated figures is not marginal. “I know 100% of confirmed cases where the gap is as much as 3-5x,” Stevenson said, noting that some startups count revenue from short-term pilots that may not convert, or assume customers will pay for features still under development, despite no guarantee those features will be delivered or adopted.

Examples include counting a full year of revenue from a contract that allows the customer to opt out after one month, treating a free three-month pilot as three months of real revenue, or booking revenue during the development phase of a feature that the customer has not yet committed to purchase. Stevenson recounted speaking with an investor who frequently sees early-stage companies claiming a million dollars in ARR, only to discover upon review that the figure is based entirely on unconverted pilot programs.

The issue drew public agreement from other investors and founders. Equal Ventures partner Rick Zullo replied to Stevenson’s post, stating, “This is rampant and it’s honestly distorting the benchmarks for everyone.” FPV Ventures partner Nikunj Kothari added, “I have stopped looking at headline number ” indicating that even experienced investors are becoming skeptical of publicly reported ARR figures.

Stevenson warned that journalists, who typically lack access to a startup’s internal contracts, may unintentionally amplify the deception by taking companies at their word when they report ARR. He urged reporters to probe whether a startup’s ARR figure reflects only “live” revenue—money already invoiced—or if it includes “contracted ARR” based on uncertain future income. He suggested that some venture capitalists may be complicit in the practice, noting a “silent pact” between founders and VCs to avoid discussing the distinction with the press and to consistently use the higher number for greater media coverage.

The consequences of this trend extend beyond misleading investors. Stevenson argued that when one startup inflates its revenue metrics, competitors may feel pressured to follow suit to avoid appearing inferior, creating a cycle of exaggeration that distorts industry benchmarks. This can lead to misallocated capital, risky business decisions, and confusion among employees and customers trying to assess the true state of the market.

He also connected the issue to broader skepticism about the economic viability of AI ventures, noting that overstating the financial performance of AI companies—whether large tech firms investing in infrastructure or startups building applications on top of AI models—only adds to market hype and increases the risk of a speculative bubble forming around unproven revenue claims.