How VCs and founders use inflated ‘ARR’ to crown AI startups

Scott Stevenson, co-founder and CEO of legal AI startup Spellbook, recently exposed what he termed a "huge scam" among AI startups: the inflation of publicly announced revenue figures. He highlighted that many AI startups are reporting impressive revenue records by employing dishonest metrics, a practice seemingly supported by major investment funds misleading journalists for PR. This sentiment resonated widely within the AI startup community, sparking considerable discussion and attracting attention from investors and founders.

Annual Recurring Revenue (ARR) is a vital metric historically used to represent the annual revenue from active, contracted customers. However, many AI companies are reportedly manipulating this metric. Instead of true ARR, they often substitute "contracted ARR" (CARR), which includes revenue from signed customers who have not yet been onboarded, making it a much "squishier" metric more susceptible to inflation.

Several sources—including founders, investors, and finance professionals—have confirmed this practice is common. Investors are often aware of these exaggerations, and some even participate, citing competitive pressures. The lack of formal audits for ARR by accountants, who focus on historical rather than future revenue, further enables this discrepancy.

The core problem with using CARR as ARR is the inclusion of revenue before a product is fully implemented. If implementation is delayed or fails, contracted revenue may never materialize. Instances have been reported where a significant portion of a startup's reported ARR came from contracts not yet deployed or actively paying customers, with some cases even involving lengthy free pilots being counted as immediate revenue. This inflates figures and creates a misleading impression of a startup's financial health and growth momentum.