AI startups accused of inflating ARR with “committed” revenue and run‑rate math

What happened

It has been reported that a wave of AI startups are reclassifying future or short‑term income as Annual Recurring Revenue (ARR) to create a faster growth narrative, according to a TechCrunch investigation reposted by ifeng. The pushback began after Scott Stevenson, co‑founder and CEO of Spellbook, publicly warned that headline ARR figures were outpacing real, deployed customer revenue — and the comment sparked wide debate among founders, VCs and finance teams. TechCrunch said it interviewed more than a dozen industry insiders who described systematic use of alternative metrics to magnify topline numbers.

How they do it

Startups are reportedly swapping traditional ARR — contracted, deployed, and billable annual revenue — for CARR (Committed ARR), which counts signed but not yet delivered or active contracts. Bessemer Venture Partners (BVP) already cautioned in 2021 that CARR must be adjusted for churn and discounting before it becomes meaningful. But some founders and investors, it has been reported, list CARR as ARR with little or no reconciliation. Others use short‑period Annualized Revenue Run Rate (also abbreviated ARR) by extrapolating a month’s or quarter’s revenue to a 12‑month figure, despite highly variable usage‑based billing in many AI business models. Reportedly, some CARR disclosures have been as much as 70% higher than realized ARR, and in certain cases free trial periods have been counted as full ARR — claims that many sources said their boards tacitly accepted.

Why it matters

Why does this matter? Because ARR has long been a core signal of SaaS health and revenue quality. Inflating it distorts hiring, fundraising and M&A decisions, and can entrench valuations that depend on continuation of that “growth” narrative. Several investors told TechCrunch they prefer not to publicly challenge overstated numbers because inflated metrics can help portfolio companies attract talent, customers and follow‑on capital — and, in turn, polish the investor’s track record. This creates a short‑term feedback loop that rewards opacity.

Consequences and pushback

Not everyone is comfortable with the practice. Some founders and CEOs are pushing back, insisting on reporting only deployed, billable ARR and warning that the sector risks a credibility deficit if numbers don’t convert into cash. As the generative‑AI boom draws intense global capital and regulatory scrutiny, transparency over revenue definitions will matter more — for investors, buyers and for cross‑border policy debates about how to evaluate and regulate fast‑moving AI firms. If growth cannot be delivered, the cost may not only be writedowns but a broader erosion of trust in an industry that depends on credibility to scale.