AI startups routinely report CARR as real ARR
Key insights
- One VC told TechCrunch that CARR at some AI startups runs 70% above actual recognized revenue.
- Counting full-year value of monthly opt-out contracts and logging free pilots as paid are the two primary inflation tactics.
- ARR is not a GAAP metric, making these practices legal and currently immune to audit requirements.
Why this matters
Valuation multiples and late-stage funding rounds across the AI sector are being benchmarked against ARR figures that may systematically overstate true recurring revenue by half or more, meaning portfolio markups and secondary prices are built on unverified numbers. For founders raising their next round, the normalization of CARR-as-ARR creates competitive pressure to inflate or lose the headline comparison, distorting the entire market signal that investors use to allocate capital. Technical leaders evaluating build-versus-buy decisions or partnership terms with AI vendors are now working with pricing and stability signals that may not reflect the vendor's actual contracted revenue base or customer retention.
Summary
Spellbook CEO Scott Stevenson has gone public accusing AI startups of systematically misrepresenting their revenue by reporting CARR (Contracted ARR) in place of actually recognized ARR, a gap one VC told TechCrunch can run as high as 70% between the two figures.
The mechanics are specific: companies count the full annualized value of monthly opt-out contracts, treat free pilots as paid revenue, and present CARR as if it were booked income. Because ARR is not a GAAP metric, none of this triggers an audit or legal liability, and major VC funds are reportedly encouraging the practice to manufacture headline numbers for press cycles.
Essentially: (Spellbook, unnamed VC-backed AI startups) have normalized a reporting standard where the number in the press release and the number in the bank account are materially different.
- One VC cited instances where CARR ran 70% above actual recognized revenue across portfolio companies.
- Monthly opt-out contracts being annualized and counted in full inflates ARR without any committed customer obligation.
- Free pilots logged as paid revenue means customer acquisition costs are effectively hidden inside inflated top-line figures.
The downstream effect is that valuation multiples, fundraising rounds, and secondary-market prices are all being set against a revenue base that has never been independently verified.
Potential risks and opportunities
Risks
- Late-stage investors holding secondary positions in AI unicorns could face write-downs if due diligence in upcoming down-rounds exposes the CARR-to-ARR gap at portfolio companies.
- AI startups that have used inflated ARR to negotiate enterprise contracts with revenue-based pricing floors could face clawbacks or renegotiation when actual recognized revenue is disclosed.
- VC funds that publicly cited portfolio ARR figures in LP reports or fundraising decks face potential LP disputes if those numbers are shown to reflect CARR rather than booked revenue.
Opportunities
- Revenue intelligence platforms (Mosaic, Maxio, Subscript) can differentiate by offering GAAP-reconciled ARR dashboards that explicitly separate CARR, pipeline, and recognized revenue for CFO buyers under investor scrutiny.
- Due diligence firms and technical audit providers gain leverage to expand scope in AI-sector transactions, with specific demand for contracted-versus-recognized revenue waterfall analysis.
- Investors who can accurately normalize ARR across AI portfolios now have a structural information advantage for pricing secondary transactions and identifying overvalued late-stage rounds before the correction.
What we don't know yet
- Whether any of the named VC funds have issued internal portfolio guidance defining ARR reporting standards, and if so when those policies were adopted.
- Which specific AI startups Stevenson's allegation is based on, and whether any have publicly responded or corrected prior ARR disclosures.
- Whether the SEC or PCAOB has any active review of non-GAAP metric disclosures in AI-sector fundraising materials as of mid-2026.
Originally reported by techcrunch.com
Read the original article →Original headline: TechCrunch Investigation: AI Startups Are Systematically Inflating ARR Metrics — CARR Counted as Revenue When Much of It Never Materializes