How AI Startups Hallucinate Their Revenue: CARR Vs. ARR
Why It Matters
Inflated ARR can mislead investors and overvalue AI startups, risking capital allocation errors.
Key Takeaways
- •AI startups often inflate ARR using future contract values.
- •True ARR requires fixed-price, non‑cancellable annual contracts for revenue.
- •Usage‑based and service revenue should not be counted as ARR.
- •Media and investors frequently accept overstated ARR without scrutiny.
- •Focus on GAAP revenue, retention rates, and margins for valuation.
Summary
The video examines how AI startups manipulate revenue metrics, especially ARR versus contracted ARR, and why this creates misleading growth signals.
It explains that legitimate ARR must come from fixed‑price, non‑cancellable annual contracts, not from month‑to‑month or usage‑based fees. The presenter shows Excel models where using year‑three contract values inflates ARR from $4‑5 million to $9 million, while GAAP revenue stays under $1 million.
Citing Scott Stevenson’s tweet and a Soundhound AI investor deck, the speaker highlights real‑world examples where companies label token fees and service fees as subscription revenue, blurring the line between ARR and actual cash earnings.
The takeaway for investors is to prioritize GAAP revenue, gross and net retention, and margin analysis over headline ARR numbers, applying lower valuation multiples to usage‑heavy AI firms.
Comments
Want to join the conversation?
Loading comments...