AI startups and investors are accused of inflating ARR figures

In short: A TechCrunch report says some AI startups and investors are publicly overstating revenue by using looser definitions of “ARR.”

What's going on

“ARR” stands for annual recurring revenue, a common way software companies describe steady, repeating income from customers. Think of it like saying, “If nothing changes, this is what we should make in a year from current paying customers.” TechCrunch reports that some AI startups are stretching what counts as ARR when they talk to the public.

One common tactic is to use “contracted ARR,” also called “committed ARR” or CARR, and label it as ARR. CARR can include signed deals where the customer is not fully set up yet, or may still be in a trial period. That is closer to counting money you expect to get, rather than money that is actually coming in.

Sources told TechCrunch that some companies have reported large ARR milestones even when only a fraction came from active paying customers. One former employee said their company even counted a yearlong free pilot as ARR, even though the customer could cancel before paying. A venture capitalist told TechCrunch they have seen cases where CARR was 70% higher than true ARR.

TechCrunch also notes confusion with another “ARR” meaning, annualized run-rate revenue. This takes a short period of sales and multiplies it to estimate a full year, which can be shaky for AI tools where spending can jump up and down.

What to watch

The big risk is that headlines about “record revenue” can shape who gets hired, who wins customers, and who attracts more investment, even if the numbers are not apples to apples. Watch for startups to clearly separate money already being paid from money that is only signed or predicted, especially as more AI companies try to raise funding or prepare for public-market scrutiny.

Source: TechCrunch AI