RTP Global investor says ARR alone no longer tells the story for AI startups
In a Crunchbase News column, RTP Global's Kenza Zayani argues that annualized revenue figures blur committed and usage-based income at AI companies, and that growth-stage investors will increasingly probe retention, contracted revenue and margins.
Annual recurring revenue is losing its standing as a sufficient measure of AI businesses, according to Kenza Zayani, an early-stage investor at London-based RTP Global. She makes the case in a column published by Crunchbase News on November 18. [1]
Zayani ties the timing to October, when, she writes, reports that OpenAI's annualized revenue was running below earlier estimates hit AI-linked stocks and renewed scrutiny of AI valuations. TechDots has not independently reviewed those reports; the account here is hers. [1]
Her core argument is structural. ARR suited software sold on predictable annual subscriptions, she writes, but most AI companies now combine a subscription base with consumption charges tied to API calls, tokens or compute, and a growing number bill per outcome. Annualizing that mix, in her view, merges revenue customers have committed to with revenue that depends on next month's usage. [1]
Zayani says what is labelled ARR is often a run rate: recent revenue multiplied to a yearly figure. She also flags that companies selling through partners may count either the full value of those sales or only their share, which she says can produce very different numbers for the same business. [1]
To illustrate, she contrasts two hypothetical companies each reporting a $10 million run rate: one built on contracted revenue from customers expanding steadily over a year, the other on a few customers whose consumption spiked in a single quarter. The headline number, she argues, does not distinguish them. [1]
Zayani says the right yardstick depends on stage. At pre-seed and seed, a steep trajectory can signal demand. Later, she argues, investors need customer concentration, the split between contracted and consumed revenue, predictability, gross margins at scale, and above all retention, expansion and churn. [1]
She does not expect ARR to disappear, but predicts more investors will ask what a figure counts, how much is committed and how retention looks by cohort, and says founders who arrive with those answers will be better placed. [1]
What we couldn't confirm
- The reports on OpenAI's annualized revenue referenced by Zayani are not named or independently confirmed in the source; no figures are given.
- The source is a single investor opinion column; no other investors, founders or data are cited to show how widely the view is shared.
- No market data is provided on the size of the October decline in AI-linked stocks.
- RTP Global's own portfolio exposure to AI companies is not detailed in the column.
Sources
- Crunchbase News: What To Measure When ARR Isn’t EnoughSource published · TechDots found it
Statements labelled as a company's claim have not been independently confirmed by TechDots.
Get Startups stories by email
Related coverage
ShareChat parent Mohalla Tech tops ₹1,000 Cr revenue as losses widen
Mohalla Tech, which runs ShareChat, Moj and QuickTV, reported its first year above ₹1,000 Cr in operating revenue in FY26, with QuickTV now a major income source. Net loss widened 79% to ₹1,978 Cr, and management's FY27 targets remain unverified by independent sources.
Indian women founders in deeptech cite traction but describe funding and credibility hurdles
HerStory interviewed eight women founders across seven Indian deeptech companies, who describe promising traction but also funding hurdles and gender-based assumptions. Separately, six Indian startups were picked for Hub71's Immersion Programme in Abu Dhabi.