Stripe's payments data shows its top AI customers accelerating from 120% to 175% growth

Compounded, the two rates put the cohort's revenue at about six times its starting level after two years, since 2.2 times 2.75 is 6.05 ^[1]. In B2B the normal pattern runs the other way, with growth rates decaying as companies get bigger ^[2]. Stripe's basis for the claim is that it processes payments for most of the fastest-growing AI companies and so sees their revenue data ^[3].

How the 175% was built decides what it means. Anthropic's $30 billion run rate is 15 times Cursor's $2 billion and 75 times Lovable's $400 million, using the figures as given ^[2]. If the cohort rate is growth in combined revenue, a single company that size can carry it. If it is a median of each member's growth, it describes a typical company. Membership matters too: a group labelled top at the end of 2026 will tend to show acceleration, because the members that slowed are the likeliest to have dropped out of it. SaaStr's account of the session does not say how Stripe defined the cohort or whether the same companies appear in both years.

The acceleration at the named companies holds up on its own figures. Lovable quadrupled from $100 million to $400 million in eight months ^[4], and Cursor doubled its run rate in three ^[5]. I think the cohort rate is a weaker number than those. The counter-case is that Stripe sees enough of the market for selection effects to wash out, and that the top of the AI market really is getting faster as it gets bigger. A same-company figure from Stripe that stays above 120% next year would settle it against me, and would mean the decay assumption is wrong for this group.

The 48% foreign share rests on more evidence, or rather on more numbers from the same source. AI companies reached 42 countries in their first year and 120 by year three ^[8], and Gamma, based in San Francisco, did $100 million in its first year with most of it from outside the U.S. ^[10]. The older B2B plan was to win the home market, reach product-market fit at scale and then hire a general manager in Dublin or London ^[9]. On Stripe's figures, these companies are selling in dozens of countries during the year the old plan reserved for the home market ^[8][9].

Stripe puts the lift from localized pricing at 18% of cross-border revenue ^[11]. For a company that has not yet localized and has a 48% foreign share, that is about 8.6% of total revenue, or 0.48 times 0.18 ^[3]. One added local payment method is worth 7% or more, by Stripe's count ^[11]. Both are Stripe's own measurements of changes made at checkout, and the test SaaStr attributes to Josebachvili is whether a customer in Brazil can pay in reais with Pix ^[16].