The $11.38 median

The median US business on Ramp's corporate-card platform spends $11.38 per employee per month on AI. The top 1% spend $7,449 — about 650 times more, and 14.1% more than the month before. That distribution, not the vendor race, is the story in Ramp's reworked June index, which switches from asking whether a business pays for AI to how hard it is leaning in.

Even the top 1% of AI spenders pay, per employee, less than half an engineer's monthly salary.

The numbers come from aggregated, anonymized transactions across more than 70,000 firms — first-party receipts rather than a survey, which is why they're worth reading even though Ramp owns the dataset and no outside party has audited it. They describe Ramp's customers, who skew startup and finance-forward, not US business at large.

The shape they reveal is a thin, fast-moving frontier pulling away from a near-flat field. A small set of firms is now spending engineer-scale money on models; the typical firm is barely AI-pilled at all. Ramp's own deflating footnote: even that top 1% spends less than half an engineer's monthly salary — so the pitch that companies will soon pay as much for AI as for the people it augments is, for now, not what the cards show.

In the background, Anthropic holds its lead over OpenAI in share of businesses paying — 41% to 39.5%, a second straight month ahead after the gap first opened in May. But that counts firms with a subscription, not dollars or seats, and Ramp notes the heaviest spenders don't pick a side: they juggle several frontier and open models at once rather than betting the budget on one lab.