The most AI-obsessed companies spend $7,500 per employee per month. The median spends $11.

The top 1% of US companies by AI adoption spend $7,500 per employee per month on AI tools and compute. The median firm spends $11.38. That 680x gap, drawn from the Ramp AI Index, is the clearest picture yet of how unevenly AI spending is distributed across American business.

Ramp describes the top 1% as “AI-pilled.” These firms are not yet spending more on AI than on people. A software engineer in the US earns roughly $16,000 per month, more than double the $7,500 figure. But the trajectory is steep. Among the top 1%, AI spend per employee grew 14.1% in the last month alone.

The top 10% of firms spend about $611 monthly per employee. That is roughly the cost of a few enterprise AI seats plus some API usage. The median, $11.38, is basically one subscription. For the vast majority of companies, AI spending is still a rounding error on their software budgets.

The data lands alongside growing anecdotes from the extreme end. An Nvidia executive recently said the cost of compute now exceeds the salaries of his employees. Mercor’s CEO said the startup spends more on tokens for internal agents than on headcount. Enterprise AI bills have tripled despite a 98% drop in per-token prices, because agentic tools drive consumption 18.6 times higher per developer.

The paradox is real. Tokens are cheap. Usage is not. A simple linear workflow in 2023 cost about $0.04 per interaction. An orchestrated agentic system in 2026 costs roughly $1.20, about 30 times more. Microsoft found individual engineers spending $500 to $2,000 a month on Claude Code tokens. Uber blew through its entire 2026 AI coding budget by April.

The AI-pilled firms tend to mix and match, bouncing between multiple frontier models and platforms that give them access to cheaper open-source alternatives. That behaviour suggests the power users are not loyal to any single provider. They are optimising for cost and capability simultaneously.

The question is whether the $7,500 figure is a ceiling or a floor. If agentic AI keeps expanding the scope of what companies automate, and if token spending becomes the third major cost centre after people and software, the top 1% may look like the median within a few years. For now, though, most of American business is still spending less on AI than on a decent lunch.