AI-Pilled Firms Spend $7,500 Per Employee a Month

Corporate AI adoption is no longer moving along one smooth curve. It has split into a canyon. Most companies are still nibbling at AI, while a small group is spending at a level that turns experimentation into a new operating cost.

Ramp calls these companies “AI-pilled.” According to Ramp’s June 2026 analysis, the top 1% of AI-spending firms were spending about $7,500 per employee each month. The top 10% spent about $611. Median spending was just $11.38. That gap says more about organizational conviction than tool availability.

Spending Identity

AI-pilled captures a corporate identity built around aggressive AI use: multiple frontier models, coding agents, APIs, open-source access platforms, and experiments running across the organization. The large software bill is one consequence.

In the month covered by the TechCrunch report, spending among this top 1% rose another 14.1% per employee. These companies were not settling on one enterprise chatbot. They were mixing models and services, trying to put AI into as many workflows as possible.

The enthusiasm also reveals how concentrated corporate AI spending remains. Ramp’s AI Index measures observed payments by American businesses using its corporate cards and bill-pay systems. Its methodology notes that free tools and employee purchases made through personal accounts may not appear in the data.

Even with that limitation, the split is striking. One group is treating AI as a new operating layer. Much of the market is still buying roughly one modest subscription per employee.

Cost Before Proof

High spending is not automatically waste. A company that spends $7,500 per worker could be replacing far more expensive labor, launching products faster, or producing revenue that justifies every token.

But spending intensity is not the same as return. The figure arrives as companies are struggling to connect AI usage to shipped products, lower costs, or higher revenue. It also lands alongside a new vocabulary of restraint: token budgets, model routing, usage caps, observability, and token rationing.

That tension separates AI-pilled from tokenmaxxing. AI-pilled measures the scale of a company’s appetite. Tokenmaxxing describes the behavior that appetite can encourage, including internal pressure to use more AI simply because heavy usage signals commitment.

Extreme Users

Every technology cycle produces power users. What makes this one unusual is that autonomous systems can spend money continuously. An agent can call another model, revise its work, search again, or run an entire coding task while the meter keeps turning.

The top 1% may be showing where corporate operations are headed. They may also be demonstrating how quickly AI costs can escape conventional software budgets.

$7,500 per employee per month is the price tag on a belief: use enough AI now, and the return will eventually catch up.