Justin PenningtonJustin Pennington
All posts

The Median Business Spends $16 a Month Per Employee on AI. Price Isn't the Problem.

Ramp's latest AI Index shows token prices down 21% in a month while the typical company still spends $16.32 per employee. The gap between the top 1% and everyone else is a workflow gap, not a budget gap.

Ramp published its November AI Index yesterday, and one number from it stuck with me. Among businesses paying for AI, the median firm spends $16.32 per employee per month. The top 1% of firms spend a median of $10,199.20.

That's not a rounding difference. That's two different ways of running a company.

What changed

Ramp's lead economist, Ara Kharazian, tracks AI spending across more than 85,000 American businesses using Ramp's card and bill-pay data. The October numbers:

  • Total AI spend hit a new high but grew only 1.65% from September.
  • The average price per million tokens fell 21% in a single month, and token volume keeps growing faster than token spend.
  • About 1% of businesses drive roughly 80% of the revenue going to OpenAI and Anthropic, and that group is mostly venture-backed tech, finance and technical-services firms.
  • Engineering teams account for 72% of classified AI spend. The fastest-growing new category is customer success and support, which went from 0.8% of classified spend in June to 13.3% in October.

Ramp is upfront that its customer base skews toward growth-oriented, tech-forward companies, so if anything the typical business is spending less than this.

Why it matters for operators

The usual explanation for slow AI adoption is cost. This data says otherwise. The work is getting cheaper every month, and the median company still isn't buying much of it.

Here's my read: the top 1% aren't spending more because they're richer. They're spending more because they've handed AI an actual job. Their engineering teams run coding agents all day. That's a defined workflow with inputs, outputs and someone checking the result. Spend follows work.

The median company bought a few chat seats. People open them when they remember to. That's $16 a month of curiosity, not $16 a month of operations.

The support numbers back this up. Support is the first department outside engineering where spending is climbing, and it's also the first department where most businesses already have a clear process: a ticket comes in, someone answers it, someone checks whether it got resolved. When the process exists, AI has somewhere to go.

My take

If you run a small or mid-size business, don't read "AI spend is flattening" as "AI isn't working." Read it as "most companies haven't picked the workflow yet."

Falling prices make the decision easier, not harder. Pick one process that already has a clear trigger and a clear finish line, like inbound leads, quotes, invoice follow-ups or service tickets. Write down who owns it today, what "done" looks like and where the records live. Then put AI inside that process, with a person approving anything that leaves the building.

That's how we build at Infraxio. Growth7 starts with the lead and the follow-up, not the chat window. IFX Bid starts with the bid package and the deadline. IFX Hub keeps the records in one place so the agent is working from the same data your team is. The model is the cheap part now. The process around it is where the value is.

The gap between $16 and $10,000 isn't going to close by waiting for another price cut. It closes when you decide what work you're handing over.

Sources