Home / News

Ramp data: AI usage keeps rising while company spending on AI stalls

For a small business, the model bill is now the smallest line in the cost of an AI assistant: according to Ramp, the median company spends $12-15 per employee per month on AI. Most of the budget for a small-business AI assistant goes into building and connecting the workflow, not into tokens.

That figure comes from a new Ramp spending report covered by TechCrunch. Token volumes keep climbing, but spend grew only 1.5% across August and September, after 42% in July. The cause is a price war: OpenAI halved API prices for GPT-6 Sol and Luna, Anthropic cut prices for Fable 5.1 and Opus 5.5, and heavy users route simple tasks to cheaper models.

What this means for a small business

Ramp's economist says a doctor's office or a car dealership won't need anything close to the $10,000 per employee the top 1% spends. That matches my projects. In a lead-intake or invoice pipeline, most steps are yes/no classification and field extraction, and a light model handles them at low cost. I keep a heavy model like Opus for the one step that needs judgment, such as flagging a risky contract clause, and put hard limits on long-running agent tasks. That routing logic and the integrations are what drives a project's cost, not API usage.

Why it matters: When you budget for an AI assistant, plan the one-time build carefully. Treat the monthly model cost as small but capped, and set spending limits and model routing from day one.

Source: TechCrunch