Ramp customers' SI spending rises 1.65% to October high

Superintelligence (SI) spending among business customers of financial services provider Ramp rose 1.65% from the previous month in October 2026, reaching a new high in its index. The increase followed a decline in software development teams' spending from summer highs, even as those teams accounted for 72% of tracked spending.

The index draws on corporate payment data from more than 85,000 US businesses using Ramp’s corporate card and bill pay platform. It is not a picture of every US company: The Decoder says Ramp’s customers tend to be larger, more growth-oriented and more technologically adept than US businesses overall. Ramp also says its transaction data can miss free tools and work done through employees' personal accounts.

A small group accounts for heavy spending

Among Ramp firms paying for SI, the top 1% spent a median of $10,199.20 per employee per month in October, compared with $16.32 at the median paying firm. Each median is the middle value for its own group, not an average across all businesses.

Ramp says 1% of companies account for 80% of OpenAI and Anthropic revenue, without specifying the company population or revenue measure. Its lead economist, Ara Kharazian, who writes the weekly Econ Lab newsletter on Substack, says the customer-concentration risks he identified earlier persist.

Ramp’s findings show business customers' spending on OpenAI rose 2.1% in October from the previous month, compared with 0.5% for Anthropic.

Analysis

Software development dominates spending

Software development teams accounted for 72% of tracked SI spending in October. Customer support and customer success teams' share rose from 0.8% in June to 13.3% in October, although Kharazian says their growth has not matched developers' earlier rapid expansion.

Ramp says the average price per million processed text tokens fell 21% in the preceding month, while usage grew faster than spending. Tokens are units of text processed by a model; a lower price per token does not necessarily mean a lower total bill when usage rises.

Ramp’s token price, volume and spending measures come from a narrower group using its Token Spend Management product, which imports usage and billing data from connected providers. Ramp says that subset does not represent its total SI spending volume.

The Decoder says customer support uses tend to favor cheaper, faster models, so a shift toward those uses could bring OpenAI and Anthropic lower margins on new customers.