US Corporate AI Spending Reaches a Record, but Growth Cools to 1.65%

US corporate AI spending reached a new high in October 2026, but the month-over-month gain was small. The Ramp AI Index shows spending rising 1.65 percent, which tells you less about broad adoption than about who is doing the spending. Here's the thing: the average is being pulled by a narrow group of heavy users, while most companies are changing their budgets only slightly.

A split between heavy users and everyone else

The clearest signal in the data is the gap between the top of the distribution and the middle. The top one percent of companies by AI spending paid a median of $10,200 per employee per month, up nine percent. A typical company spent $16.32 per employee, an increase of less than a dollar from the previous month. Those two numbers describe very different businesses. One group is building AI into core workflows, while the other is still experimenting with a handful of subscriptions.

Ramp's index draws on payment data from more than 85,000 US companies, which gives it real breadth. But it measures what passes through company cards and accounts. Free tools and usage through personal accounts are likely missing, so the figures understate some kinds of employee experimentation. Ramp also notes that its customer base leans toward growth-oriented, tech-savvy and larger firms compared with US businesses overall. That means AI adoption across the wider economy is probably lower than these charts suggest, and the typical-company figure is best read as a ceiling for smaller, less digital firms rather than a national average.

Ramp's separate metrics on token volume and spending come only from companies using its Token Spend Management product, so they don't represent total AI spending on the platform. Treat those numbers as directional.

OpenAI pulls ahead of Anthropic on spending growth

Business spending on OpenAI rose 2.1 percent month over month, compared with 0.5 percent for Anthropic. On adoption, Anthropic reached 44.8 percent of tracked companies, up 1.1 percentage points, while OpenAI held steady at 40.6 percent. Overall adoption of AI services rose to 56.8 percent. Those two metrics don't contradict each other. Anthropic is winning more new accounts, while OpenAI's existing customers are expanding their usage faster.

Ara Kharazian, Ramp's chief economist, says the structural risks he flagged in earlier analysis still hold. About one percent of companies generate 80 percent of revenue for both OpenAI and Anthropic. That concentration matters for anyone reading these labs' growth numbers. A small cohort of large accounts can carry a revenue line, and if that cohort changes its buying, the picture shifts quickly. The top tier is mostly venture-backed firms in tech, finance and technical services. Outside those sectors, adoption is barely spreading.

The index also includes a surprise. SpacexSI grew 2.2 percent, the fastest rate among software companies in Ramp's dataset, and holds a 6.3 percent share on the platform. My read is that this says more about the sample than about the company's models, which Ramp's framing suggests aren't the main draw. The growth is probably tied to the coding tool Cursor, which belongs to SpacexSI. A share that high in a tech-skewed sample shouldn't be extrapolated to the broader market.

Software teams still dominate, but support is catching up

Software development teams accounted for 72 percent of all tracked AI spending in October. That share is down from summer highs, though, and the decline is worth watching. It suggests that coding remains the anchor use case while other functions begin to add real volume.

The most notable shift is in customer support and customer success. Their share of AI spending climbed from 0.8 percent in June to 13.3 percent in October. Ramp links the jump to AI chatbots deployed in customer service, which is a plausible explanation given how easily those tools can be rolled out against existing help desk workflows. Leadership teams also increased spending slightly.

Kharazian's assessment is measured. Customer support growth is nowhere near the pace development teams posted earlier this year. That's a useful reminder that a fast-rising category in percentage terms can still be small in dollar terms, and that a function's share of spending depends heavily on where it started.

Falling prices, rising model tiers and thinner margins

Prices are moving in the opposite direction from spending. The average price per million processed text tokens fell 21 percent in the latest month, according to Ramp. Usage is growing faster than spending, so each unit of AI work is getting cheaper. Ramp attributes the decline to competitive pressure between OpenAI and Anthropic, along with open-weight models from China.

Yet the mix is shifting toward premium models. The share of spending on the most expensive models, which Kharazian groups as Opus, Fable, Sol and Astra, edged up to 69.5 percent. Those two trends can coexist. Companies pay less per token but route more of their work to flagship models, so their total bills still rise.

That combination has consequences for the labs. Customer support and similar use cases tend to favor cheaper, faster models over the most expensive flagships. If new customers come from those functions, the labs may see lower margins per account even as their customer counts grow. In practice, the economics of AI services will depend on whether premium workloads keep their share as cheaper models absorb simpler tasks.

What to watch next

Three things will tell us more than the next monthly headline. First, whether the middle of the market starts moving, since a typical company spending $16.32 per employee suggests most businesses are still early. Second, whether customer support's climb from 0.8 to 13.3 percent of spending continues or plateaus once the easiest chatbot deployments are done. Third, whether falling token prices keep pace with rising usage, which would confirm that cheaper AI is expanding the market rather than just repricing existing demand.

For developers and decision-makers, the practical takeaway is that AI spending is growing fastest where it's already deep, and that most of the economy hasn't caught up yet. The gap between heavy adopters and everyone else is the number to track.