AI spending per employee falls 10% at top firms in August
Data from payments firm Ramp shows a slowdown in business AI adoption and a significant drop in per-employee spending among the heaviest users, as price
Business spending on artificial intelligence tools slowed in August, according to data from the payments company Ramp. The firm's survey of 70,000 companies shows 56% of its customers paid for AI products last month, a rise of just 0.4% from July.
This slowdown mirrors a pattern from last year. Ramp's data showed little growth in AI adoption between August and October 2025 before activity picked up again later in the year.
A Leading Indicator of Market Health
Ramp's figures may overstate overall business adoption due to its technology-focused client base. A separate U.S. Census Bureau survey updated on August 23 indicates only 22% of businesses report using AI. However, Ramp's direct spending data is considered a potential leading indicator for the market.
The August timing is significant, as many industries slow down for summer vacations. Ramp economist Ara Kharazian noted this could explain the lull but also pointed to other warning signs for companies reliant on AI usage revenue.
Price Cuts Drive Down Spending
A major shift was a nearly 10% decline in AI spending per employee among the top 1% of AI-using firms in Ramp's sample. Spending fell to $7,205 per employee. Kharazian linked this drop to falling token costs as leading AI labs cut prices.
Average token costs have declined sharply from a 2026 peak. The data shows a clear shift in pricing.
| Metric | Value |
|---|---|
| Average token cost (Aug 2026) | $0.68 per million tokens |
| Peak token cost (Mar 2026) | $1.15 per million tokens |
OpenAI and Anthropic have driven these price reductions. The data suggests the labs have not yet compensated for lower prices with sufficiently higher usage volume.
Customer Preference for Older, Cheaper Models
The same price incentives are influencing model choice. Kharazian said many customers are opting for older, less expensive models instead of the latest, most powerful frontier releases. Examples include OpenAI's ChatGPT 5.6-Terra and Anthropic's Sonnet.
This trend poses a potential problem for AI developers. Employees at frontier labs have indicated that a significant portion of a new model's massive training cost is recouped in the first weeks after its release. Slower adoption of new models threatens that financial dynamic.
Kharazian summarized the competitive effect. "We are showing that competition between OpenAI and Anthropic is making AI more accessible, and also driving the price down for companies-and not only driving the price down, but driving spend down at the top 1% of companies," he said.
Limited Impact from Open-Weight Models
Despite discussion about open-weight models challenging frontier labs, their direct impact on business spending remains limited. In August, only 6.4% of AI-spending businesses used model-serving or inference platforms. This share is growing steadily but is not yet large enough to influence broader business adoption trends.
The focus for AI labs is consequently shifting. Explaining the spending drop also clarifies the industry's intensified focus on winning over non-technical users with AI co-working tools.
For AI model-builders and large cloud providers with massive chip orders, the August slowdown could be a concerning sign. Kharazian offered a final perspective. "It depends on who you are in the market," he noted. "If your company is using AI, it's great."
