Payments company Ramp, which tracks spending across roughly 70,000 companies, reported that business adoption of AI tools slowed in August. Ramp’s latest data shows 56% of its customers paid for AI products in August, an increase of just 0.4 percentage points from July.
Past patterns and possible explanations
This is not the first time Ramp’s metrics have indicated a slowdown: the company’s AI index showed little to no adoption growth between August and October last year, before picking up again later in the year. The August data may partly reflect seasonality, since many firms and employees take vacations during the month.
Why the slowdown matters
Large investments in AI infrastructure by frontier labs and hyperscalers rest on expectations of substantial revenue to justify those costs. Usage has grown rapidly so far—especially as software engineers adopted agentic coding tools—but if adoption eases, associated revenue growth could also slow.
Ramp’s sample and broader measures
Ramp’s clientele is skewed toward tech-active businesses, so its 56% figure probably overstates overall market adoption. For comparison, a US Census Bureau ongoing survey of AI adoption, updated on August 23, shows only 22% of businesses reporting AI use. Nevertheless, Ramp’s dataset is one of the few direct spending indicators available and may act as a leading signal.
Token costs, price cuts and the top 1%
Ara Kharazian, an economist at Ramp, highlighted a notable warning: AI spending per employee among the top 1% of firms in Ramp’s sample fell nearly 10%, to $7,205. That drop could reflect vacation-related lower activity, but it also aligns with falling token costs. After price cuts from OpenAI and Anthropic, average token costs have dropped to $0.68 per million tokens, compared with a 2026 peak of $1.15 per million tokens in March.
Ramp’s data suggest labs have not yet offset those price cuts with higher usage. Many customers are choosing older, cheaper models—such as OpenAI’s ChatGPT 5.6-Terra and Anthropic’s Sonnet—rather than the newest frontier releases. Employees at frontier labs have said much of a new model’s training cost is recovered in the first weeks after release, so slower adoption could disrupt that recovery dynamic.
Model-serving platforms remain a small but growing share
Despite talk that open-weight models could threaten frontier labs, only 6.4% of AI-spending businesses used model-serving or inference platforms in August. That share is steadily rising but not yet large enough to drive broad business adoption dynamics.
Kharazian’s takeaway
“We are showing that competition between OpenAI and Anthropic is making AI more accessible, and also driving the price down for companies—and not just driving the price down, but driving spend down at the top 1% of companies that previously the market was expecting to drive much of the growth going forward,” Ara Kharazian said.
For model-builders and hyperscalers with large hardware commitments, the August dip could be alarming if it signals a longer-term slowdown. But as Kharazian noted, the impact depends on your position in the market: for companies already using AI, lower prices and wider accessibility are beneficial, while for those relying on high early spending to recoup heavy infrastructure costs, the trend could be problematic.
Conclusion
Ramp’s August spending data point to a pause in AI spending growth, which may be partly seasonal. Still, the decline in token prices and reduced spending among the top-spending firms could have meaningful implications for revenue expectations at frontier labs and hyperscalers if the trend persists.
