Payments company Ramp, which tracks spending data at 70,000 companies, reports that business adoption of AI products slowed in August. 56% of Ramp customers paid for AI products in August, up only 0.4% from the month before. Ramp‘s AI index saw a similar lull last year, with little to no growth between August and October before picking up again. The article cautions that even small slowdowns matter: frontier labs and hyperscalers are betting huge amounts of AI infrastructure investment on enough future revenue to pay it back. Usage has grown steeply so far, especially as software engineers adopted agentic coding tools, but slower adoption would likely slow revenue too.

Ramp‘s figures may overstate overall adoption because its customer base is tech-heavy. An ongoing US Census Bureau survey updated on August 23 shows only 22% of businesses report using AI. Ramp‘s data is not necessarily representative of the market, but the piece describes it as one of the few direct spending data sets available and potentially a leading indicator. August vacations could explain the doldrums, but Ramp economist Ara Kharazian flags other warning signs. AI spend per employee in the top 1% of firms fell nearly 10% to $7,205. That may reflect vacations, but it also reflects falling token costs: average token costs have declined to $0.68 per million tokens, compared with a peak of $1.15 per million tokens in March. OpenAI and Anthropic have cut prices, and the labs have yet to make up for the cuts with growing volume.

The same incentives are pushing customers toward older, cheaper models such as OpenAI‘s ChatGPT 5.6-Terra and Anthropic‘s Sonnet instead of newer frontier releases. Employees at frontier labs have said much of the cost of training is recouped in the first weeks of a new model’s release, and slower adoption could threaten that dynamic. For all the talk of open-weight models threatening the frontier labs, only 6.4% of AI-spending businesses used model-serving or inference platforms in August; that share is growing steadily but not fast enough to drive broader adoption dynamics.

Kharazian said the data shows competition between OpenAI and Anthropic making AI more accessible and driving price down, while also driving spend down at the top 1% of companies that the market expected to drive much of the growth going forward. That helps explain why AI labs are focusing on winning over non-technical users for AI co-working tools. The article calls the August data point a possible blip, but says it could be a bad sign for model-builders or hyperscalers with large chip orders. For companies using AI, Kharazian notes, lower prices are good: “it depends on who you are in the market. If your company is using AI, it’s great.”