Business spending on AI tools barely grew in August, according to new data from the payments company Ramp. The firm tracks spending across 70,000 companies, and its latest survey shows 56% of customers paid for AI products in August, up just 0.4% from the month before.
The slowdown is not unprecedented. Ramp’s AI index showed little to no growth between August and October last year, before adoption picked up again as the year closed. But the stakes are higher now. The massive investment in AI infrastructure by frontier labs and hyperscalers depends on steady revenue growth, and even a small dip in adoption can ripple through the market.
The per-employee slump at top firms
The most striking figure in Ramp’s data concerns the heaviest spenders. Among the top 1% of firms in the sample, AI spend per employee fell nearly 10% in August, dropping to $7,205.
The decline likely reflects two forces. The first is seasonal: August is when much of the industry is on vacation, and token usage tends to dip. The second is structural: token prices have fallen sharply as OpenAI and Anthropic cut costs.
Average token costs now sit at $0.68 per million tokens, down from the 2026 peak of $1.15 per million tokens in March. The price cuts have not yet been offset by growing volume, and the data suggests the labs have not made up the difference with increased usage.
Ramp economist Ara Kharazian has flagged other warning signs for companies that depend on token spend, including the decline at the top 1% of firms.
Cheaper models win the month
The price pressure is changing customer behavior. Many businesses are choosing older, cheaper models like OpenAI’s ChatGPT 5.6-Terra and Anthropic’s Sonnet instead of the more powerful frontier releases.
That choice matters for the labs. Employees at frontier labs have said much of the cost of training a new model is recouped in the first weeks of its release. Slower adoption of the newest models could threaten that dynamic.
Kharazian framed the shift in plain terms. “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,” he said.
A narrow slice of the market
Ramp’s figures may overstate overall adoption. Its clientele is tech-heavy, and a separate ongoing US Census Bureau survey updated on August 23 shows just 22% of businesses report using AI.
Ramp’s survey is not necessarily representative of the broader market, but it is one of the few direct spending data sets available, and it may serve as a leading indicator.
The data also shows limited movement in one area that has drawn attention: open-weight models and inference platforms. 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 the dynamics of broader business adoption.
What the slowdown means
The slowdown is a warning sign for companies that depend on token spend, but its meaning depends on where you sit.
For model-builders and hyperscalers with hundreds of billions of dollars in chips on order, a sustained slowdown would be bad news. Revenue growth is the assumption underneath those orders.
For businesses using AI, the picture is brighter. Falling prices and competitive pressure from OpenAI and Anthropic mean AI tools are more accessible.
Kharazian put it directly: “it depends on who you are in the market. If your company is using AI, it’s great.”
August at a glance
| Metric | August figure | Comparison |
|---|---|---|
| Share of Ramp customers paying for AI | 56% | Up 0.4% from July |
| AI spend per employee, top 1% of firms | $7,205 | Down nearly 10% from prior month |
| Average token cost | $0.68 per million tokens | Down from $1.15 peak in March 2026 |
| Businesses using AI (Census survey) | 22% | Updated August 23 |
| Businesses using model-serving platforms | 6.4% | Growing steadily |
The question is whether August was a seasonal blip or the start of a trend. Last year’s data suggests adoption can recover after a slow summer. But the price cuts are new, and the market has not yet shown whether volume will catch up to them.
The next few months will tell. If adoption accelerates in September and October, August will look like a vacation artifact.
For now, the data offers a mixed verdict. The AI buildout continues at a furious pace, but the spending that justifies it is growing more slowly than before. Whether that is a summer doldrum or a warning sign is a question the market will answer on its own.
Source: techcrunch.com
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.