The adoption of artificial intelligence (AI) tools by businesses appeared to slow in August, according to spending data collected from 70,000 companies by Ramp, a payments company. The data shows that 56% of Ramp's customers spent on AI products in August, a rise of just 0.4% from the previous month. This marks the second time Ramp's metrics have indicated a slowdown in AI adoption, with similar trends observed earlier this year. However, given the rapid pace of AI development, even minor declines can raise concerns about future growth. The significant investment in AI infrastructure by leading tech companies, often called "frontier labs" and "hyperscalers," depends on the assumption that there is enough revenue to justify these costs. Usage of AI tools has grown sharply, especially among software engineers using AI-powered coding assistants. However, if adoption slows, revenue could follow suit. Ramp’s data may overstate the overall trend because its customer base tends to be more tech-savvy. A separate, ongoing survey by the U.S. Census Bureau, updated in late August, found that only 22% of businesses report using AI. While Ramp's data isn't fully representative of the entire market, it remains one of the few direct sources of spending data and could serve as an early indicator of trends. The August data comes from a time when much of the industry is on vacation, which may explain the slower pace of spending. However, Ramp economist Ara Kharazian points to other concerning trends. For example, AI spending per employee in the top 1% of firms in his sample dropped nearly 10% to $7,205. This could be due to lower costs from companies like OpenAI and Anthropic, which have reduced the price of using AI models. Average token costs have fallen to $0.68 per million tokens, compared to a peak of $1.15 in March 2026. The data suggests that these price cuts have not yet been offset by increased usage. Despite the lower costs, many businesses are opting for older, cheaper AI models like OpenAI's ChatGPT 5.6-Terra and Anthropic's Sonnet instead of newer, more advanced tools. Frontier labs, which develop cutting-edge AI models, have previously seen significant returns in the early weeks after launching new models. However, if adoption continues to slow, this financial pattern could be disrupted. While some companies are exploring open-source AI models, only 6.4% of AI-spending businesses used model-serving or inference platforms in August, a number that is growing but not fast enough to drive broader adoption. Kharazian noted that the competition between OpenAI and Anthropic has made AI more accessible and reduced costs for businesses, particularly for the top 1% of spenders. This shift could be a concern for companies with massive investments in AI infrastructure, but it could be a positive sign for businesses that are already using AI. “It depends on who you are in the market,” Kharazian said. “If your company is using AI, it's great.”