AI Spending Per Employee Falls 10% at Top Firms as August Adoption Stalls
Ramp data shows AI adoption barely grew in August as top firms cut per employee spend nearly 10%, raising revenue concerns for model labs and hyperscalers.
Summary
Ramp spending data from 70,000 companies shows 56% of customers bought AI products in August 2026, up only 0.4% from July. A similar plateau from August to October 2025 preceded renewed year end growth, while vacations may explain this August’s weakness. Ramp’s technology focused customer base likely overstates adoption, since a US Census Bureau survey updated August 23 found only 22% of businesses using AI, but its direct spending data may provide an early signal.
Among the top 1% of AI spending firms, expenditure per employee fell nearly 10% to $7,205. Average token prices dropped from their March 2026 peak of $1.15 per million tokens to $0.68 after OpenAI and Anthropic cuts, without enough volume growth to compensate. Customers increasingly favor cheaper older models, including OpenAI’s ChatGPT 5.6-Terra and Anthropic’s Sonnet, over frontier releases, potentially weakening labs’ ability to recoup training costs during a model’s first weeks. Only 6.4% of AI spending businesses used model serving or inference platforms in August, despite steady growth. Ramp economist Ara Kharazian said competition is improving affordability but reducing spending among companies expected to drive expansion, intensifying labs’ push for nontechnical users of AI coworking tools. Lower costs benefit customers but threaten model builders and hyperscalers with hundreds of billions of dollars in chips on order.
Positives
- 56% of Ramp customers purchased AI products in August 2026, still rising 0.4% from July despite the slowdown.
- Average token prices fell to $0.68 per million tokens from a March 2026 peak of $1.15.
- OpenAI and Anthropic competition is making AI cheaper and more accessible to business customers.
- 6.4% of AI spending businesses used model serving or inference platforms in August, a share that continues to grow.
- August to October 2025 stagnation was followed by renewed adoption growth at the end of that year.
Risks & concerns
- Top 1% AI spending firms cut expenditure per employee nearly 10% to $7,205 in August.
- AI labs have not offset lower token prices with sufficient growth in usage volume.
- Customers increasingly choose cheaper older models such as ChatGPT 5.6-Terra and Sonnet over frontier releases.
- Slower early adoption could impair frontier labs’ ability to recover new model training costs within the first weeks.
- Model builders and hyperscalers face weaker revenue signals while carrying hundreds of billions of dollars in chip orders.