AI spend per employee at top firms fell 10% in August: Ramp

The slowdown in AI adoption metrics from Ramp offers a counterpoint to the relentless infrastructure buildout by frontier labs. These companies are investing heavily based on expectations of rising revenue, but the data suggests that price cuts are not yet being compensated by increased volume. The falling cost of tokens has made AI more accessible, but it also means that revenue per user may be shrinking. This could complicate the business case for expensive frontier models if customers continue to opt for cheaper, older versions.

The pattern is not entirely new; Ramp's index showed a similar plateau last year between August and October. However, the current context is different, with a greater emphasis on price competition between major providers like OpenAI and Anthropic. The data also hints that the market may be relying on a few high-spending firms to drive growth, and their recent pullback is a potential warning.

The future will depend on whether adoption accelerates again, as it did last year. For model-builders and hyperscalers, the stakes are high—if the slowdown persists, revenue might not justify the massive investments. For businesses using AI, the current environment is favorable, with lower costs and more options, making this a period of opportunity.