Menlo Ventures Finds Agents Driving $177B AI Spend

The report’s strongest signal is that enterprise AI is shifting from seat-based software to work-based consumption, with agents turning token usage into the new unit of demand.

• –AI spending is scaling far faster than adoption, suggesting power users and autonomous workloads—not broad employee rollout—are driving the market.
• –Frontier labs capture the dollars, but open-source models capture 19% of usage, creating a meaningful opportunity for lower-cost inference and specialized deployment.
• –Coding remains the clearest AI-native category, while established software markets are seeing AI layered onto incumbent systems rather than fully displaced.
• –Developers should treat token budgets, caching, observability, and agent workflow controls as core production infrastructure.
• –The $177 billion figure is an annualized estimate based on survey data and bottom-up modeling, so its direction is more reliable than its precision.