AI at Work

Enterprise AI spending reaches $177 billion driven by agentic usage

Enterprise AI spending is projected to reach $177 billion in 2026, a fivefold increase from 2025, driven by agentic capabilities. Anthropic leads business API spend, while the market shifts from per-seat to per-token pricing models.

· 1 source · Enterprise AI spend

Menlo’s latest report, based on a survey of 400 U.S. enterprise decision makers responsible for $4 billion in AI spend, estimates enterprise AI spending reached almost $177 billion in 2026, up from $33 billion in 2025 and $1.7 billion in 2023. The growth rate accelerated to five times from 2025 to 2026, reversing a slowdown seen in 2025. Average spend per enterprise increased 4.7 times, while the share of enterprises using AI rose from 55% to 65%. Companies spent 2.5 times their budgeted amounts, as the market shifted from per-seat to per-token pricing models. Anthropic captured 55% of business API spend, up from 40% in 2025, while OpenAI recovered to 31% and Google fell to 9%. Frontier models accounted for 63% of spend, up from 35%, as buyers prioritized capability over cost. Open-source models represented 4% of spend but 19% of tokens, with Kimi dominating open-source spending at 61%. The report suggests that the acceleration in spend is driven by agentic capabilities and that the current concentration in frontier models is an early adoption phase that will eventually broaden to include more efficient models.