Agentic AI Sends Enterprise Token Costs Soaring

A Futurum report sponsored by QumulusAI finds that agentic AI drives token consumption 10 to 100 times higher than simple inference calls, creating budget crises for enterprises. Futurum forecasts agent and reasoning inference will grow 219% this year, with total inference spending rising from $120 billion in 2025 to $885 billion by 2030. A survey of 824 AI decision-makers shows 66% of AI compute already runs on reserved or owned infrastructure, not on-demand cloud. Unpredictable costs have caused some organizations to abandon useful AI tools entirely, a governance failure that risks slowing broader adoption.