Goldman Sachs Predicts Agentic AI Token Consumption to Surge 24x by 2030

Goldman Sachs has released a report forecasting massive growth in agentic AI over the next 15 years. The investment bank predicts that token consumption will increase 24 times by 2030 compared to 2026 levels. By 2040, agentic AI alone could drive a 55-fold increase in token consumption.

Token consumption to grow 24 times by 2030

The report estimates that AI queries will rise from 5 billion in 2025 to 23 billion in 2030. By that year, 30 percent of all queries are expected to be agentic. The bank also notes that costs per token computation are dropping by 60 to 70 percent annually, driven by new chip designs.

Goldman Sachs warns that poor data quality could lead to wasted resources as agentic AI scales. The bank's analysis highlights the importance of reliable data to avoid inefficiencies in the growing AI ecosystem.

The report comes as the AI industry rapidly expands, with major investments in infrastructure and hardware. Goldman Sachs' projections underscore the potential for agentic AI to reshape computing demands over the coming decades.