Goldman Sachs Says Token Use Hits 120 Quadrillion Monthly. Nobody Knows What To Charge. Of Course They Don't.

Goldman Sachs forecasts token consumption will increase 24 times between 2026 and 2030, reaching 120 quadrillion tokens per month as companies deploy AI agents. Per-token costs have plummeted, but total consumption has skyrocketed, leaving buyers unable to predict costs and sellers unsure how to price. Oliver King-Smith, founder of smartR AI, notes smaller organizations can at least hide behind flat-fee personal subscriptions.

This illustrates what economists call the usage elasticity problem. When unit cost drops but volume explodes unpredictably, traditional pricing models break down. The lesson: agentic systems multiply consumption non-linearly because multiple AI agents interact and trigger each other. If you are buying AI services, assume your token bill will behave nothing like your current software subscriptions.

Goldman Sachs provided the token consumption forecast. Oliver King-Smith of smartR AI offered the flat-fee workaround for smaller organizations. The broader market of AI buyers and sellers remains, by their own admission, lost.