The AI industry has shifted from a subsidized model to a "token shortage" era. This forces all companies, from AI providers to enterprise users like Uber, to prioritize cost-effective usage. Business models are now usage-based, making architectural and financial efficiency paramount.

The narrative of "off the charts" AI demand is misleading. Major AI providers like OpenAI are "burning tens of billions of dollars," indicating they are not charging the true cost for their services. A realistic picture of demand will only emerge once they are forced to price for profitability, which could significantly cool the market.

The current model of paying per AI token is a temporary phase. Drawing a parallel to computing history, any resource constraint that requires payment eventually moves to the user's local device and becomes free. On-device AI processing will follow this pattern, ultimately eliminating token costs.

Despite enterprises hitting AI budget limits, the market is not collapsing. Competition is forcing AI providers to lower token prices, triggering the Jevons paradox: as a resource's cost falls, its consumption increases, sustaining demand for underlying infrastructure like NVIDIA chips.

The cost of AI, priced in "tokens by the drink," is falling dramatically. All inputs are on a downward cost curve, leading to a hyper-deflationary effect on the price of intelligence. This, in turn, fuels massive demand elasticity as more use cases become economically viable.

Anthropic is ending subsidized token usage for third-party tools, reflecting a market shift from seat-based to usage-based pricing. This move is a direct consequence of compute demand exceeding supply, ending a brief 'golden age' of cheap, large-scale experimentation for developers.

The "golden age" of cheap, plentiful AI experimentation is over due to token shortages and high costs. This new "trade-offs era" forces companies to justify AI expenses, which slows the pace of human replacement, buys time for adaptation, and forces the market toward more sustainable, realistic pricing models.

Countering the narrative of insurmountable training costs, Jensen Huang argues that architectural, algorithmic, and computing stack innovations are driving down AI costs far faster than Moore's Law. He predicts a billion-fold cost reduction for token generation within a decade.

The current affordability of AI tokens is not sustainable; it's propped up by venture capital funding AI companies operating at a loss. Businesses should treat this as a temporary window for aggressive learning and experimentation before prices inevitably rise to reflect true operational costs.