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.
Related Insights
The massive capital expenditure by hyperscalers on AI will likely create an oversupply of capacity. This will crash prices, creating a golden opportunity for a new generation of companies to build innovative applications on cheap AI, much like Amazon utilized the cheap bandwidth left after the dot-com bust.
The cost for a given level of AI performance halves every 3.5 months—a rate 10 times faster than Moore's Law. This exponential improvement means entrepreneurs should pursue ideas that seem financially or computationally unfeasible today, as they will likely become practical within 12-24 months.
AI companies operate under the assumption that LLM prices will trend towards zero. This strategic bet means they intentionally de-prioritize heavy investment in cost optimization today, focusing instead on capturing the market and building features, confident that future, cheaper models will solve their margin problems for them.
The cost for a given level of AI capability has decreased by a factor of 100 in just one year. This radical deflation in the price of intelligence requires a complete rethinking of business models and future strategies, as intelligence becomes an abundant, cheap commodity.
The comparison of the AI hardware buildout to the dot-com "dark fiber" bubble is flawed because there are no "dark GPUs"—all compute is being used. As hardware efficiency improves and token costs fall (Jevons paradox), it will unlock countless new AI applications, ensuring that demand continues to absorb all available supply.
While the per-unit cost of using AI has plummeted, total enterprise spending has soared. This is a classic example of the Jevons paradox: efficiency gains and lower prices are unlocking entirely new use cases that were previously uneconomical, leading to a net increase in overall consumption and total expenditure.
Even for complex, multi-hour tasks requiring millions of tokens, current AI agents are at least an order of magnitude cheaper than paying a human with relevant expertise. This significant cost advantage suggests that economic viability will not be a near-term bottleneck for deploying AI on increasingly sophisticated tasks.
Big tech companies are offering their most advanced AI models via a "tokens by the drink" pricing model. This is incredible for startups, as it provides access to the world's most magical technology on a usage basis, allowing them to get started and scale without massive upfront capital investment.
As AI gets exponentially smarter, it will solve major problems in power, chip efficiency, and labor, driving down costs across the economy. This extreme efficiency creates a powerful deflationary force, which is a greater long-term macroeconomic risk than the current AI investment bubble popping.
Arvind Krishna forecasts a 1000x drop in AI compute costs over five years. This won't just come from better chips (a 10x gain). It will be compounded by new processor architectures (another 10x) and major software optimizations like model compression and quantization (a final 10x).
