Stocks● Neutral
AI price cuts lift OpenAI usage and revenue, TD Cowen data shows
OpenAI slashed GPT-5.6 Luna by 80%. TD Cowen data shows usage jumped 14-fold and revenue rose 34%, a pattern that echoes the Jevons Paradox of the 19th century.
Continue with
OpenAI slashed the price of its top-tier GPT-5.6 Luna model by 80% and cut its mid-range Terra model by 20% in late July. Two weeks later, usage of both models had jumped far more than prices had fallen.
TD Cowen analysts studied transaction data from OpenRouter, a service that lets developers route queries across different AI models. They found that the effective price per token for Luna – the amount developers actually pay after OpenAI's cut – fell roughly tenfold. Consumption rose about 14-fold.
Terra followed a similar pattern. Its effective per-token price fell about threefold. Usage climbed roughly fivefold.
The revenue math worked in both cases. TD Cowen estimated that OpenAI's Luna revenue increased about 34% in the two weeks after the cut, compared with the seven-day period before it. Terra revenue was up roughly 45%.
An 80% price reduction that still produces a revenue gain is unusual in most industries. The logic that explains it traces back to a 19th-century observation by economist William Stanley Jevons. He noticed that improvements in coal-burning efficiency did not reduce coal consumption. They drove it higher, because cheaper energy unlocked new uses.