Topic
The per-token pricing model AI providers use to charge for model usage, with separate input and output rates, discounts and resale units.
Jefferies' Chris Wood says US hyperscalers' $695 billion of guided 2026 capex looks unsustainable as cheap Chinese open-source models win share. He blames a slump in the price of AI output, with demand intact, while guidance climbs toward $870 billion for 2027.
Reality
The largest token allowances in China go to developers, not shoppers. The real pressure on Western AI pricing sits in API list prices that run 60% to 90% below OpenAI and Anthropic.
Reality
The AI capex cycle rests on the labs capturing pricing power as compute demand grows. Ramp's corporate card data has the effective enterprise price of a million tokens moving the other way since March.
Reality
One trade report puts Claude's top tier at 10x to 50x DeepSeek's per-token price. Anthropic's answer, a cheaper tier plus a quarter off realized prices, defends volume by surrendering price.
Reality
Google put the same 3.7 Flash behind its API, Vertex, Gemini Enterprise and AI Mode in Search on August 13. The payoff is fewer evaluations to run, not a new capability tier.
Reality