OpenAI Is Taking the “Crack Cocaine” Approach to Pricing
TL;DR
Futurism frames OpenAI’s pricing shift as a break from the flat-rate promise: cheap or broad access first, then possible billing based on actual token and compute consumption. The piece cites audited figures saying OpenAI’s net loss rose from 5 billion dollars in 2024 to 39 billion dollars in 2025, while noting that some of the loss may be accounting-related. Sam Altman recently described AI as a utility like electricity or water, implying users may eventually pay through a meter rather than a simple monthly plan.
Nauti's Take
The crack metaphor is deliberately ugly and overstated, but the business point lands: AI was sold as unlimited magic even though it behaves like infrastructure with recurring marginal costs. OpenAI helped create that misunderstanding through broad flat-rate access.
Metered pricing is the reality check, and for many teams it will matter more than the next model benchmark.
Briefingshow
This hits the core economics of AI: users have been trained to expect flat-rate access while providers still pay real compute costs per request. If those subsidies fade, the impact is not just higher prices. Experiments, agent workflows and automations become budget decisions instead of background habits.
Sources