AI models have reached commodity status. Token prices fell 41% while usage surged 50% since July, and the strategic consequences are becoming visible. OpenAI cut prices on Luna by more than 80% to win share; frontier models’ share of tokens slipped from 53% in August into the mid-40s; open-source demand shifted from frontier-dominated to medium- and small-tier models; spending among the top 1% of adopters fell 9.7% in August to $7,205 per employee per month; and decision models like TypeSafe‘s Jev are winning LLM calls at $0.04 per million input tokens—roughly 75x cheaper than standard LLMs for routing, gating, and classification.

Given this environment, the author recommends three moves. First, partner and resell: OpenAI partnered with Baseten to resell open-weight models, and Elon tweeted that the Grok bot would use the best model for a task, not just SpaceXSI models. Reselling creates a high-margin revenue stream by collecting a toll on products without bearing the cost to serve, since no company can provide the best model for every use case. Second, control the user interface: distribution becomes the moat. Labs should build token-hungry harnesses with memorable mononyms like Dots, Bot, and Muse, while providing the best models for the job so users stay; this makes ads and commerce an important B2C revenue driver. Third, capture user data for intelligent routing and training infrastructure: reselling other models while routing tasks across them aggregates information useful for training future models. Winning relative share becomes the only game that matters when aggregate dollar spend cools while token volume compounds.

OpenAI‘s recent surge toward a roughly $70 billion run rate highlights the dynamic: aggressive price cuts and marketplace aggregation let it recapture share within a boat length of Anthropic. In a commoditizing market, the spoils do not go to the lab with a marginal benchmark lead, but to the platform that aggregates volume.