Token prices collapsing, regulation rising, AI's pricing power looks fragile
Falling token prices and increasing regulatory pressures are weakening AI companies' pricing power, raising concerns about the sustainability of current business models in the artificial intelligence sector.
Background
- "Token" in AI refers to the chunks of text (words, syllables, or characters) that language models process. Companies like OpenAI and Anthropic charge per token, making it the fundamental unit of AI pricing.
- The article examines how AI companies' pricing power is weakening: token prices are dropping due to competition (especially from cheaper open-source models like Meta's Llama) and rising regulatory pressure (e.g., EU AI Act, US executive orders).
- This matters because the AI industry has been valued largely on the assumption that companies can maintain high margins on API access. If token prices keep falling—commoditizing the core AI product—the business models of leading AI firms become fragile.
- Key context: The "OpenAI moment" in late 2022 sparked an investment boom, but the landscape has since fragmented. Open-source models and regulatory compliance costs are squeezing profits, raising questions about whether AI will follow the pattern of cloud computing (where margins eventually thinned) or something new.