Cheaper AI alternatives are eroding the pricing power that underpins OpenAI and Anthropic's expected IPO valuations north of $800 billion.

Chinese AI labs are charging a fraction of American competitors for comparable work. Artificial Analysis benchmarked each lab's most capable model on the same 10 evaluations: Anthropic's Claude cost $4,811, OpenAI's ChatGPT $3,357, DeepSeek $1,071, Kimi $948, and Zhipu's GLM $544—making Claude nearly nine times more expensive than the cheapest Chinese alternative.

Enterprises are adopting an 'advisor model' strategy: a cheap open-source model handles routine tasks, and only calls out to frontier models from OpenAI or Anthropic when needed. On OpenRouter, a marketplace for AI model access, Chinese models grew from about 1% of usage in 2024 to more than 60% in May.

Google and other vendors are selling cost reduction as a product. Google CEO Sundar Pichai said that if large Google Cloud customers shifted 80% of their workloads from frontier models to Gemini 3.5 Flash, they would save more than $1 billion a year. Figma is selling features that cut customers' token consumption by 20 to 30%.