Chinese open-weight AI models have turned inference into a commodity, questioning the sustainability of frontier labs’ 60%-80% markups, though OpenAI and Anthropic may face less pressure than benchmarks suggest. Jesse Zhang, co-founder of customer-service chatbot maker Decagon, argued that running open models requires technical finesse many corporations would rather pay to avoid, citing the adage “no one gets fired for buying IBM.” Most enterprises could build infrastructure in-house, but prefer handing off the problem to Anthropic or OpenAI. Data from OpenRouter, a unified API endpoint for inference providers, supports this: since mid-May, open-weight models’ share of token usage has climbed nearly one-third to 75%. However, in estimated dollars spent on the platform, closed-source models still dominate open-weights.