Broadcom’s Custom Chip for OpenAI Signals Nvidia’s Inference Dominance Is Ending
OpenAI has deployed “Jalapeño,” a custom Broadcom-manufactured chip designed to efficiently run inference for ChatGPT and other products, with broader deployment targeted for end of 2026. The move signals that major AI labs are treating chip dependency as a structural risk, following patterns set by Google and Amazon while reducing reliance on Nvidia.









