OpenAI Turns Up the Heat With Jalapeño, Its First Custom AI Chip
Summary
OpenAI unveiled Jalapeño, its first custom AI chip, developed with Broadcom to speed up and lower the cost of large language model inference. The chip is designed specifically for chatbot and LLM workloads, marking a major step toward greater control of OpenAI’s hardware stack and reduced dependence on Nvidia. OpenAI says early versions are already being tested in its labs, including with GPT-5.3-Codex-Spark, and claims the chip offers more compute with less energy, though no benchmarks were released. Jalapeño is the first in a planned multi-generation compute platform that will begin deployment in data centers later this year, with future versions aimed at supporting gigawatt-scale AI infrastructure with Microsoft and other partners.
