OpenAI and Broadcom unveil Jalapeño LLM inference chip

OpenAI and Broadcom on June 24, 2026, unveiled Jalapeño, OpenAI's first custom accelerator for large language model inference, the companies said in an announcement posted on OpenAI's website. The chip, described as OpenAI's first Intelligence Processor, was architected around OpenAI's vision for the future of LLM inference and developed from design to production in nine months. Early testing shows that the first-generation accelerator will deliver performance per watt substantially better than current state-of-the-art. It was delivered to OpenAI CEO Sam Altman and President Greg Brockman by Broadcom President and CEO Hock Tan and President Charlie Kawwas. The accelerator was developed with partners Broadcom and Celestica. Engineering samples are running ML workloads in the lab at production target frequency and power, including GPT-5.3-Codex-Spark. Jalapeño is the first AI accelerator in a multi-generation compute platform the companies are building together for deployment at gigawatt scale with data center partners, the announcement said.