OpenAI Launches Jalapeño AI Chip With Broadcom, Already Running GPT-5.3-Codex-Spark Workloads

OpenAI unveils its first-generation AI inference chip, Jalapeño, developed with Broadcom, already powering GPT-5.3-Codex-Spark workloads with superior performance per watt, reducing reliance on Nvidia while positioning OpenAI as a major player across the entire AI hardware and datacenter stack.

• OpenAI unveils Jalapeño, its first-generation AI inference chip developed with Broadcom, which is already running machine learning workloads including GPT-5.3-Codex-Spark and delivers performance per watt that is substantially better than current state-of-the-art.
• Jalapeño is designed to work with all LLMs, reduces reliance on Nvidia, and is built to limit data movement while balancing compute, memory, and networking — positioning OpenAI as a potential player in the broader datacenter hardware market.
• The chip is part of OpenAI's broader full-stack strategy, which also includes Guaranteed Capacity compute access and an open-source GPU networking standard, embedding the company into every layer of the AI industry and making it a foundational cornerstone of AI innovation.