OpenAI and Broadcom unveiled Jalapeño, OpenAI's first custom AI inference chip, targeting 50% cheaper LLM inference with deployment at gigawatt-scale data centers by end-2026. Nine-month development cycle accelerated by OpenAI's own models.

OpenAI and Broadcom have unveiled Jalapeño, OpenAI's first Intelligence Processor, announced on June 24, 2026. The chip is purpose-built for LLM inference—the computational process where an AI model generates responses to users, executes coding tasks, and powers applications like ChatGPT, Codex, and APIs. Designed from scratch for modern LLM inference rather than repurposed from general-purpose silicon, Jalapeño is part of a wider compute plan involving Broadcom and Celestica. The first deployment is planned by the end of 2026, with expansion expected in subsequent years.

Engineering samples are already running machine learning workloads in the laboratory, including GPT-5.3-Codex-Spark. The co-development process, spanning from design to manufacturing tape-out, took nine months, with OpenAI's own models helping to accelerate portions of the design and optimization work.

OpenAI President and Co-Founder Greg Brockman stated: "Jalapeño is part of our long-term full-stack infrastructure strategy to make compute more abundant, resulting in AI which is faster, more reliable, more affordable for people and businesses, and can be used to solve more important problems."

Broadcom contributes silicon and networking expertise, while Celestica handles board, rack, and system integration. Broadcom CEO Hock Tan noted that the partnership will enable the deployment of "gigawatt scale data centers with Microsoft and other partners beginning in 2026."

For India's AI market, the chip holds significance because faster inference capabilities can directly influence the pricing, accessibility, and reliability of AI tools available to users and businesses.