OpenAI and Broadcom Unveil 'Jalapeño,' a Custom AI Chip Built for LLM Inference That Breaks Speed Records in Development

OpenAI and Broadcom unveil 'Jalapeño,' a custom AI accelerator chip purpose-built for LLM inference that reportedly breaks speed records with a nine-month development cycle and delivers superior performance per watt, with deployment planned by end of 2026.

• OpenAI and Broadcom unveil 'Jalapeño,' OpenAI's first custom AI accelerator chip, purpose-built from the ground up for LLM inference and designed to deliver performance per watt substantially better than current state-of-the-art hardware.
• The chip goes from initial design to manufacturing tape-out in just nine months, marking what is believed to be the fastest ASIC development cycle ever achieved in high-performance advanced semiconductors, with OpenAI's own AI models helping accelerate parts of the design process.
• Jalapeño is the first step in a multi-generation compute platform set for initial deployment by end of 2026, with partners Broadcom and Celestica, aiming to make AI inference faster, cheaper, and more accessible for users, developers, and businesses worldwide.