OpenAI and Broadcom unveil LLM-optimized inference chip
OpenAI Blog · 2026-06-24
OpenAI and Broadcom unveiled Jalapeño, OpenAI's first custom LLM inference chip co-developed in nine months, delivering substantially better performance per watt than current accelerators, with gigawatt-scale deployment planned for end of 2026.
Appears in
Extraction
Topics: custom-siliconllm-inferenceopenai-infrastructureai-chips
Claims
- Jalapeño is OpenAI's first custom Intelligence Processor, designed from scratch for LLM inference rather than adapted from a general-purpose accelerator.
- Early testing shows Jalapeño delivers substantially better performance per watt than current state-of-the-art AI accelerators.
- OpenAI and Broadcom co-developed the chip from design to tape-out in nine months, which they claim is the fastest ASIC development cycle ever achieved in high-performance advanced semiconductors.
- OpenAI used its own AI models to accelerate parts of the chip design and optimization process.
- Jalapeño is the first in a multi-generation platform targeting gigawatt-scale data center deployment with Microsoft and other partners beginning in 2026.
Key quotes
"The world is moving to a compute-powered economy. Jalapeño is part of our long-term full-stack infrastructure strategy to make compute more abundant." — Greg Brockman, OpenAI President
"Jalapeño was designed from the ground up for LLM inference using detailed insights from our close collaboration with OpenAI researchers." — Richard Ho, OpenAI hardware lead
"By co-developing our industry-leading silicon directly with OpenAI, we are enabling the deployment of gigawatt scale data centers with Microsoft and other partners beginning in 2026." — Hock Tan, Broadcom CEO