Post by @openai

Designed from the ground up by OpenAI and brought to production with @Broadcom, Jalapeño is purpose-built for the LLM workloads powering ChatGPT, Codex, the API, and future agentic products. While OpenAI is still measuring final performance, early testing shows that Jalapeño will deliver performance per watt substantially better than current state-of-the-art. We'll release a detailed technical report once testing is complete. Jalapeño was co-developed from initial design to manufacturing tape-out in just nine months—the fastest ASIC development cycle ever achieved in high-performance advanced semiconductors. The same OpenAI models served to users today were used to accelerate parts of the design and optimization process. We believe if AI can help engineers design better chips faster, it will lower the cost of compute and help democratize access to advanced AI. We believe that chips are foundational to the AI economy, and that making compute more abundant will result in faster and more reliable AI that can be used to solve more problems. Building our own expands our full-stack platform from products to models to infrastructure, and will help us scale intelligence, serve more people, and expand access to AI.