New
AI & RoboticsMore in AI & Robotics→
OpenAI Pushes Into Custom AI Hardware With Jalapeño
Key Takeaways
- OpenAI and Broadcom say Jalapeño is a custom accelerator built from scratch for LLM inference.
- The companies claim early tests showed substantially better performance per watt, though results have not been independently verified.
- OpenAI says the chip went from design to tape-out in nine months, with large-scale deployment planned for late 2026.
DE
DT Editorial Team··via the-decoder.com