OpenAI, Broadcom Unveil Jalapeño Chip
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI and Broadcom unveiled Jalapeño, an LLM-optimized inference chip developed from design to manufacturing tape-out in nine months
- OpenAI's own AI models aided the chip's design process, marking OpenAI's first custom silicon
- The first-generation Jalapeño accelerator shows performance per watt substantially better than current state-of-the-art in early testing
- The chip targets 50% cheaper inference for large-scale LLM workloads, per Tech Times' coverage
- Tom's Hardware describes Jalapeño as a "massive reticle-sized ASIC" built in an ultra-fast nine-month development cycle
- Deployment is targeted for the end of 2026
Why it matters: OpenAI's first custom chip enters direct competition with Nvidia on inference economics, with a 50% cost reduction target and end-of-2026 deployment. The nine-month design-to-tape-out cycle, accelerated by OpenAI's own AI models, shows custom AI silicon can be built far faster than the multi-year cycles that have defined GPU generations.
