OpenAI and Broadcom unveiled their first jointly developed AI chip, called Jalapeño, a processor built for inference that early samples show running at roughly 50% lower cost than typical AI GPUs, according to Broadcom CEO Hock Tan. OpenAI said samples are running in its labs at target power and performance with its GPT-5.3-Codex-Spark model, with the chip designed in about nine months, aided partly by AI tools, before going to TSMC for manufacturing. The companies are targeting initial deployment by the end of 2026, with finished chips integrated into data centers run by Microsoft and other partners and server systems built by Celestica. Tan placed Jalapeño on par with Nvidia’s Blackwell line and Google’s TPUs, and said demand has grown strong enough that his earlier deployment projections may prove conservative.
