OpenAI, Broadcom unveil chip to run models faster, cheaper
OpenAI and Broadcom have unveiled their first custom AI chip, named “Jalapeno,” designed to run AI models faster and at significantly lower cost. Early tests suggest around 50% cost savings compared to traditional GPUs. The chips, developed to reduce reliance on Nvidia, will be deployed in large data centres operated with Microsoft and other partners later this year, marking a major step in OpenAI’s strategy to build dedicated AI hardware infrastructure.
