OpenAI and Broadcom Forge Custom AI Chip Alliance OpenAI is partnering with Broadcom to co-develop custom AI…

OpenAI and Broadcom Forge Custom AI Chip Alliance OpenAI is partnering with Broadcom to co-develop custom AI accelerators targeting roughly 10 gigawatts of compute capacity, with initial deployments expected in 2026 to ease GPU bottlenecks and lower inference costs. OpenAI will design the chips and systems while Broadcom handles development and manufacturing, leveraging its custom ASIC and networking expertise and an order reportedly around $10 billion. The move mirrors strategies by Google, Amazon, Microsoft, and Meta to integrate hardware-software stacks, aiming for efficiency and supply assurance rather than replacing Nvidia for training workloads. Broadcom’s shares surged on the news amid rising AI infrastructure spending