Overview
- OpenAI and Broadcom signed a multi‑year term sheet to develop and deploy racks of OpenAI‑designed accelerators integrated with Broadcom’s Ethernet networking.
- Broadcom will handle development, manufacturing, and large‑scale deployment, with installations planned across OpenAI’s facilities and partner data centers.
- Rollouts are scheduled to begin in the second half of 2026 and conclude by the end of 2029, targeting a total of 10 gigawatts of AI capacity.
- Broadcom shares jumped roughly 10–12%, adding about $150 billion in market value after the announcement, while financial terms were not disclosed.
- OpenAI president Greg Brockman said the company used its own models to discover chip design optimizations that cut area and accelerated the design timeline, as analysts flag supply, financing, and energy constraints for a 10 GW buildout.