Overview
- The collaboration covers up to 10 gigawatts of OpenAI‑designed accelerators and full rack systems built on Broadcom’s Ethernet networking stack, with initial deployments slated for the second half of 2026 and rollout through 2029.
- Broadcom’s stock rose about 9%–10% in early trading following the announcement, while the companies did not disclose financial terms.
- OpenAI will design the chips and systems and work with Broadcom to develop and deploy them, a move the company says will improve efficiency and reduce compute costs.
- Broadcom executive Charlie Kawwas said OpenAI is not the previously referenced $10 billion mystery customer from September, clarifying market speculation.
- The deal adds to OpenAI’s recent multivendor commitments of roughly 33 GW across Nvidia, AMD, Oracle and others, compared with just over 2 GW in operation today, as analysts flag financing complexity and scale constraints as key risks.