OpenAI and Broadcom have officially announced a strategic collaboration aimed at deploying 10 gigawatts of computing power based on artificial intelligence accelerators designed by OpenAI itself. This multi-year agreement involves the joint development of next-generation systems and Ethernet solutions.
The partnership targets the creation of scalable and energy-efficient infrastructure, which is expected to be fully ready by 2029. The primary focus of the collaboration lies in the hardware necessary to support future AI models.
This information is based exclusively on a statement from the publisher OpenAI News. At present, there are no independent confirmations or additional sources detailing the technical specifications or financial terms of the deal.
editorial commentary
Why it matters
A likely consequence will be increased dependence of the AI market on alliances between model developers and chipmakers. The next observable signals will be announcements of specific chip architectures or the start of data center construction. The main uncertainties are created by implementation timelines and the possibility of overcoming physical power consumption constraints by 2029.