OpenAI, Broadcom Announce Partnership to Jointly Design and Develop AI Chips

OpenAI and Broadcom announced a multi-year strategic collaboration on Monday. As part of the deal, the two companies will jointly design and develop chips and systems to power the world's growing artificial intelligence (AI) compute demands. The two companies will collaborate to develop 10 gigawatts of custom AI accelerators. The move also marks the San Francisco-based AI giant's second official foray into the hardware space, after the under-development AI device, which is being built in partnership with Jony Ive.

OpenAI to Design AI Chips

In a post, the ChatGPT maker announced a strategic partnership with Broadcom to develop 10GW worth of custom AI accelerators, highlighting that the two companies have signed a term sheet for the same. In this collaboration, OpenAI will be designing the chipsets that can handle heavy AI workflows, while Broadcom will fabricate and deploy them. The San Jose-based semiconductor giant will also provide Ethernet solutions to help the AI system scale.

OpenAI says that by designing its own chips and systems, it can apply its learnings from developing large language models (LLMs) and products such as the ChatGPT and Sora apps. Essentially, these systems will be optimised for the AI giant's workloads. Additionally, these systems will be deployed across the company's facilities and partner data centres to meet the compute demand of third parties as well.

“Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI's potential and deliver real benefits for people and businesses. Developing our own accelerators adds to the broader ecosystem of partners, all building the capacity required to push the frontier of AI to provide benefits to all humanity,” said Sam Altman, Co-Founder and CEO of OpenAI.

While the financial details of the deal were not disclosed, the post stated that Broadband will have exclusive rights to bring the racks and the Ethernet solutions for the AI chips. Interestingly, this marks the fourth partnership forged by OpenAI to scale the compute for its AI needs. Oracle, Nvidia, and AMD are the other partners.

OpenAI said that ChatGPT has grown to more than 800 million weekly active users, alongside the adoption of its services across enterprises, small businesses, and developers. The rapid growth has exponentially increased the requirement for more servers to handle the workload.