OpenAI to Develop Custom AI Chips with Broadcom
OpenAI has entered a multi-year partnership with semiconductor manufacturer Broadcom to create custom-designed AI accelerator chips. This strategic move aims to build the specialised infrastructure necessary to power future advancements in artificial intelligence.
Under the agreement, OpenAI will be responsible for the design of the new AI accelerators and systems. Broadcom will then handle the development and deployment of this hardware. The project’s goal is to deploy 10 gigawatts of these custom systems, a significant undertaking in computing infrastructure.
The timeline for this initiative is set to begin in the second half of 2026, with the full deployment expected to be completed by the end of 2029. The new systems will be installed in data centres operated by OpenAI and its partners. To connect these powerful accelerators, the infrastructure will rely entirely on Broadcom’s networking technology, including its Ethernet, PCIe, and optical connectivity solutions.
This collaboration reflects a broader industry trend where major AI companies are designing their own hardware to gain a competitive edge. By creating custom chips, OpenAI can tailor the hardware specifically to the needs of its advanced AI models, potentially leading to greater efficiency and new capabilities. The companies have indicated that this partnership is a critical step toward building the capacity required to develop artificial general intelligence (AGI).
