Photo by Dzenina Lukac on Pexels

In a bid to secure its AI infrastructure and reduce dependence on Nvidia, OpenAI is collaborating with Broadcom to develop custom computer chips tailored for its data centers. This partnership aims to furnish OpenAI with the substantial computing power required for demanding applications like ChatGPT and Sora. The deal will see OpenAI, leveraging its own chip designs, deploying “10 gigawatts of custom AI accelerators.” Production is slated to commence in the latter half of 2026 and extend through 2029. This strategic move mirrors a wider industry trend where tech giants like Meta, Google, and Microsoft are increasingly investing in custom chip development to fortify their supply chains and mitigate reliance on singular providers.