OpenAI Partners with Broadcom to Develop Custom AI Processors, Aiming to Transition from Nvidia GPUs, Report Says

OpenAI’s ChatGPT, backed by Nvidia’s powerful GPUs, shows no signs of disappearing. To cut costs, OpenAI reportedly plans to collaborate with Broadcom to design proprietary AI chips for internal use starting next year. This partnership is underscored by Broadcom CEO Hock Tan’s announcement of securing $10 billion in AI system orders from a “new customer,” hinting at significant revenue growth by 2026. Currently, OpenAI heavily relies on Nvidia for model training, contributing to Nvidia’s $115.2 billion revenue in the last fiscal year. As competition grows, companies like Amazon and Google have developed in-house components to reduce chip expenses, a strategy OpenAI aims to emulate with Broadcom’s expertise. However, OpenAI will remain dependent on Nvidia GPUs for its existing software infrastructure, indicating a prolonged reliance on Nvidia as it transitions to a new hardware platform. This shift offers little hope for budget-conscious gamers amid rising chip costs.

Source link