The editor of Downcodes learned that OpenAI is actively deploying in the field of AI chips, striving to get rid of dependence on existing chip suppliers, reduce costs and improve efficiency. According to Reuters, OpenAI is working with TSMC and Broadcom to develop in-house AI chips and has begun using AMD and Nvidia chips for AI model training. It plans to launch its first self-developed AI chip in 2026. This move marks an important step for OpenAI in independent research and development of AI hardware.
OpenAI has been working with Broadcom for months to develop an AI chip for running models, which could be available as early as 2026. At the same time, OpenAI plans to use AMD chips for model training through Microsoft's Azure cloud platform.
OpenAI previously relied almost entirely on Nvidia GPUs for training, but chip shortages and delays, as well as high training costs, prompted OpenAI to explore alternatives. OpenAI abandoned plans to build a network of chip manufacturing factories and instead focused on in-house chip design.
OpenAI’s move reflects the AI field’s increasing emphasis on autonomous and controllable hardware, and also indicates that competition in the AI chip field will become more intense in the future. By independently developing chips, OpenAI is expected to further improve the performance and efficiency of its AI models and reduce operating costs. This is of great significance to the future development of OpenAI.