OpenAI plans to produce its first artificial intelligence chip in 2026. The company is partnering with TSMC and Broadcom to develop its own in-house AI chip, and will also begin using AMD chips alongside Nvidia to train its AI models.
According to Reuters, OpenAI initially abandoned plans to build a network of factories for chip manufacturing and instead focused on in-house chip design. The company has been working with Broadcom for several months to develop a special chip for controlling artificial intelligence models, which is expected to be launched in 2026.
In addition, OpenAI plans to use AMD chips through Microsoft's Azure cloud platform in the training process. The company previously relied mostly on Nvidia GPUs, but chip shortages, delays and high training costs have forced the company to explore alternatives.
