OpenAI is said to be preparing to develop its inaugural AI chip in 2026.

OpenAI is said to be collaborating with TSMC and Broadcom to develop a proprietary AI chip, while also starting to utilize AMD chips in conjunction with Nvidia's for the training of its AI systems.

OpenAI has reportedly decided to halt its plans to create a network of chip manufacturing factories, at least for the time being, according to Reuters. The organization will instead concentrate on developing its own chip designs. For several months, OpenAI has been collaborating with Broadcom to produce an AI chip intended for model execution, which is anticipated to be available by 2026, as per the report.

In the interim, OpenAI intends to utilize AMD chips via Microsoft’s Azure cloud platform for the training of its models. Previously, the company predominantly depended on Nvidia GPUs for this purpose; however, challenges such as chip shortages, delays, and the escalating costs associated with training have prompted OpenAI to seek alternative solutions, as reported by Reuters.