OpenAI, one of the tech world’s most powerful players, is extending its AI strength to the hardware side. The company has taken a significant step toward designing and manufacturing its own AI chips. This initiative is crucial for reducing dependence on giants like Nvidia and controlling operational costs.

OpenAI is preparing to produce its own custom chip

OpenAI will launch this new chip in 2026. The company collaborated with semiconductor giant Broadcom on this project. According to financial sources, OpenAI will not sell these chips to an external customer. The chips will be used solely to train and run the company’s own AI models, such as ChatGPT.

This strategic move has actually been something OpenAI has been working on for months. The company, which was previously rumored to be working with Broadcom and TSMC on a custom chip, is now utilizing AMD chips in addition to Nvidia to meet its growing infrastructure needs.

Broadcom CEO Hock Tan announced that they received an AI infrastructure order exceeding $10 billion from an undisclosed customer in the last quarter. Tan stated that their revenue from this area will increase significantly in fiscal 2026. Industry analysts are certain that this mysterious customer is OpenAI.

Large technology companies such as Google, Amazon, and Meta are also taking similar steps. The rapid rise in AI training and operating costs is pushing tech giants to develop their own hardware solutions. OpenAI’s move stands out as a prime example of companies seeking in-house solutions to address rising costs.