OpenAI tipped to launch its first in-house chip by the year-end to take on Nvidia

OpenAI

OpenAI, the company behind ChatGPT, is poised to enter the custom component business in 2025 and reduce the dependence on Nvidia by developing its own artificial intelligence (AI) silicon chip. Moreover, the company aims to finalise the chip’s design within the next few months and subsequently send its design for fabrication to Taiwan Semiconductor Manufacturing Co. (TSMC), utilising its 3-nanometer process technology.

According to a report from Reuters, the initial version of this custom chipset from OpenAI is expected to focus primarily on running AI models and will be integrated into OpenAI’s infrastructure on a limited scale. TSMC has been named as the manufacturing partner, and insiders indicate that the company is in the final stages of design, with plans to commence manufacturing in the coming months.

The report adds that the company also reportedly plans to develop future versions of the chip with more advanced processors and capabilities. The process of giving the first design of a chipset to the chip factory, which is called ‘Tapping Out,’ costs tens of millions of dollars, and it will also take more than six months to produce a finished chip. However, the artificial intelligence giant plans to invest heavily in this project to accelerate production and manufacturing.

OpenAI's plan to send its design to TSMC this year shows how the company wants to end its dependence on Nvidia or other AI chip manufacturers. Moreover, this processor is being designed by an in-house team at OpenAI, led by Richard Ho, who previously led Google's custom AI chip program, and supported by Broadcom.