OpenAI Nears Completion of First Custom AI Chip Design

The company aims to reduce reliance on Nvidia by developing in-house AI chips, with plans to finalize the design and begin production by 2026.

OpenAI is finalizing the design of its first custom AI chip, which could be sent for manufacturing in the next few months.
The chip is being developed in collaboration with Broadcom and may be fabricated using TSMC's advanced 3-nanometer process technology.
If successful, the chip will initially have a limited role in running AI models, with potential future use in training AI systems.
The project seeks to decrease OpenAI's dependence on Nvidia's AI hardware and provide the company with greater control over its AI infrastructure.
Experts note the ambitious timeline and high costs involved, with mass production potentially beginning in 2026 if initial tests succeed.