According to Reuters, the company is expected to send the chips to TSMC for fabrication shortly as it aims for mass production in 2026. The move forms part of the company’s efforts to become less reliant on chips from NVIDIA.
The report suggests that TSMC will be producing OpenAI’s AI chip on its 3nm process, featuring a systolic array, HBM, and strong networking – very similar to NVIDIA’s design.
While neither OpenAI and TSMC have commented on these reports, sources cited by Reuters have suggested that OpenAI’s AI chip, though capable of training and running models, will initially have a limited role. Rather its development by OpenAI is seen as being a strategic move to gain leverage over suppliers.
OpenAI’s in-house team, led by ex-Google chip expert Richard Ho, has doubled to 40 members and is co-developing the AI chip with Broadcom, according to Reuters.
OpenAI joins the likes of Microsoft and Meta in developing in-house chips which is seen as a response to soaring costs and their over-reliance on NVIDIA.
