OpenAI is Reportedly Working on its First In-House AI Chip

OpenAI is reportedly working on its first in-house AU chip to lessen its reliance on Nvidia, which controlled approximately 80% of the AI chip market as of last year. According to a new report from Reuters, the ChatGPT maker may finalize the design of its first 3nm AI chip in the next few months so it can enter mass production at TSMC in 2026.

The new chip is reportedly being developed by a team of 40 OpenAI employees collaborating with Broadcom. Richard Ho, the company’s new Head of Hardware, is leading that effort after previously working on silicon solutions for Google’s infrastructure and cloud services.

“OpenAI’s in-house AI chip, while capable of both training and running AI models, will initially be deployed on a limited scale, and primarily for running AI models. The chip will have a limited role within the company’s infrastructure,” Reuters explained citing sources familiar with the project.

There’s currently a very high demand for Nvidia’s AI chips as companies like OpenAI, Microsoft, Meta, Google, and others are investing billions of dollars into AI data centers. With the recently-announced Stargate Project, OpenAI and its partners SoftBank, Oracle, and MGX plan to invest $ 500 billion in new AI infrastructure in the US over the next four years.

However, designing cutting-edge AI chips that can rival Nvidia’s offerings is going to be challenging for a company. Reuters reports that OpenAI may spend $500 million to develop the first version of the 3nm chip, with costs possibly doubling to create the software and peripherals that will use it.

“Inside OpenAI, the training-focused chip is viewed as a strategic tool to strengthen OpenAI’s negotiating leverage with other chip suppliers,” the report said. “After the initial chip, OpenAI’s engineers plan to develop increasingly advanced processors with broader capabilities with each new iteration.”