Umbilical cord to Nvidia is being cut OpenAI has its own AI processor manufactured

From Manuel Christa | Translated by AI 1 min Reading Time

Related Vendors

OpenAI plans to complete the development of its first own AI chip this year and is already close to the tape-out. This is reported by Reuters, citing "informed sources."

OpenAI aims to become more independent from Nvidia and is developing its own AI processor.(Image: AI-generated)
OpenAI aims to become more independent from Nvidia and is developing its own AI processor.
(Image: AI-generated)

The ChatGPT developer has been working on its own AI accelerator for some time. The goal is to create long-term alternatives to the highly demanded Nvidia chips that currently dominate the market. According to Reuters, OpenAI is close to the so-called "tape-out"—the completion of the development phase before production in a chip factory. Manufacturing is to be carried out by the Taiwanese semiconductor manufacturer TSMC using the 3-nm process.

The development is led by Richard Ho, a former Google engineer. His team at OpenAI has grown to 40 people in recent months and is working in collaboration with Broadcom on the chip architecture.

None of the involved companies have officially confirmed any of this. However, a tape-out would be hard to conceal. It costs tens of millions of US dollars and takes about half a year. Until a functioning processor is manufactured, it can easily take one to one and a half years.

Strategic importance for OpenAI

A proprietary chip could give OpenAI more leverage in negotiations with Nvidia. At the same time, it is primarily intended to be used for inference—that is, the execution of trained AI models. However, the development of powerful own chips remains a challenge: Google and Meta have been investing in their own hardware for years and are still facing technical hurdles. No other major competitor has yet succeeded in developing a competitive chip, despite larger teams being engaged in this effort, Reuters further reports.

Despite high development costs estimated at half a billion US dollars, a successful tape-out could pave the way for OpenAI to mass production. However, the first chip is likely to be used only to a limited extent. Widespread deployment not only requires further iterations of the design but also significant expansion of the infrastructure (via Reuters). (mc)

Subscribe to the newsletter now

Don't Miss out on Our Best Content

By clicking on „Subscribe to Newsletter“ I agree to the processing and use of my data according to the consent form (please expand for details) and accept the Terms of Use. For more information, please see our Privacy Policy. The consent declaration relates, among other things, to the sending of editorial newsletters by email and to data matching for marketing purposes with selected advertising partners (e.g., LinkedIn, Google, Meta)

Unfold for details of your consent