OpenAI in cooperation with Broadcom makes a dedicated artificial intelligence chip

OpenAI has announced that it is developing its own chips in cooperation with Broadcom to provide the computing power of artificial intelligence data centers.

According to The Verge, the agreement between OpenAI and Broadcom is part of ChatGPT’s efforts to reduce dependence on Nvidia chips and ensure sufficient computing power to support applications such as ChatGPT and Sora, as well as fulfilling OpenAI’s mission to develop super-intelligent artificial intelligence.

OpenAI makes a dedicated chip with the aim of reducing dependence on Nvidia

According to OpenAI, the proprietary chip design allows the company to embed the knowledge and experience gained from developing leading models and products directly into the hardware, unlocking a new level of capability and intelligence.

The partnership will also allow OpenAI to develop and deploy 10 gigawatts of dedicated artificial intelligence accelerator capacity using its own chips and systems. To better understand this number, it should be said that the output of a typical nuclear reactor is about one gigawatt.

Broadcom is going to start deploying equipment racks from the second half of 2026, and this cooperation will be completed by the end of 2029. “This collaboration is a critical step in building the infrastructure needed to unlock the potential of artificial intelligence and deliver real benefits to people and businesses,” said Sam Altman, CEO of OpenAI.

The news of OpenAI’s cooperation with Broadcom is announced after this artificial intelligence company announced the signing of an agreement of 6 gigawatts with AMD and 10 gigawatts with Nvidia. The new OpenAI agreements come after the company changed its exclusive agreement with Microsoft to provide AI computing power.

The design and production of proprietary chips has become a trend in the technology industry, and the giants in this field are going to manufacture and design their own chips with the aim of reducing dependence on chip manufacturing companies and reducing costs.

Currently, these projects do not seem to be a threat to Nvidia’s dominance of the artificial intelligence chip market, and the company still holds a large part of this market.