OpenAI, Broadcom to collaborate on next-gen custom chips amid AI bubble concerns

OpenAI and Broadcom have announced a strategic partnership to produce its first custom artificial intelligence processors in the midst of growing concerns over the AI bubble.

The announcement came on the heels of the tech race where the companies are striving hard to secure the computing power needed to meet surging service demands and maintain dominance.

Both California-based companies will deploy custom “AI accelerators” in the next year.

As per agreement, OpenAI would design the chips and Broadcom will start its deployment in the second half of 2026.

Moreover, the tech giants are also planning to launch 10 gigawatts worth of custom chips, which will consume power equivalent to the needs of more than 8 million US households or five-fold the electricity produced by the Hoover Dam.

Although both companies have not revealed the financial terms of the deal, it is the latest massive deal between OpenAI and Broadcom.

Earlier this week, OpenAI also entered a 6-gigawatt AI chip supply deal with AMD. Prior to this deal, Nvidia announced plans to invest up to $100 billion in OpenAI, equipping it with 10-gigawatts of data centers.

The circular financing deals have raised the concerns of the AI bubble as many experts, IMF, and Bank of England fear that it is about to burst.

"Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI's potential," OpenAI CEO Sam Altman said in a statement.

Altman also described it as “a gigantic amount of computing infrastructure to serve the needs of the world to use advanced intelligence.”

Following the announcement, the shares of Broadcom rose more than 10 percent as per Reuters.