OpenAI in Talks With Broadcom About AI Chip Development

According to a report from The Information, AI startup OpenAI is in talks with a few chip designers, including Broadcom, about developing its own AI chip to cut reliance on chip market dominator Nvidia and stay ahead in the AI race.

Sources familiar with the matter told The Information that OpenAI plans to venture into the AI chip production market with its own AI server chip amid the high costs and shortage of graphic processing units it uses to develop AI models like ChatGPT and DALL-E3. To speed up its chip production, the Microsoft-backed startup is hiring former Google employees who have created the search giant’s proprietary AI chip, the tensor processing unit.

Despite the support from former Google team members, the report argues that OpenAI’s AI chip can hardly compete with that of the largest AI chipmaker, NVIDIA. While OpenAI might be far from becoming an AI chip leader from the ranks of NVIDIA with customers like Microsoft and Google, developing its own AI server chip could make it more self-reliant and enhance its AI development efforts.

According to a source familiar with the matter, Microsoft’s $13 billion investment in OpenAI won’t be enough to feed the startup’s chip-making ambitions. Previously, it was reported that OpenAI CEO Sam Altman was seeking between $5 and $7 trillion in investments to start a chip-making business. To raise such a massive amount, OpenAI would likely need to rely on multiple commercial partnerships.

“OpenAI is having ongoing conversations with industry and government stakeholders about increasing access to the infrastructure needed to ensure AI’s benefits are widely accessible,” a spokesperson for OpenAI told the Information. “This includes working in partnership with the premier chip designers, fabricators, and the brick-and-mortar developers of data centers,” the spokesperson added without going into much detail.

Broadcom has yet to comment on The Information’s claims.