OpenAI Partners with Broadcom to Build Its First AI Chip as Competition with Nvidia Heats Up
OpenAI has announced a new partnership with Broadcom to develop its first in-house artificial intelligence processors, marking another major move in the company’s push to secure greater computing power for its fast-growing AI services.
Following the announcement, Broadcom shares surged more than 10% on Monday.
Under the deal, OpenAI will design the processors, while Broadcom will handle development and deployment, starting in the second half of 2026. The companies plan to roll out 10 gigawatts of custom AI chips, enough to power over 8 million U.S. households or roughly five times the output of the Hoover Dam.
While this move highlights OpenAI’s ambition to expand its hardware capabilities, analysts say it is unlikely to immediately challenge Nvidia’s dominance in the AI accelerator market, given the complexity and scale of chip design and manufacturing.
This collaboration is part of a broader wave of AI chip investments across the tech industry, as companies race to develop custom hardware to support increasingly advanced AI systems.
Just last week, OpenAI revealed a 6-gigawatt chip supply deal with AMD, including an option to acquire a stake in the chipmaker. The news came days after reports that Nvidia plans to invest up to $100 billion in OpenAI and supply it with data-center infrastructure boasting at least 10 gigawatts of capacity.
“Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI’s potential,” said OpenAI CEO Sam Altman.
The financial details of the agreement were not disclosed. However, experts note that the 2026 rollout timeline is ambitious. A one-gigawatt data center can cost between $50 billion and $60 billion, according to Nvidia CEO Jensen Huang, with Nvidia hardware typically accounting for more than half of that cost.
Gadjo Sevilla, an analyst at eMarketer, said OpenAI’s strong investor confidence makes it well-positioned to raise the required funds. Financing will likely involve a combination of funding rounds, strategic investments, pre-orders, and Microsoft’s continued backing, he added.
Broadcom Gains as Custom Chip Market Accelerates
The partnership, first reported by Reuters last year, puts OpenAI alongside other cloud giants like Google and Amazon, which are also developing custom AI chips to reduce reliance on Nvidia’s supply-constrained processors.
Although Microsoft and Meta have also pursued similar efforts, their in-house chips have not yet matched Nvidia’s performance, according to industry reports.
Still, the growing demand for AI-specific hardware has turned Broadcom into one of the biggest beneficiaries of the generative AI boom — with its stock up nearly sixfold since late 2022.
In September, Broadcom revealed a $10 billion custom AI chip order from an undisclosed client, which some analysts speculated was OpenAI. However, a company executive later suggested that OpenAI was not the unnamed customer.
Both companies confirmed that the new AI processor systems will be fully deployed by the end of 2029, using Broadcom’s Ethernet and networking technologies — a direct alternative to Nvidia’s InfiniBand networking solution.
