OpenAI has announced a major partnership with semiconductor company Broadcom to design and develop 10 gigawatts of custom artificial intelligence chips and systems, a scale of power equivalent to the electricity consumption of a large city. The collaboration highlights the increasing energy demands of the global AI boom and OpenAI’s push to expand its computing capacity.
The deal marks OpenAI’s latest move to strengthen its hardware capabilities as demand for its AI tools continues to surge. The company already works with Nvidia and AMD, but this new partnership signals a deeper commitment to developing its own specialized chips to support next-generation AI models.
“Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI’s potential and deliver real benefits for people and businesses,” said Sam Altman, OpenAI’s co-founder and CEO, in a press release announcing the partnership.
The deployment of the new AI accelerator and network systems is expected to begin in the second half of 2026. According to Reuters, the project’s total energy use would match that of about 8 million U.S. households, fueling growing concern about the environmental impact of AI technologies.
A 2024 U.S. Department of Energy report found that data centers could consume between 6.7% and 12% of total U.S. electricity by 2028, up from 4.4% in 2023. Altman previously noted that a single ChatGPT query uses roughly as much energy as a lightbulb operating for a few minutes. Generating realistic video content through advanced models like Sora 2, he added, consumes far more.
Despite these concerns, the announcement sent Broadcom’s stock soaring 12% on Monday morning. The news reinforced investor optimism in the chipmaker’s growing role in the AI hardware race.
OpenAI’s partnership with Broadcom comes as its products reach unprecedented popularity. ChatGPT now attracts about 800 million weekly active users, and its recently launched video generation app, Sora, is expanding even faster. To sustain this growth, OpenAI has been working to secure long-term access to advanced chips and computing power.
Developing custom accelerators with Broadcom will allow OpenAI to tailor its hardware for AI workloads, improving performance and efficiency while reducing reliance on external suppliers. In its announcement, the company said the partnership enables it to “embed what it’s learned from developing frontier models and products directly into the hardware, unlocking new levels of capability and intelligence.”
Analysts say the collaboration reflects a broader industry trend in which leading AI companies seek tighter control over both software and hardware to gain a competitive edge. By designing its own chips, OpenAI aims to increase processing speed, cut costs, and enhance scalability for training and deploying massive AI models.
Hints of this deal surfaced earlier when Broadcom CEO Hock Tan revealed during an earnings call that the company had landed a new $10 billion customer, widely believed to be OpenAI. The confirmation cements Broadcom’s position as a major contender in the fast-evolving AI chip market, which has been dominated by Nvidia.
As the AI industry continues to expand, securing a reliable chip supply has become essential. With this partnership, OpenAI moves closer to achieving greater independence over its hardware ecosystem while laying the groundwork for more advanced, energy-intensive systems that will define the next chapter of artificial intelligence.
Related Readings:
