OpenAI and semiconductor manufacturer Broadcom have signed one of the largest computing infrastructure deals in history, a partnership that will jointly develop and deploy 10 GW of AI computing power over the next four years.
OpenAI and semiconductor manufacturer Broadcom have signed one of the largest computing infrastructure deals in history, a partnership that will jointly develop and deploy 10 GW of AI computing power over the next four years.
According to The Wall Street Journal, this deal, estimated to be worth several billion dollars, is intended to meet OpenAI's growing demand for computing resources to train its future large language models.
Broadcom shares responded to the news by rising almost 10%, increasing the company's market capitalization to over $1.5 trillion.
Under the terms of the agreement, Broadcom will begin manufacturing GPUs designed exclusively for use in OpenAI systems. Production of these chips will begin in the second half of 2026.
The new infrastructure will include server racks and Broadcom Ethernet-based networking equipment that will be installed in the data centers of OpenAI and its partners.
With this deal, as well as partnerships with Nvidia and AMD , the total computing power that OpenAI plans to deploy will reach 26 GW — more than double New York City's summer electricity needs.
OpenAI founder and CEO Sam Altman said the company plans to create up to 250 GW of new capacity by 2033. This investment could cost more than $10 trillion.
Recall that thanks to new deals, OpenAI 's valuation reached $500 billion , overtaking Elon Musk's SpaceX.
