OpenAI and Nvidia’s $100B AI Plan Needs Power of 10 Nuclear Reactors
▼ Summary
– OpenAI and Nvidia announced a partnership to deploy at least 10 gigawatts of Nvidia systems for OpenAI’s AI infrastructure, with Nvidia planning to invest up to $100 billion.
– The first gigawatt of systems is scheduled to come online in the second half of 2026 using Nvidia’s Vera Rubin platform.
– The 10-gigawatt scale is unprecedented, equaling the power output of roughly 10 nuclear reactors and dwarfing current data center energy consumption.
– Nvidia’s stock rose nearly 4% following the announcement, adding approximately $170 billion to its market capitalization.
– This partnership follows a separate $5 billion investment Nvidia made in Intel the previous week to co-develop custom data center and PC products.
A landmark collaboration between OpenAI and Nvidia aims to fundamentally reshape the landscape of artificial intelligence infrastructure. The companies have unveiled a strategic partnership centered on deploying a staggering 10 gigawatts of Nvidia systems to power OpenAI’s future AI development. This monumental undertaking, backed by a planned Nvidia investment of up to $100 billion, represents an unprecedented scale of computational ambition. The first phase, delivering one gigawatt of capacity utilizing Nvidia’s upcoming Vera Rubin platform, is scheduled to become operational in the latter half of 2026.
OpenAI CEO Sam Altman emphasized the foundational role of this compute power, stating that it will serve as the bedrock for the future economy. He confirmed the infrastructure will be leveraged to achieve new AI breakthroughs and deploy them at a massive scale for people and businesses worldwide. The sheer magnitude of the project is difficult to overstate. Nvidia CEO Jensen Huang provided a striking comparison, noting that the planned 10 gigawatts equates to the power consumption of between four and five million graphics processing units. This figure matches Nvidia’s total GPU shipments for the entire current year and doubles the volume from the previous year, leading Huang to simply describe the endeavor as “a giant project.”
To fully grasp the energy requirements, consider that 10 gigawatts is equivalent to the output of approximately 10 standard nuclear power plants. This level of demand completely overshadows today’s largest data centers, which typically consume between 50 and 100 megawatts. The planned AI infrastructure would effectively require as much electricity as several major metropolitan areas combined, highlighting the immense power needs of next-generation artificial intelligence.
This partnership arrives on the heels of OpenAI’s explosive growth, which now sees 700 million weekly active users engaging with its technologies. The market responded positively to the news, with Nvidia’s stock climbing nearly 4% and adding roughly $170 billion to its market value in a single day. The agreement solidifies Nvidia’s position as OpenAI’s preferred partner for compute and networking, existing alongside OpenAI’s other strategic relationships with Microsoft, Oracle, SoftBank, and participants in the recently announced Stargate project.
Interestingly, this announcement comes just one week after Nvidia disclosed a separate $5 billion investment in Intel, acquiring a 4% stake in its longtime rival. That move is part of a plan for the two chipmakers to co-develop custom products for data centers and personal computers, signaling a period of intense strategic maneuvering within the industry.
(Source: Ars Technica)
