The AI boom has turned Nvidia into an unstoppable giant. While companies like OpenAI, Anthropic, Meta and other AI labs are racing to get an edge, all of them depend on Nvidia for the raw computing power needed to run their LLMs. But now, Nvidia CEO Jensen Huang says chips alone are no longer enough to meet the demands of the AI boom. They also need vast amounts of land, power and data-centre infrastructure. So Nvidia has now decided to sell– lease to be specific– these resources to the AI companies.

In a long post on X titled “Securing the Infrastructure of Intelligence”, Huang says Nvidia is now expanding its focus to the physical infrastructure needed to deploy AI. “” AI factories require a full stack of critical resources: advanced chips, packaging, memory, and networking – as well as land, power and shell,”he wrote. By “shell”, Huang is referring to the physical data-centre infrastructure needed to house and run AI systems.

Huang says AI companies are growing so quickly that getting enough compute is becoming a challenge in itself. “Their growth is increasingly constrained not by algorithms or customer demand, but by the availability of compute,” he says. That is where Nvidia now wants to step in. It wants to help AI labs secure the land, power and infrastructure needed to deploy more AI compute.

The company has already started putting this plan into action through a partnership with SB Energy at the PORTS-Pike Technology Campus in Portsmouth, Ohio. The site will host Nvidia AI factories, with OpenAI as the tenant under a 20-year lease. During this time, OpenAI will build and operate the AI factory, which will use Nvidia’s full-stack DSX AI factory platform, including its GPUs, CPUs, networking and infrastructure software.

The next big AI plan of Nvidia

So, what exactly is Nvidia doing here? The initial plan is to build an AI factory with 4.25 gigawatts of capacity at PORTS-Pike. In simple terms, that means the site will have a huge amount of computing infrastructure and power dedicated to running AI systems. Nvidia says it could later secure another 3.75 gigawatts, taking the site’s potential capacity to 8 gigawatts.

The first phase of the project is expected to come online between 2028 and 2030. The AI factory will use Nvidia’s full-stack DSX platform, including its GPUs, CPUs, networking and infrastructure software. This means the site is being designed not just as a data centre, but as a large-scale facility built specifically to run Nvidia-powered AI computing.

OpenAI is the first customer

But Nvidia is not simply paying for the entire site and handing it over to OpenAI. Huang reveals that Nvidia will provide support for defined portions of the lease and power payments, along with a specified residual-value commitment. OpenAI will remain the tenant and will pay the lease. In other words, Nvidia is helping make sure the land, electricity and physical infrastructure needed for its AI systems are available, rather than taking on the full cost of the project.

Nvidia is also investing $1.5 billion in SB Energy as part of the arrangement. The bigger idea, according to Huang, is to secure sites that can keep hosting Nvidia’s AI systems as the technology gets upgraded. So, instead of building infrastructure for just one generation of chips, the same site could host newer and more powerful Nvidia systems over several generations.

– Ends

Published By:

Divya Bhati

Published On:

Aug 17, 2026 18:58 IST