The AI factory is the infrastructure shaping the next era of artificial intelligence. It transforms energy, data, and advanced computing into intelligence that powers businesses, industries, and countries worldwide.
In the AI economy, computing is revenue.
Building an AI factory requires a complete infrastructure stack, including land, power, and shell capacity (LPS), along with advanced chips, semiconductor packaging, memory, networking, and software.
Just as NVIDIA uses its scale, long-term demand visibility, and supply-chain partnerships to secure critical semiconductor resources, the company is applying the same disciplined approach to securing dedicated LPS capacity for NVIDIA AI factories.
NVIDIA is partnering with SB Energy to secure LPS capacity for NVIDIA computing at the premier PORTS-Pike Technology Campus in Portsmouth, Ohio. OpenAI will serve as the tenant.
LPS: The Next Strategic Resource for AI Infrastructure
For most NVIDIA customers, securing land, power, and shell capacity has long been a central part of their data center and AI infrastructure strategy.
The world’s largest cloud service providers and investment-grade companies have the balance sheets, infrastructure expertise, and long-term contracts required to secure LPS independently. These organizations build and operate AI factories powered by NVIDIA accelerated computing, networking, systems, and software.
This model will continue to represent the majority of NVIDIA’s business.
However, frontier AI labs have different infrastructure requirements.
Frontier AI labs are experiencing extraordinary demand for AI training and inference computing. In many cases, that demand is growing faster than their balance sheets and long-term credit profiles can support. These companies may have strong customer demand and rapidly increasing revenue, but they often lack the decades-long infrastructure contracts and investment-grade financing capacity needed to develop their own AI factories.
As a result, their growth is increasingly constrained by computing availability rather than by algorithms or customer demand.
For these companies, additional computing capacity can deliver more intelligence, products, users, and revenue. NVIDIA is helping provide the infrastructure required to power this growth cycle.
PORTS-Pike: A Long-Term Site for NVIDIA AI Computing
OpenAI will build and operate a world-class AI factory at PORTS-Pike. The facility will use NVIDIA’s full-stack DSX AI Factory platform, including NVIDIA GPUs, CPUs, networking, and infrastructure software.
The initial deployment is expected to provide 4.25 gigawatts of AI factory capacity. Each generation of NVIDIA AI factory systems deployed at PORTS-Pike is expected to represent approximately 1.5 million NVIDIA GPUs and approximately $150 billion to $200 billion in revenue for NVIDIA. Over more than 20 years, the site can support multiple technology upgrade cycles.
This long-term approach is economically important. NVIDIA’s commitment to LPS supports the longevity of the AI factory site, while the NVIDIA computing systems inside the facility can be upgraded repeatedly. Each new generation can deliver greater production, more intelligence, and improved economics.
NVIDIA may also expand the PORTS-Pike arrangement beyond the initial 4.25 gigawatts to secure the site’s remaining 3.75 gigawatts of capacity.
OpenAI and NVIDIA Expand AI Computing Capacity
More broadly, OpenAI is committed to significant deployment of NVIDIA AI infrastructure through 2030. OpenAI’s existing and planned initiatives represent approximately 12 gigawatts of NVIDIA computing capacity, with the potential to increase to approximately 16 gigawatts if NVIDIA expands the PORTS-Pike agreement beyond the original 4.25 gigawatts.
At these levels, the opportunity represents approximately $600 billion in NVIDIA computing by 2030.
Key Questions About the PORTS-Pike Agreement
What does NVIDIA guarantee, and for how long?
NVIDIA will support approximately 4 gigawatts of PORTS-Pike’s LPS infrastructure for 20 years, helping secure the site for the exclusive deployment of NVIDIA computing.
NVIDIA’s support is limited to a defined portion of lease and electricity payments, as well as a specified residual value guarantee. It does not cover the full cost of the site or all of the tenant’s obligations.
The guarantee will take effect gradually as data centers come online between 2028 and 2030. As OpenAI makes lease payments and capacity becomes operational, NVIDIA’s remaining exposure will decline.
Why is NVIDIA supporting PORTS-Pike?
Land, power, and shell capacity are significant constraints in the deployment of AI factories. NVIDIA is selectively reserving premier sites that can host multiple generations of NVIDIA computing and meet sustained customer demand.
The productive life of a site can be extended through successive generations of NVIDIA systems. Each generation is designed to deliver more intelligence, computing performance, and revenue than the generation before it.
Is this a circular loan?
No. OpenAI will pay the lease fee.
NVIDIA is using its scale and long-term demand visibility to secure PORTS-Pike as a site for NVIDIA computing. This approach reflects the same discipline NVIDIA applies to supply-chain management: visibility into customer demand helps secure critical inputs and support long-term production capacity.
What happens to PORTS-Pike if OpenAI no longer uses the site?
NVIDIA computing is versatile, fungible, and widely adopted. If necessary, the capacity could be made available to other eligible tenants across NVIDIA’s global ecosystem, including cloud service providers, enterprises, AI labs, and startups.
Through CUDA, NVIDIA extends the value of its computing platform beyond hardware. CUDA provides developers and NVIDIA engineers with a common software environment for the continuous improvement of deployed systems.
CUDA makes NVIDIA computing versatile, while that versatility makes the computing capacity fungible. Greater fungibility can improve availability and durability, making NVIDIA computing a productive asset that can be rented, redeployed, and financed.
The value of a high-quality site such as PORTS-Pike is not limited to one customer or one generation of computing. NVIDIA’s standardized platform, broad developer ecosystem, and large base of potential users support the company’s ability to redeploy production capacity over time.
How much LPS capacity will NVIDIA secure?
NVIDIA’s approach will be strategic and disciplined.
Most NVIDIA customers will continue to secure their own LPS capacity. The majority of facilities hosting NVIDIA computing will continue to be secured directly by cloud service providers, enterprises, sovereign AI builders, and other customers.
NVIDIA will selectively focus on exceptional sites where visible, sustained demand can support multiple generations of NVIDIA computing infrastructure.
Building the Intelligence Infrastructure
PORTS-Pike represents the next step in NVIDIA’s evolution as an AI infrastructure company.
NVIDIA began by developing accelerated computing chips and has since expanded into complete systems, networking, CUDA, and full-stack AI factory platforms. The company is now helping secure the critical physical infrastructure required to build and operate these AI factories.
NVIDIA is a full-stack AI infrastructure platform.
By investing in a durable foundation for AI factories, NVIDIA is helping customers deploy the world’s most productive computing platform across multiple technology generations.
By securing the critical resources needed to host NVIDIA computing, the company can help the world’s most innovative organizations build the AI factories powering the age of intelligence.
Source: blogs.nvidia.com