top of page

Inside PORTS-Pike: How NVIDIA and OpenAI Are Building an 8-Gigawatt Powerhouse for Frontier AI

The race to build increasingly capable artificial intelligence systems is entering a new phase. Computing power remains central, but access to GPUs alone is no longer enough. The next generation of AI requires enormous quantities of electricity, specialized data centers, high-capacity networks, suitable land, cooling systems, transmission infrastructure, financing and long-term operational planning. The emerging PORTS-Pike Technology Campus in Pike County, Ohio, illustrates this transformation. OpenAI, NVIDIA, SB Energy and the U.S. Department of Energy are collaborating on an AI infrastructure project expected to secure approximately 8 gigawatts of IT capacity, creating one of the most significant physical foundations for frontier AI development. The project is more than another hyperscale data center. It demonstrates how the economics of artificial intelligence are increasingly being shaped by physical infrastructure, and why land, power and the data center shell are becoming strategic resources alongside chips, memory and networking. For the AI industry, the implications extend far beyond Ohio. The AI Race Is Becoming an Infrastructure Race The first phase of modern AI competition centered heavily on algorithms, datasets and model architectures. As foundation models became increasingly capable, accelerated computing became the dominant constraint. That constraint is now broadening. A company can have access to advanced GPUs and still struggle to deploy them at scale if it cannot secure sufficient electricity, transmission capacity, cooling infrastructure and physical space. Frontier AI laboratories are particularly exposed because their demand for training and inference capacity can grow faster than traditional infrastructure planning cycles. This creates a new economic relationship between computing and infrastructure. A gigawatt-scale AI factory requires an ecosystem rather than a conventional server facility. Electricity must reach the site reliably, computing equipment must be installed and cooled, networking must connect enormous clusters, and the entire system must be engineered for continuous operation. The result is an emerging model in which AI infrastructure increasingly resembles industrial infrastructure. The PORTS-Pike project reflects that shift by combining: Large-scale electrical capacity Dedicated data center infrastructure NVIDIA accelerated computing High-speed networking Advanced cooling systems Long-term capacity commitments New transmission infrastructure Workforce development Community investment The physical infrastructure therefore becomes part of the AI company's competitive strategy. PORTS-Pike Could Become a Major Frontier AI Compute Hub OpenAI has entered an agreement to secure approximately 8 gigawatts of IT capacity at PORTS-Pike. The development is expected to occur progressively rather than appearing as a fully operational facility at once. The first 800 megawatts are expected to become available in 2028, with further expansion requiring additional generation, transmission infrastructure and other supporting systems. The broader buildout is expected to continue through 2032. NVIDIA's announcement separately describes an initial 4.25-gigawatt AI factory deployment, with the possibility of securing another 3.75 gigawatts. NVIDIA estimates that each generation deployed at the site could involve approximately 1.5 million GPUs, illustrating the extraordinary scale involved. This is important because AI factories are not static assets. A conventional data center can operate with relatively stable hardware for years. AI infrastructure has a different lifecycle. GPU generations evolve rapidly, networking technologies improve, power efficiency changes and software increasingly determines how effectively physical resources are utilized. A site capable of supporting multiple generations of computing hardware can therefore remain strategically valuable long after its initial equipment becomes obsolete. The economics of a multigenerational AI factory NVIDIA estimates that each generation deployed at PORTS-Pike could represent approximately $150 billion to $200 billion in NVIDIA revenue, depending on the configuration and scale of deployment. Across OpenAI's existing and planned commitments, NVIDIA describes approximately 12 gigawatts of compute through 2030, with a potential increase to roughly 16 gigawatts if the additional PORTS-Pike capacity is secured. At the levels described by NVIDIA, the opportunity could represent approximately $600 billion of NVIDIA compute through 2030. These figures illustrate why infrastructure decisions are increasingly becoming strategic financial decisions. Securing a site today can create the physical foundation for several generations of future computing demand. Why Land, Power and Shell Are Becoming Strategic Assets The concept of LPS, land, power and shell, is becoming increasingly important in AI infrastructure. Land determines where an AI factory can be built. Power determines how much compute it can operate. The shell provides the physical environment in which computing systems can be deployed. Historically, companies often treated these elements as facilities-management considerations. The explosive growth of AI is changing that equation. Power availability can determine whether a project proceeds at all. Grid connections and transmission projects can require long planning horizons. Suitable land must meet technical, environmental and logistical requirements. Data center construction itself requires substantial capital and specialized engineering. This creates a potential bottleneck between demand for AI and the physical capacity needed to satisfy it. For major cloud providers and investment-grade enterprises, securing these resources independently is generally more feasible because they possess established financing capacity and long-term infrastructure relationships. Frontier AI companies can face a different challenge. Their computing requirements may expand at extraordinary rates while their infrastructure financing history is comparatively young. NVIDIA's decision to provide selective support for LPS infrastructure therefore represents an attempt to address a specific bottleneck in the AI supply chain. NVIDIA Is Expanding Beyond Chips The PORTS-Pike strategy also demonstrates how NVIDIA's role in the AI ecosystem has expanded. NVIDIA originally became synonymous with GPUs. Its position today is much broader, encompassing CPUs, networking, systems, software, CUDA and complete AI factory architectures. Now the company is moving toward securing some of the physical infrastructure required to deploy those systems. NVIDIA says it will provide support associated with the initial 4.25-gigawatt deployment and separately invest $1.5 billion in SB Energy. Its support includes defined portions of lease and power payments and a residual-value commitment rather than assuming the entire cost of the project. The structure is strategically significant. NVIDIA is effectively attempting to ensure that infrastructure exists when customers require large quantities of NVIDIA compute, while limiting its exposure through phased deployment. The guarantees become effective as data center capacity enters service between 2028 and 2030, while NVIDIA's remaining exposure declines as lease payments and capacity come online. This is closer to infrastructure risk management than conventional hardware sales. OpenAI's Long-Term Compute Strategy For OpenAI, the project addresses one of the most consequential challenges facing frontier AI laboratories, access to sustained computing capacity. OpenAI says it will use the site for frontier training and growing product demand. The company will pay for capacity as completed portions become available and plans to finance its commitments through business revenue, cash flow and capital raised from investors. The arrangement also separates ownership and usage. SB Energy will build, own and operate the data center under a 20-year lease to OpenAI. OpenAI becomes the customer, while the facility will exclusively host NVIDIA AI compute infrastructure. That structure provides OpenAI with long-term access to infrastructure without requiring it to own every physical component of the facility. It also reflects a broader trend toward AI infrastructure becoming an asset that can be financed, leased and potentially redeployed across generations. Why NVIDIA Believes the Infrastructure Can Remain Valuable One of the central risks in technology infrastructure is obsolescence. Hardware can lose value quickly when newer generations arrive. NVIDIA's argument is that AI compute has an important characteristic that can mitigate this risk, versatility. Because NVIDIA's hardware and software platform is widely adopted, a facility designed around NVIDIA infrastructure could potentially serve different customers over time if the original tenant's requirements change. CUDA is particularly important to this thesis because it provides a common software ecosystem connecting developers, applications and NVIDIA hardware. That ecosystem can increase the usefulness of installed infrastructure beyond a single customer or GPU generation. In practical terms, a large AI factory is not necessarily valuable only because of the specific chips installed today. Its power capacity, networking architecture, physical location and supporting infrastructure can remain useful while successive generations of computing equipment are installed. This creates a potential distinction between hardware depreciation and infrastructure durability. Ohio's Economic Transformation The PORTS-Pike project also has a significant regional dimension. The development is expected to generate approximately 35,000 construction jobs during its six-year buildout through 2032, followed by approximately 2,500 long-term operating jobs. OpenAI has committed an additional $40 million community grant fund, building on SB Energy's previously announced $40 million commitment. The funding is intended to support locally identified priorities including schools, public safety, health care, utilities, workforce training, housing, veterans' services, small businesses and working families. The project is also expected to generate hundreds of millions of dollars in state and local tax revenue over its lifetime. This is strategically important because the AI infrastructure boom is creating a new form of regional industrial development. Communities that secure large data center investments can gain construction activity, technical employment, infrastructure improvements and new demand for local suppliers. However, the long-term value depends on whether these projects create durable skills and businesses rather than only temporary construction activity. The workforce strategy at PORTS-Pike is therefore notable. The project is working with building trades, educational institutions, apprenticeship programs, veterans' organizations and workforce groups to develop pathways into construction, technical and operational careers. Energy and Water Are Central to AI Infrastructure The scale of AI factories makes environmental and utility considerations unavoidable. OpenAI says SB Energy will cover the full cost of grid upgrades and new transmission infrastructure required to serve the project, rather than transferring those costs to regional ratepayers. The project will also use closed-loop, air-cooled cooling systems designed to recirculate water instead of relying on cooling towers that continuously consume water. OpenAI says that once the cooling system is filled, ongoing water consumption is expected to be comparable to that of an office building serving a similar population, although the final expected water use will be publicly reported after site design is completed. Energy remains the more fundamental challenge. The first phase can largely use existing AEP infrastructure, but subsequent development is expected to require new power generation, including natural gas generation, along with additional transmission infrastructure. That requirement highlights an uncomfortable reality of AI expansion: digital intelligence ultimately depends on physical energy systems. The PORTS-Pike Project Could Become a Blueprint for AI Factories NVIDIA and OpenAI plan to collaborate on the design, testing and commissioning of the facility and publish a technical white paper covering lessons from the project. The focus includes resilient infrastructure design, component qualification and software-level workload management. That combination is significant because AI cluster reliability cannot be solved solely through better hardware. When thousands or millions of accelerators operate as a coordinated system, interruptions in one part of the infrastructure can affect workloads across the cluster. Power systems, networking, cooling, hardware reliability and software scheduling must therefore operate as an integrated system. The future AI factory will increasingly resemble a highly optimized industrial production environment, except its output is intelligence rather than physical goods. What PORTS-Pike Means for the Future of AI The Ohio project provides a useful lens through which to understand the next stage of artificial intelligence. The competitive advantage of AI companies will increasingly depend on the ability to convert capital, electricity, data and computing hardware into reliable intelligence at scale. That means the industry's critical resources are expanding. Traditional AI bottlenecks Emerging AI factory requirements Algorithms Compute capacity Training data Electricity GPUs Land Model architecture Data center shell Software Transmission infrastructure Cloud access Cooling and water systems Engineering talent Skilled physical workforce Capital Long-term infrastructure financing The strategic lesson is clear. AI infrastructure is no longer simply a support function behind the technology. It is becoming part of the technology's competitive foundation. For NVIDIA, securing selected LPS capacity could strengthen its ability to deploy multiple generations of accelerated computing. For OpenAI, the project provides a long-term foundation for frontier training and inference. For Ohio, it represents a potential transition from a historically industrial site toward a new form of digital industrialization. For the broader AI economy, PORTS-Pike demonstrates that the next generation of intelligence will be built not only in laboratories and software environments, but also on enormous physical sites connected to power grids and industrial supply chains. Conclusion: The Physical Foundation of the Intelligence Era The AI industry has spent years emphasizing models, algorithms and chips. The next competitive frontier may be far more physical. Land must be secured. Power must be generated and transmitted. Data centers must be constructed. Cooling systems must operate efficiently. Networks must remain reliable. Hardware must be upgraded repeatedly. Communities must receive tangible economic benefits. The PORTS-Pike project brings all of those requirements together in one large-scale development. Its importance therefore extends beyond the number of GPUs or gigawatts involved. It represents a transition toward AI factories as long-lived industrial assets, capable of supporting successive generations of computing infrastructure. As Dr. Shahid Masood and the expert team at 1950.ai continue examining the evolution of artificial intelligence and advanced computing, developments such as PORTS-Pike deserve attention because they reveal a fundamental truth about the next phase of AI: intelligence may be digital in form, but its expansion increasingly depends on physical infrastructure. The companies that can reliably secure that infrastructure may gain an advantage that is as important as the companies building the models themselves. Further Reading / External References OpenAI joins PORTS-Pike project https://openai.com/index/openai-joins-ports-pike-project/ Securing the Infrastructure of Intelligence https://blogs.nvidia.com/blog/securing-the-infrastructure-of-intelligence/

The race to build increasingly capable artificial intelligence systems is entering a new phase. Computing power remains central, but access to GPUs alone is no longer enough. The next generation of AI requires enormous quantities of electricity, specialized data centers, high-capacity networks, suitable land, cooling systems, transmission infrastructure, financing and long-term operational planning.


The emerging PORTS-Pike Technology Campus in Pike County, Ohio, illustrates this transformation. OpenAI, NVIDIA, SB Energy and the U.S. Department of Energy are collaborating on an AI infrastructure project expected to secure approximately 8 gigawatts of IT capacity, creating one of the most significant physical foundations for frontier AI development.


The project is more than another hyperscale data center. It demonstrates how the economics of artificial intelligence are increasingly being shaped by physical infrastructure, and why land, power and the data center shell are becoming strategic resources alongside chips, memory and networking.

For the AI industry, the implications extend far beyond Ohio.


The AI Race Is Becoming an Infrastructure Race

The first phase of modern AI competition centered heavily on algorithms, datasets and model architectures. As foundation models became increasingly capable, accelerated computing became the dominant constraint.

That constraint is now broadening.


A company can have access to advanced GPUs and still struggle to deploy them at scale if it cannot secure sufficient electricity, transmission capacity, cooling infrastructure and physical space. Frontier AI laboratories are particularly exposed because their demand for training and inference capacity can grow faster than traditional infrastructure planning cycles.


This creates a new economic relationship between computing and infrastructure.

A gigawatt-scale AI factory requires an ecosystem rather than a conventional server facility. Electricity must reach the site reliably, computing equipment must be installed and cooled, networking must connect enormous clusters, and the entire system must be engineered for continuous operation.

The result is an emerging model in which AI infrastructure increasingly resembles industrial infrastructure.


The PORTS-Pike project reflects that shift by combining:

  • Large-scale electrical capacity

  • Dedicated data center infrastructure

  • NVIDIA accelerated computing

  • High-speed networking

  • Advanced cooling systems

  • Long-term capacity commitments

  • New transmission infrastructure

  • Workforce development

  • Community investment

The physical infrastructure therefore becomes part of the AI company's competitive strategy.


PORTS-Pike Could Become a Major Frontier AI Compute Hub

OpenAI has entered an agreement to secure approximately 8 gigawatts of IT capacity at PORTS-Pike. The development is expected to occur progressively rather than appearing as a fully operational facility at once.

The first 800 megawatts are expected to become available in 2028, with further expansion requiring additional generation, transmission infrastructure and other supporting systems.


The broader buildout is expected to continue through 2032.

NVIDIA's announcement separately describes an initial 4.25-gigawatt AI factory deployment, with the possibility of securing another 3.75 gigawatts. NVIDIA estimates that each generation deployed at the site could involve approximately 1.5 million GPUs, illustrating the extraordinary scale involved.


This is important because AI factories are not static assets.

A conventional data center can operate with relatively stable hardware for years. AI infrastructure has a different lifecycle. GPU generations evolve rapidly, networking technologies improve, power efficiency changes and software increasingly determines how effectively physical resources are utilized.


A site capable of supporting multiple generations of computing hardware can therefore remain strategically valuable long after its initial equipment becomes obsolete.

The economics of a multigenerational AI factory

NVIDIA estimates that each generation deployed at PORTS-Pike could represent approximately $150 billion to $200 billion in NVIDIA revenue, depending on the configuration and scale of deployment.


Across OpenAI's existing and planned commitments, NVIDIA describes approximately 12 gigawatts of compute through 2030, with a potential increase to roughly 16 gigawatts if the additional PORTS-Pike capacity is secured.

At the levels described by NVIDIA, the opportunity could represent approximately $600 billion of NVIDIA compute through 2030.

These figures illustrate why infrastructure decisions are increasingly becoming strategic financial decisions. Securing a site today can create the physical foundation for several generations of future computing demand.


Why Land, Power and Shell Are Becoming Strategic Assets

The concept of LPS, land, power and shell, is becoming increasingly important in AI infrastructure.

Land determines where an AI factory can be built. Power determines how much compute it can operate. The shell provides the physical environment in which computing systems can be deployed.


Historically, companies often treated these elements as facilities-management considerations. The explosive growth of AI is changing that equation.

Power availability can determine whether a project proceeds at all. Grid connections and transmission projects can require long planning horizons. Suitable land must meet technical, environmental and logistical requirements. Data center construction itself requires substantial capital and specialized engineering.

This creates a potential bottleneck between demand for AI and the physical capacity needed to satisfy it.


For major cloud providers and investment-grade enterprises, securing these resources independently is generally more feasible because they possess established financing capacity and long-term infrastructure relationships.

Frontier AI companies can face a different challenge.

Their computing requirements may expand at extraordinary rates while their infrastructure financing history is comparatively young. NVIDIA's decision to provide selective support for LPS infrastructure therefore represents an attempt to address a specific bottleneck in the AI supply chain.


NVIDIA Is Expanding Beyond Chips

The PORTS-Pike strategy also demonstrates how NVIDIA's role in the AI ecosystem has expanded.

NVIDIA originally became synonymous with GPUs. Its position today is much broader, encompassing CPUs, networking, systems, software, CUDA and complete AI factory architectures.

Now the company is moving toward securing some of the physical infrastructure required to deploy those systems.

NVIDIA says it will provide support associated with the initial 4.25-gigawatt deployment and separately invest $1.5 billion in SB Energy. Its support includes defined portions of lease and power payments and a residual-value commitment rather than assuming the entire cost of the project.

The structure is strategically significant.


NVIDIA is effectively attempting to ensure that infrastructure exists when customers require large quantities of NVIDIA compute, while limiting its exposure through phased deployment. The guarantees become effective as data center capacity enters service between 2028 and 2030, while NVIDIA's remaining exposure declines as lease payments and capacity come online.

This is closer to infrastructure risk management than conventional hardware sales.


OpenAI's Long-Term Compute Strategy

For OpenAI, the project addresses one of the most consequential challenges facing frontier AI laboratories, access to sustained computing capacity.

OpenAI says it will use the site for frontier training and growing product demand. The company will pay for capacity as completed portions become available and plans to finance its commitments through business revenue, cash flow and capital raised from investors.

The arrangement also separates ownership and usage.

SB Energy will build, own and operate the data center under a 20-year lease to OpenAI. OpenAI becomes the customer, while the facility will exclusively host NVIDIA AI compute infrastructure.


That structure provides OpenAI with long-term access to infrastructure without requiring it to own every physical component of the facility.

It also reflects a broader trend toward AI infrastructure becoming an asset that can be financed, leased and potentially redeployed across generations.


Why NVIDIA Believes the Infrastructure Can Remain Valuable

One of the central risks in technology infrastructure is obsolescence. Hardware can lose value quickly when newer generations arrive.

NVIDIA's argument is that AI compute has an important characteristic that can mitigate this risk, versatility.

Because NVIDIA's hardware and software platform is widely adopted, a facility designed around NVIDIA infrastructure could potentially serve different customers over time if the original tenant's requirements change.

CUDA is particularly important to this thesis because it provides a common software ecosystem connecting developers, applications and NVIDIA hardware.

That ecosystem can increase the usefulness of installed infrastructure beyond a single customer or GPU generation.


In practical terms, a large AI factory is not necessarily valuable only because of the specific chips installed today. Its power capacity, networking architecture, physical location and supporting infrastructure can remain useful while successive generations of computing equipment are installed.

This creates a potential distinction between hardware depreciation and infrastructure durability.


Ohio's Economic Transformation

The PORTS-Pike project also has a significant regional dimension.

The development is expected to generate approximately 35,000 construction jobs during its six-year buildout through 2032, followed by approximately 2,500 long-term operating jobs.


OpenAI has committed an additional $40 million community grant fund, building on SB Energy's previously announced $40 million commitment.

The funding is intended to support locally identified priorities including schools, public safety, health care, utilities, workforce training, housing, veterans' services, small businesses and working families.

The project is also expected to generate hundreds of millions of dollars in state and local tax revenue over its lifetime.


This is strategically important because the AI infrastructure boom is creating a new form of regional industrial development. Communities that secure large data center investments can gain construction activity, technical employment, infrastructure improvements and new demand for local suppliers.

However, the long-term value depends on whether these projects create durable skills and businesses rather than only temporary construction activity.

The workforce strategy at PORTS-Pike is therefore notable. The project is working with building trades, educational institutions, apprenticeship programs, veterans' organizations and workforce groups to develop pathways into construction, technical and operational careers.


Energy and Water Are Central to AI Infrastructure

The scale of AI factories makes environmental and utility considerations unavoidable.

OpenAI says SB Energy will cover the full cost of grid upgrades and new transmission infrastructure required to serve the project, rather than transferring those costs to regional ratepayers.


The project will also use closed-loop, air-cooled cooling systems designed to recirculate water instead of relying on cooling towers that continuously consume water.

OpenAI says that once the cooling system is filled, ongoing water consumption is expected to be comparable to that of an office building serving a similar population, although the final expected water use will be publicly reported after site design is completed.


Energy remains the more fundamental challenge.

The first phase can largely use existing AEP infrastructure, but subsequent development is expected to require new power generation, including natural gas generation, along with additional transmission infrastructure.

That requirement highlights an uncomfortable reality of AI expansion: digital intelligence ultimately depends on physical energy systems.


The PORTS-Pike Project Could Become a Blueprint for AI Factories

NVIDIA and OpenAI plan to collaborate on the design, testing and commissioning of the facility and publish a technical white paper covering lessons from the project.

The focus includes resilient infrastructure design, component qualification and software-level workload management.

That combination is significant because AI cluster reliability cannot be solved solely through better hardware.


When thousands or millions of accelerators operate as a coordinated system, interruptions in one part of the infrastructure can affect workloads across the cluster. Power systems, networking, cooling, hardware reliability and software scheduling must therefore operate as an integrated system.

The future AI factory will increasingly resemble a highly optimized industrial production environment, except its output is intelligence rather than physical goods.


What PORTS-Pike Means for the Future of AI

The Ohio project provides a useful lens through which to understand the next stage of artificial intelligence.

The competitive advantage of AI companies will increasingly depend on the ability to convert capital, electricity, data and computing hardware into reliable intelligence at scale.

That means the industry's critical resources are expanding.

Traditional AI bottlenecks

Emerging AI factory requirements

Algorithms

Compute capacity

Training data

Electricity

GPUs

Land

Model architecture

Data center shell

Software

Transmission infrastructure

Cloud access

Cooling and water systems

Engineering talent

Skilled physical workforce

Capital

Long-term infrastructure financing

The strategic lesson is clear. AI infrastructure is no longer simply a support function behind the technology. It is becoming part of the technology's competitive foundation.

For NVIDIA, securing selected LPS capacity could strengthen its ability to deploy multiple generations of accelerated computing. For OpenAI, the project provides a long-term foundation for frontier training and inference. For Ohio, it represents a potential transition from a historically industrial site toward a new form of digital industrialization.


For the broader AI economy, PORTS-Pike demonstrates that the next generation of intelligence will be built not only in laboratories and software environments, but also on enormous physical sites connected to power grids and industrial supply chains.


The Physical Foundation of the Intelligence Era

The AI industry has spent years emphasizing models, algorithms and chips. The next competitive frontier may be far more physical.

Land must be secured. Power must be generated and transmitted. Data centers must be constructed. Cooling systems must operate efficiently. Networks must remain reliable. Hardware must be upgraded repeatedly. Communities must receive tangible economic benefits.

The PORTS-Pike project brings all of those requirements together in one large-scale development.


Its importance therefore extends beyond the number of GPUs or gigawatts involved. It represents a transition toward AI factories as long-lived industrial assets, capable of supporting successive generations of computing infrastructure.


As Dr. Shahid Masood and the expert team at 1950.ai continue examining the evolution of artificial intelligence and advanced computing, developments such as PORTS-Pike deserve attention because they reveal a fundamental truth about the next phase of AI: intelligence may be digital in form, but its expansion increasingly depends on physical infrastructure.

The companies that can reliably secure that infrastructure may gain an advantage that is as important as the companies building the models themselves.


Further Reading / External References

OpenAI joins PORTS-Pike project

Securing the Infrastructure of Intelligence



Comments

bottom of page