SoftBank’s 10 GW Ohio Campus Marks a Turning Point for AI Infrastructure

March’s unveiling of a 10-gigawatt AI data center campus at the former Portsmouth Gaseous Diffusion Plant in Pike County, Ohio is more than another large-scale AI factory announcement. It offers one of the clearest signals yet of where AI infrastructure is headed.

At this scale, data centers are no longer being planned as real estate projects anchored by available utility service. They are being conceived as integrated systems, where power generation, transmission, and compute are developed in parallel as a single coordinated platform.

The result is a new class of infrastructure that looks less like a data center campus and more like a strategic utility-and-industry complex.

The Emergence of the AI Utility Model

Under the plan, SB Energy, the SoftBank-affiliated developer, will pair a 10 GW compute campus with up to 10 GW of new power generation, including 9.2 GW of natural gas capacity, alongside $4.2 billion in transmission investments in partnership with AEP Ohio.

The U.S. Department of Energy is leasing federal land for the project and positioning it as a national model for expanding AI capacity while limiting the burden of grid-upgrade costs on ratepayers. In announcing the project, DOE described it as an effort to “revitalize federal land in southern Ohio to advance energy and technology infrastructure and lower energy prices.”

The scale alone places the development in a category of its own. A 10 GW campus is not simply larger than a traditional hyperscale buildout. It is large enough to force a rethinking of how AI infrastructure is financed, sited, powered, and politically justified. Reuters described the project as a “colossal” pairing of gas-fired generation and AI infrastructure on federal land, while the Associated Press reported that construction could begin this year at what DOE is now branding the PORTS Technology Campus.

The project also carries an explicit geopolitical dimension. DOE said the generation plan includes $33.3 billion in Japanese-backed funding tied to the natural gas component, while Reuters linked that financing to the broader U.S.-Japan Strategic Trade and Investment Agreement. SoftBank framed the initiative as the launch of the Portsmouth Consortium under that agreement, identifying a group of Japanese and U.S. companies expected to participate in the development.

A groundbreaking ceremony for the campus was held on March 20, 2026 in Piketon, Ohio. Early filings indicate an initial 800 MW phase has been submitted for interconnection, with an operational target of 2028.

From Data Center Campus to Power-Anchored Development

The Ohio project treats AI load not simply as a customer of the grid, but as the anchor for a purpose-built power district. While elements of this approach have emerged in recent large-scale developments, the scale and level of integration on display in Piketon mark a significant escalation.

For years, data center development followed a familiar sequence: secure land, lock in incentives, negotiate for utility service, and expand in phases as power becomes available. That model is now under strain. AI training and inference clusters are pushing project sizes beyond what many utilities can support on conventional timelines, particularly in regions constrained by transmission capacity.

The Piketon project reflects a different approach. Rather than waiting for the grid to catch up, developers are proposing generation, transmission, substations, and pipeline infrastructure as part of the initial buildout. This shift is already beginning to influence how utilities evaluate large-load interconnections, including projects that propose to bring their own power behind the meter.

DOE’s fact sheet indicates the plan includes new 765-kV transmission lines, four substations, and interstate gas pipeline development, with long-lead high-voltage electrical equipment already secured. These are not incremental upgrades. They are foundational elements of a new infrastructure model.

The result is something closer to a vertically integrated AI utility zone than a conventional data center campus. The project is being designed around dedicated energy systems and purpose-built grid infrastructure, not a standard interconnection request.

At gigawatt scale, that distinction becomes decisive. The next generation of AI campuses will be defined not just by access to land and capital, but by the ability to secure firm power, transmission capacity, critical electrical equipment, and regulatory alignment in parallel.

From Cold War Asset to AI Industrial Platform

DOE’s fact sheet frames the redevelopment as a transformation of a former Cold War national security asset into a new engine for American AI leadership. The department is leasing federal land at the Portsmouth Gaseous Diffusion Plant to an SB Energy-affiliated entity, with SB Energy also committing to accelerated cleanup and remediation at the site. DOE’s language is notably direct: Portsmouth, once a cornerstone of national defense through uranium enrichment, is now being positioned to help the United States “win the AI race.”

That framing is revealing. It signals how Washington increasingly views AI infrastructure, not simply as commercial capacity, but as strategic industrial capability. In that context, data center development begins to resemble earlier eras of federal focus on energy systems, semiconductor manufacturing, and defense supply chains.

Reuters noted that the site produced enriched uranium for weapons during the Cold War and that billions of dollars have already been spent on decontamination. The Associated Press reported that Portsmouth is one of 16 federal sites identified by DOE as candidates for future data infrastructure development. Taken together, these signals point to a broader federal posture: the reuse of controlled industrial land as a platform for large-scale AI deployment.

Piketon is therefore more than an available parcel. It is a federally controlled site with industrial legacy, existing grid proximity, redevelopment momentum, and symbolic weight. That combination makes it a compelling template for the next phase of federal land reuse in support of AI and advanced computing infrastructure.

Power as the Primary Constraint

The defining issue in AI infrastructure is no longer compute. It is power.

The Ohio project makes that explicit. The plan pairs a 10 GW data center campus with up to 9.2 GW of natural gas generation, creating a combined system that Reuters describes as among the largest integrated power-and-compute developments in the world. AEP Ohio has indicated that initial power delivery to the site is expected later in the decade.

At this scale, power, not processors, becomes the gating constraint. The industry continues to focus on GPU supply, interconnect architectures, and liquid cooling roadmaps. But those considerations are secondary if firm gigawatts cannot be delivered on commercially viable timelines. The Ohio development is structured around that reality, bringing generation, transmission, and compute online as a coordinated system rather than a sequential process.

This is the emergence of a power-first model for AI campuses. Instead of adapting to grid availability, projects of this scale are increasingly being designed around dedicated energy supply from the outset. That shift reflects a broader recalibration across the industry, where access to power is becoming the primary determinant of where and how AI infrastructure can be deployed.

The reliance on natural gas will be controversial, but it reflects current constraints on speed, scale, and reliability.

Dispatchable gas generation remains one of the few options capable of delivering multi-gigawatt capacity within the timeframes AI developers are targeting. Nuclear continues to attract interest as a long-term solution, but deployment timelines remain extended. Industry signals reinforce this trend, with turbine manufacturers such as GE Vernova reporting a growing share of demand tied to data center and AI-driven power requirements.

Renewables can reduce carbon intensity, but they cannot independently meet the need for continuous, multi-gigawatt firm capacity without large-scale storage and balancing resources. For developers targeting guaranteed availability within this decade, natural gas remains the most readily deployable option, despite the political and environmental tradeoffs it introduces.

AEP and the Cost Allocation Model

If the generation plan explains the engineering logic, the AEP structure speaks to the political one. At the center is one of the most contested questions in the data center market: who pays for the transmission and grid upgrades required to serve large new loads?

Utilities, regulators, consumer advocates, and large-load customers are increasingly divided on this issue. Data center developers point to economic development benefits, including jobs and tax revenue. Consumer advocates counter that residential ratepayers should not subsidize infrastructure built primarily to serve hyperscale demand.

The Ohio arrangement is being positioned as a response to that conflict. DOE states that SB Energy and AEP Ohio are partnering on $4.2 billion in new transmission infrastructure, with SB Energy committing to fund those investments rather than passing costs through to ratepayers. AEP has echoed that position, indicating the structure is intended to avoid upward pressure on transmission rates for Ohio customers.

Whether that outcome holds will depend on regulatory review and execution. But the structure itself is significant. It frames a model in which large-load developers directly fund the transmission infrastructure required to support their projects, rather than relying on broader cost recovery mechanisms.

That makes the project more than a construction milestone. It positions it as a potential policy template. If validated, this approach could influence how utilities and regulators across the U.S. address cost allocation for AI-scale infrastructure, particularly as similar disputes intensify in constrained grid regions.

Why 765-kV Transmission Signals Scale

AEP says the project will require new 765-kV transmission infrastructure. This is not conventional distribution or even routine bulk-power expansion. Lines at this voltage class can carry significantly more capacity than standard 345-kV transmission, and AEP is one of the nation’s largest operators of 765-kV systems. Route planning is already underway, with the Ohio Power Siting Board overseeing permitting, including public input, environmental review, and land-use approvals.

The implications are substantial. First, the use of 765-kV transmission reflects the scale and credibility of the underlying load. Infrastructure of this class is not planned for speculative demand; it is built to support sustained, multi-gigawatt consumption.

Second, it highlights where the real complexity of the project resides. The challenge is not the construction of data halls. It is the coordination of transmission, substations, pipelines, permitting, environmental review, and large-scale generation within a tightly sequenced development program.

At this level, AI infrastructure becomes an exercise in system orchestration. The Ohio project is as much an infrastructure choreography problem as it is a computing deployment.

SoftBank as Infrastructure Orchestrator

Data Center Frontier has previously reported on SoftBank’s collaboration with OpenAI and Oracle on Stargate, a long-term U.S. AI infrastructure initiative that has been associated with investment projections approaching $500 billion. While the Piketon project has not been formally identified as part of that effort, the alignment in scale, participants, and regional focus is notable.

More broadly, SoftBank’s role in Ohio reflects a shift in posture. The company is not operating as a passive capital provider. It is positioning itself as an orchestrator across compute demand, international financing, and energy infrastructure.

Reuters reported that SoftBank CEO Masayoshi Son described the project as strengthening U.S. AI leadership while securing long-term energy and compute capacity. In remarks reported by The Register, Son added:

“AI will transform every industry, and the PORTS Technology Campus will help deliver the next-generation infrastructure needed to unlock those breakthroughs.”

The structure of the financing reinforces that positioning. The gas-generation component is tied to Japanese-backed capital under the U.S.-Japan Strategic Trade and Investment Agreement, linking the project to a broader framework of bilateral economic cooperation.

That dynamic moves the Ohio development beyond a conventional project-finance model. It places it within the context of coordinated industrial policy, where capital, energy, and compute infrastructure are aligned across national boundaries. In that sense, AI infrastructure is increasingly being treated as a strategic asset class, with investment decisions extending beyond technology into energy security and geopolitical positioning.

Economic Promise and Regional Reality

For Ohio, and particularly Appalachian Ohio, the project is being framed as redevelopment on a historic scale. It represents more than the cleanup of a decommissioned nuclear-era site. DOE states that the Portsmouth redevelopment could generate more than 10,000 construction jobs over four years, more than 2,000 operational roles, and tens of thousands of additional indirect jobs across manufacturing and service sectors. Ohio is already an established data center market, with AWS, Google, Vantage, Aligned, and other operators expanding their presence in the state.

That economic narrative carries weight, especially in a region shaped by industrial decline, long-term remediation work, and the search for durable reinvestment. The transformation of a former uranium enrichment site into a large-scale AI and energy hub aligns with a broader policy goal: positioning AI not only as a software-driven industry, but as a catalyst for physical and industrial renewal.

At the same time, the headline numbers warrant scrutiny. Job creation estimates tied to projects of this scale are often concentrated in the construction phase and diffused across indirect categories. The long-term regional impact will depend on how much of the operational, technical, and supply chain activity remains anchored locally once the initial buildout phase subsides.

The Social and Political Constraint

This points to the project’s most immediate fault line: not whether it is ambitious, but whether public acceptance can keep pace with that ambition.

The Associated Press reported that the announcement came just days after rural Ohio residents filed a petition seeking a statewide constitutional ban on large-scale data centers. That timing is significant. The project is entering a landscape where opposition to data center development is no longer limited to isolated zoning disputes. Concerns now extend to energy consumption, electricity pricing, land use, water resources, emissions, and the broader societal value of AI infrastructure.

The Portsmouth model addresses one dimension of that debate by structuring transmission costs to be borne by the developer rather than ratepayers. But that does not resolve the full set of issues. Questions remain around air emissions, expanded gas infrastructure, permitting impacts, and the broader justification for a project of this scale.

In that sense, the development may serve as a test case. Not only for a new model of AI infrastructure delivery, but for the industry’s ability to sustain public support as projects move into the multi-gigawatt era. The outcome will likely shape how developers, policymakers, and communities engage on future AI deployments of similar scale.

A Threshold Moment for AI Infrastructure

The Piketon announcement marks a threshold moment. What SoftBank, SB Energy, AEP Ohio, and the U.S. Department of Energy have outlined is not simply a large-scale campus. It is a working model of what AI infrastructure becomes when power constraints, political scrutiny, and geopolitical priorities converge.

At this scale, data centers are no longer discrete facilities. They are evolving into integrated energy-and-compute systems, where land, generation, substations, transmission corridors, and regulatory structures carry as much strategic weight as processors and networks.

If the project advances, its significance will extend beyond a single site. It has the potential to shape how the next generation of AI infrastructure is financed, powered, and approved. It points toward a model built on dedicated energy supply, direct investment in transmission, and the reuse of federally controlled or industrial land.

It also raises a parallel question. Whether that model can scale without triggering sustained public resistance. Shielding ratepayers from infrastructure costs addresses one dimension of the challenge, but it does not resolve broader concerns around environmental impact, land use, and the societal footprint of AI development.

The PORTS Technology Campus places a marker on the landscape. The next phase of AI will not be constrained by demand or innovation. It will be constrained by execution at the intersection of power, infrastructure, and public acceptance.

And in Piketon, the industry may be seeing one of the first full-scale attempts to build that system from the ground up.

At Data Center Frontier, we talk the industry talk and walk the industry walk. In that spirit, DCF Staff members may occasionally use AI tools to assist with content. Elements of this article were created with help from OpenAI's GPT5.

Keep pace with the fast-moving world of data centers and cloud computing by connecting with Data Center Frontier on LinkedIn, following us on X/Twitter and Facebook, as well as on BlueSky, and signing up for our weekly newsletters using the form below.