The 10-Gigawatt Ohio Megaproject
Nvidia and OpenAI are in advanced discussions to build a data center with a staggering 10 gigawatts of capacity on federal land about 50 miles south of Columbus, Ohio. The first phase, slated for completion in 2028, would deliver roughly 800 megawatts—already a massive facility—before expanding over subsequent years to its full capacity, enough electricity to power about 8.4 million households.
The financial scale is unprecedented. Nvidia would invest $250 billion into the project, enabling OpenAI to lease the computing capacity. OpenAI would then buy up to $350 billion worth of Nvidia’s chips, pushing the total commitment above $500 billion. Japanese conglomerate SoftBank’s energy subsidiary, SBEnergy, is developing the site, while Japan itself has allocated $33 billion for energy infrastructure as part of a broader U.S. investment pact.
The Wall Street Journal, which first reported the talks, cautions that no deal is final, and both the scope and financial structure could change. Still, even a scaled-down version would eclipse today’s largest data centers, underscoring how profoundly AI development is migrating from code-centric labs to colossal physical assets.
What This Superscale Bet Means for the AI Industry
The Nvidia-OpenAI Symbiosis
For Nvidia, the project is a factory for trillion-dollar chip sales, locking its most valuable AI customer into an exclusive pipeline. For OpenAI, it secures the raw compute needed to train next-generation models without ceding control to public cloud rivals. This vertical integration—hardware, infrastructure, and model developer under one roof—creates a closed loop that competitors will find difficult to replicate.
An AI Arms Race Running on Gigawatts
The Ohio project is not an outlier; it represents the new normal. Meta is expanding a Louisiana data center to 5 gigawatts, Alphabet has earmarked tens of billions for infrastructure, and Amazon increased its Anthropic investment to $25 billion. When 10 GW becomes the benchmark, only a handful of giants can play, concentrating power—literally and economically—in a narrow band of firms and their suppliers.
From Software to Heavy Industry
This deal signals that the AI story is no longer primarily about algorithms. It is about securing land, power, transmission lines, and cooling systems at a scale previously reserved for aluminium smelters or entire national grids. Regulators and communities will soon confront the reality that a single AI data center can consume more power than many midsized cities, raising profound questions about grid reliability, electricity prices, and climate commitments.
The Japanese Angle
Japan’s $33 billion infrastructure commitment through SoftBank’s SBEnergy adds a geopolitical layer. It ties U.S.-Japan technology cooperation directly to critical energy assets and positions Japanese capital as a key enabler of American AI dominance, while also giving SoftBank a strategic foothold in the next wave of global compute infrastructure.
Implications for the Tech Sector, Investors, and Policymakers
- For tech leaders: The 10 GW figure resets expectations. Firms without a secured, multi-gigawatt footprint risk being priced out of frontier AI within the decade. Site selection, power purchase agreements, and regulatory relationships are now core strategic functions.
- For investors: Monitor Nvidia’s capex and chip-sale disclosures—a finalised deal of this magnitude would lock in demand visibility for years. Smaller data center operators and power equipment suppliers may see a super-cycle, but concentration risk is acute if hyperscalers capture the bulk of new capacity.
- For energy and utility sectors: A single 10 GW colossus in Ohio will raise local wholesale power prices and strain transmission infrastructure. Expect heated debates over cost allocation, grid upgrades, and whether AI data centers should shoulder a larger share of system improvements.
- For policymakers: This project interweaves trade, energy, and national security. Site permitting on federal land, foreign investment from Japan, and the environmental impact of a 10 GW facility will all require coordinated federal and state oversight. A clear national framework for AI infrastructure siting becomes urgent before ad hoc projects overwhelm local grids.
Risk & Opportunity Assessment
| Commercial Risk | High | The $500B+ undertaking depends on a final agreement between Nvidia and OpenAI that is not yet signed, and even then must clear permitting, energy supply contracts, and construction timelines that are inherently uncertain. Cost overruns or delays could turn the projected return on such a facility sharply negative. |
| Competitive Risk | High | If Meta, Alphabet, or Amazon accelerate their own multi-gigawatt builds first, OpenAI’s exclusive compute advantage could erode, and Nvidia’s chip pipeline could be diverted to other buyers willing to pay a premium. |
| Regulatory Risk | Critical | The site is on U.S. federal land, and supplying 10 GW of power will require unprecedented interconnection studies, environmental reviews, and possibly new legislation. Opposition from local communities, grid operators, or federal agencies could block or drastically alter the project. |
| Reputation Risk | Medium | A facility consuming power equivalent to millions of homes will attract intense scrutiny over climate impacts and equity. If linked to fossil fuel generation, Nvidia, OpenAI, and SoftBank risk a consumer and political backlash against ‘AI-driven’ carbon emissions. |
| Technology Disruption | Transformational | A 10 GW data center, purpose-built for AI, could fundamentally reset the scale at which training is done, accelerating a shift to large clusters that render smaller or less energy-efficient architectures obsolete. Conversely, if novel, more efficient chip designs emerge before the 2028 first phase, the center’s capacity could be overengineered. |
| Commercial Opportunity | Transformational | For Nvidia, it guarantees chip sales exceeding the GDP of many countries. For OpenAI, locking in that compute could secure an insurmountable lead in foundation models. For SBEnergy and Japanese investors, it establishes a pivotal position in the infrastructure layer of the global AI economy. |
Comments 0