Coverage spread: 2 sources — 2 center
Lean ratings via AllSides / Media Bias-Fact-Check. How this works.
What’s being reported
Nvidia and OpenAI have reportedly held discussions about a massive data center project in southern Ohio that could cost roughly $500 billion in total, according to the Wall Street Journal, as cited by Forbes. The proposed site, located about 50 miles south of Columbus, is being developed by SBEnergy, the energy subsidiary of SoftBank, on federally owned land, and is designed to eventually provide 10 gigawatts of power capacity — a scale that would make it the largest data center project in the world by power output, far surpassing other announced projects.
Under the terms being discussed, Nvidia would provide $250 billion to help OpenAI lease the site, while OpenAI is separately in talks to purchase up to $350 billion worth of Nvidia chips, pushing the full project cost above $500 billion once construction and hardware are combined. The first phase of the site is expected to be completed by 2028, delivering around 800 megawatts of the eventual 10-gigawatt target.
Financing for the underlying power infrastructure has a notable international dimension: Japan has committed $33 billion toward the project as part of a broader pledge to invest in the United States in exchange for reduced tariffs. Under the arrangement described, the U.S. and Japan would split proceeds from power sales at the site until Japan recoups its $33 billion investment, after which the U.S. government would take 90% control of the power revenue. Commerce Secretary Howard Lutnick is reportedly playing a role in determining which companies get access to power at the site, and other major tech firms — including Anthropic, Microsoft, and Google — have also approached Lutnick expressing interest in securing capacity there.
Context: a year of enormous AI infrastructure deals
Forbes situates this potential deal within a broader pattern of enormous capital commitments among AI and cloud giants in 2025. Just before this report, Nvidia announced a separate transaction worth more than $500 billion with SK Hynix to secure AI memory chips, which includes building large-scale data centers expected to be operational by 2027. Other recent moves include Meta’s plan to expand a Louisiana data center to 5 gigawatts with more than $50 billion in total investment, Alphabet’s $80 billion capital raise for AI infrastructure (which included a $10 billion contribution from Berkshire Hathaway), and Amazon’s increased $25 billion investment in Anthropic, paired with Anthropic’s commitment to purchase over $100 billion in Amazon cloud services over the next decade. OpenAI itself had already announced a separate $20 billion, 3.2-gigawatt data center project in Effingham County, Georgia, just a week prior to the Ohio reports — a project the new Ohio site would dwarf by more than three times the power capacity.
To illustrate scale, Forbes notes that the average U.S. household consumed 865 kilowatt-hours per month in 2024, meaning a fully built 10-gigawatt facility could theoretically supply electricity equivalent to roughly 8.4 million homes.
How coverage differs
Forbes’ account, built on Wall Street Journal reporting, focuses heavily on the mechanics and scale of the deal: the financing structure between Nvidia and OpenAI, the U.S.-Japan power arrangement tied to tariff negotiations, the government’s role through Lutnick, and comparisons to other major industry deals to contextualize just how large this proposed project would be relative to existing efforts.
MarketWatch’s headline, by contrast, signals a more skeptical or analytical angle, framing the potential Nvidia-OpenAI arrangement as reviving a concerning pattern reminiscent of past tech bubbles — specifically, the practice of chipmakers or infrastructure providers financing their own customers’ purchases, which can inflate demand signals and obscure underlying business fundamentals. However, the full MarketWatch article text was not available, so the specific analyst commentary, reasoning, and any named individuals behind that bubble warning cannot be detailed here beyond what the headline indicates.
Why it matters
If finalized, this deal would represent one of the largest single infrastructure commitments in the history of the technology industry, underscoring both the immense capital intensity of the AI race and the increasingly entangled financial relationships between chipmakers, AI developers, cloud providers, and now sovereign government interests. The arrangement’s reliance on Japanese government funding tied to tariff negotiations, and the U.S. Commerce Department’s direct involvement in allocating access to the resulting power capacity, also signals a deepening intersection of AI infrastructure development with international trade policy and domestic industrial strategy. At the same time, the scale and circular nature of financing — with Nvidia potentially funding the very company that will buy its chips — has drawn early scrutiny from analysts wary of parallels to prior speculative bubbles in tech investment cycles.
Sources
Featured photo by ThisisEngineering on Unsplash