Nvidia Is No Longer Just Selling AI Chips — It Is Financing The AI Economy
OpenAI’s Giant Ohio Data Centre Could Become A $200 Billion Nvidia Machine
Nvidia’s $105 Billion OpenAI Gamble Could Reshape The AI Arms Race
Nvidia is putting its financial strength behind one of the most ambitious infrastructure projects of the artificial-intelligence era, agreeing to provide credit support that could reach $105 billion for a vast Ohio data-centre campus that will serve OpenAI. Nvidia will also invest $1.5 billion directly into developer SB Energy, while its technology is set to become the exclusive AI computing infrastructure deployed at the site.
The scale explains why the deal matters far beyond another large GPU order. OpenAI has agreed to use as much as 8 gigawatts of IT capacity at the PORTS-Pike Technology Campus in Pike County under a 20-year lease, while Nvidia says the initial deployment alone is designed for 4.25 IT-GW. The project is expected to begin coming online in phases from 2028.
Nvidia Is Moving Beyond Selling Chips
For years Nvidia’s extraordinary AI growth has depended on customers finding the money, electricity, buildings and engineering capacity required to install increasingly vast amounts of computing hardware. The Ohio agreement shows Nvidia becoming much more directly involved in solving those constraints.
Nvidia says it is securing what it calls land, power and shell capacity so that the physical infrastructure is available for its computing systems. The company will provide credit support covering the initial 4.25 IT-GW deployment and has the option to participate in the remaining 3.75 IT-GW.
That changes Nvidia’s role. It is still a semiconductor company, but increasingly it is also using its balance sheet and market position to help create the infrastructure into which those semiconductors will be sold.
Chief executive Jensen Huang has framed the strategy around a simple problem: AI demand can grow faster than the supply of suitable sites, power connections and financing. Nvidia therefore has an incentive to remove those bottlenecks before they limit the number of GPUs it can ultimately sell.
The Ohio Project Is Enormous
SB Energy plans to build and operate the PORTS-Pike campus, with OpenAI as the customer under a 20-year lease. Nvidia will supply the full-stack computing infrastructure, including GPUs, CPUs and networking through its DSX AI factory platform.
The wider development is expected to require at least 10 GW of new energy generation to support approximately 8 IT-GW of computing capacity. SB Energy and SoftBank also plan at least $4.2 billion of regional electricity-grid investment alongside AEP Ohio.
For perspective, this is not simply another collection of server buildings. It is industrial-scale infrastructure being designed around artificial intelligence, tying together power generation, grid investment, construction, networking and successive generations of Nvidia processors.
The campus is being developed around the decommissioned Portsmouth Gaseous Diffusion Plant and surrounding land in southern Ohio, with involvement from the US Departments of Energy and Commerce. Nvidia says tens of thousands of jobs could be associated with the development, while the companies have established an $80 million community-benefits commitment.
Why Nvidia Could Make Far More Than It Risks
The headline number can initially look alarming. A potential $105 billion guarantee is enormous even for Nvidia.
But it would be misleading to treat the arrangement as Nvidia simply writing OpenAI a $105 billion cheque.
The guarantee relates to infrastructure financing and the value of the project rather than Nvidia immediately paying the entire amount. The structure is designed to help SB Energy secure the capital needed to construct the facilities while limiting Nvidia’s direct upfront expenditure. Nvidia’s immediate equity investment in SB Energy is $1.5 billion.
The commercial prize could be much larger.
Huang has indicated that the initial 4.25-GW Ohio deployment could ultimately generate as much as $200 billion in Nvidia revenue. Across Nvidia’s wider planned relationship with OpenAI, including as much as 16 GW of computing capacity, Nvidia could generate approximately $600 billion in revenue from OpenAI by 2030 if those deployments occur at the scale currently contemplated.
That is the fundamental logic of the transaction.
Nvidia is effectively using part of the financial power created by the AI boom to help finance the infrastructure required for the next stage of the AI boom.
If the infrastructure is completed, OpenAI expands as expected and successive generations of Nvidia hardware continue replacing older systems, the economic return to Nvidia could dwarf the money initially committed.
The Biggest Risk Is OpenAI
The arrangement is not risk-free.
OpenAI is committing itself to extraordinary amounts of computing capacity over decades. The economics ultimately depend on AI services generating enough revenue to support enormous spending on chips, buildings and electricity.
Nvidia’s guarantee means it is taking greater exposure to that ecosystem. If OpenAI failed to meet its obligations, Nvidia could ultimately become responsible for part of the difference between the guaranteed value of the infrastructure and whatever value could be recovered through another tenant or sale.
There is therefore an important distinction between demand being genuinely generated by profitable AI usage and demand being partially sustained because Nvidia, OpenAI, infrastructure developers and financial institutions are financing each other.
That distinction has become increasingly important to investors.
Earlier discussions reportedly contemplated a Nvidia backstop of as much as $250 billion before the structure was reduced substantially. The eventual commitment of up to $105 billion therefore appears considerably more restrained than some of the arrangements previously under discussion.
The Circular-Financing Question
The uncomfortable part of Nvidia’s strategy is that the company increasingly benefits from helping customers obtain the resources required to purchase Nvidia technology.
That creates an obvious criticism.
If Nvidia finances infrastructure, the infrastructure buys Nvidia systems, and the resulting Nvidia revenue strengthens Nvidia’s ability to finance additional infrastructure, the system can begin to look circular.
Huang has rejected the description of the Ohio deal as circular financing, arguing that Nvidia is using its scale and visibility into long-term demand to secure infrastructure that otherwise represents a bottleneck.
There is merit to that argument. Financing a factory that will use your equipment is not automatically evidence of artificial demand. Industrial companies have long offered financing, guarantees and other support to stimulate deployment of their products.
The risk appears if the end customer cannot eventually generate sufficient economic returns from the equipment.
That is why OpenAI’s revenue growth, cash generation and eventual ability to finance its enormous infrastructure commitments will matter increasingly to Nvidia shareholders.
Why The Deal Is Strategically Important
The Ohio agreement may reveal something more important than its $105 billion headline.
Nvidia appears determined to stop physical infrastructure constraints from becoming the ceiling on AI growth.
The semiconductor bottleneck that dominated the early generative-AI boom is evolving. Power generation, substations, transmission capacity, available land, construction and financing are becoming increasingly important constraints.
Nvidia is responding by moving upstream.
Rather than waiting for hyperscalers and AI companies to build facilities and then competing for the chip orders, Nvidia can help secure the facility itself and ensure Nvidia systems occupy it.
At PORTS-Pike, Nvidia is expected to be the exclusive AI compute infrastructure provider.
That exclusivity matters enormously. An AI campus requiring several gigawatts of computing capacity represents years of potential demand for accelerators, networking equipment, CPUs, software and replacement hardware.
The buildings may last decades, but Nvidia’s computing systems inside them will be upgraded repeatedly.
That means one secured data-centre campus could potentially generate several hardware cycles.
What It Means For Nvidia Stock
For Nvidia shareholders, the announcement is strategically bullish but financially more complicated than a straightforward new customer order.
The positive case is powerful. Nvidia has effectively secured a gigantic future deployment site for its technology, strengthened its relationship with one of the world's largest AI customers and created the possibility of hundreds of billions of dollars of future revenue if OpenAI’s expansion proceeds as planned.
It also demonstrates confidence from Nvidia management that demand for AI computing will remain enormous deep into the next decade.
The negative case is concentrated around risk.
Nvidia is taking on financial exposure that traditional semiconductor companies normally would not. If increasingly large amounts of Nvidia’s future sales depend on investments, guarantees or financing supplied by Nvidia itself, investors could begin assigning a discount to the quality of those revenues.
The stock-market question is therefore not simply whether Nvidia can sell more GPUs.
It is whether the extraordinary amount of computing capacity being financed today eventually produces enough economically valuable AI activity to justify the investment.
The Short-Term Stock Impact Could Be Smaller Than The Headline
A $105 billion headline sounds large enough to transform Nvidia overnight, but the immediate share-price effect may be considerably more restrained.
Nvidia is already valued on expectations of vast continuing AI expenditure. Another enormous OpenAI commitment reinforces that thesis, but investors have become increasingly aware that future AI infrastructure announcements must eventually translate into sustainable revenue and free cash flow.
The strongest bullish element is the potential revenue attached to the project. If an initial 4.25-GW deployment really can represent as much as $200 billion of Nvidia revenue, the economics are potentially exceptional relative to Nvidia’s $1.5 billion direct SB Energy investment.
The principal bearish element is balance-sheet exposure.
Investors could reasonably ask how many similar guarantees Nvidia might ultimately provide, what happens during an AI spending downturn and whether infrastructure designed around extremely high-density Nvidia systems would retain sufficient value if one of the industry's largest customers ran into financial trouble.
Those concerns are unlikely to disappear.
The Longer-Term Stock Argument Is More Important
For Nvidia stock over several years, the Ohio agreement strengthens one of the company's most important advantages: Nvidia is attempting to control not only the computing platform but increasingly the conditions required for that platform to exist.
That could make its competitive moat wider.
AMD and custom chips from companies such as Google, Amazon and other hyperscalers can compete with Nvidia on hardware. It is much harder to compete with an ecosystem involving GPUs, CPUs, networking, software, infrastructure design, financing relationships and physical data-centre capacity operating together.
The greater danger is that Nvidia becomes so deeply embedded in financing the AI ecosystem that investors struggle to separate organic customer demand from demand supported by Nvidia's own capital.
Both possibilities can be true simultaneously.
The Ohio deal can be an exceptionally valuable commercial agreement and still increase the financial risk surrounding Nvidia.
What Happens Next
The first major milestone is 2028, when the initial PORTS-Pike capacity is expected to begin coming online. Construction and power development will then continue in phases as the campus moves towards its planned 8 IT-GW scale.
Investors should watch three numbers above everything else: how quickly the 4.25-GW first phase is built, how much Nvidia hardware revenue actually results from it and whether Nvidia exercises its option over the additional 3.75 GW.
The most important question, however, sits above the Ohio project itself.
Nvidia has spent the first phase of the AI revolution becoming the company that supplies the machines. It is now using its extraordinary financial strength to help build the world in which those machines will operate.
If AI demand continues compounding, the strategy could lock Nvidia into hundreds of billions of dollars of additional revenue and make its dominance even harder to challenge.
If the economics of AI fail to justify the infrastructure being built around it, the same agreements could expose Nvidia to risks that barely existed when it was simply selling chips.
That makes the $105 billion Ohio commitment far more important than another giant data-centre announcement. It is a glimpse of Nvidia's next business model — and perhaps the biggest test yet of whether the AI boom can finance the physical world it now requires.
