NVIDIA announced a $105 billion financing commitment for OpenAI's Ohio data center campus, marking one of the largest infrastructure deals in technology history. The arrangement involves NVIDIA funding land acquisition, power procurement, and building construction for a facility that could eventually house 1.5 million graphics processing units (GPUs) across 8 gigawatts of computing capacity. The deal structure reveals how AI infrastructure is being financed in 2026, but it also raises questions about whether NVIDIA is essentially funding its own future chip sales.

How Does This Three-Way Partnership Actually Work?

The PORTS-Pike Technology Campus represents a new model for AI infrastructure development. SB Energy will build and own the facility, while OpenAI will lease it for 20 years. NVIDIA's role is to finance the physical prerequisites that make construction possible. This arrangement differs from traditional cloud infrastructure, where companies like Microsoft, Amazon, and Google fund their own data centers entirely.

• NVIDIA's Financing Role: The chip maker provides over $105 billion to secure land, power procurement, and building shells that enable construction to begin.
• SB Energy's Ownership: The energy company builds and owns the completed facility, managing long-term operations and maintenance.
• OpenAI's Lease Agreement: The AI company commits to a 20-year lease, securing computing capacity for its operations and product development.

NVIDIA also made a separate $1.5 billion equity investment in SB Energy as part of the arrangement. Combined with prior agreements between NVIDIA and OpenAI, the total deal value could reach as much as $600 billion.

Why Is Wall Street Questioning This Deal?

The elephant in the room is straightforward: NVIDIA invests in infrastructure, which then buys NVIDIA's chips. That pattern looks like circular investing, a practice that raises red flags about inflated revenue figures and whether the company is artificially boosting its own sales pipeline. The concern is that NVIDIA's investment reduces barriers for OpenAI to purchase more NVIDIA hardware, creating a self-reinforcing cycle.

NVIDIA CEO Jensen Huang addressed this criticism directly in a blog post, arguing that the company "uses its scale and long-term visibility to secure PORTS-Pike to host NVIDIA compute." He compared the approach to supply-chain management, where companies secure critical inputs when they have clear visibility into customer demand. Huang also clarified that NVIDIA will only cover portions of lease and power payments, not the full cost of the site or OpenAI's operational obligations.

Jensen Huang

"This is the same discipline we apply to supply-chain management: we secure critical inputs when we have visibility into customer demand and when doing so enables long-term productive capacity," stated Jensen Huang.

Jensen Huang, CEO at NVIDIA

The distinction matters to some observers but may not satisfy skeptics. NVIDIA isn't handing cash directly to OpenAI to buy GPUs. Instead, it's financing the physical infrastructure that makes GPU deployment possible. The chips themselves are a separate transaction. However, the practical effect remains the same: NVIDIA's investment reduces the barriers for OpenAI to buy more NVIDIA hardware.

What Does This Mean for AI Infrastructure Going Forward?

The PORTS-Pike campus deal landed just days after NVIDIA announced a separate $500 billion fund with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to finance AI data center construction globally. The company has also invested in Ilya Sutskever's Safe Superintelligence startup and SK Hynix, the memory chip manufacturer critical to GPU production.

The pattern reveals a broader strategy: NVIDIA is building an ecosystem where every layer of the AI infrastructure stack has NVIDIA's fingerprints on it. The company wants to be the landlord, the builder, and the equipment supplier simultaneously. Each new deal makes the next one easier because NVIDIA can point to its growing infrastructure portfolio as proof of committed demand.

The Ohio campus represents a significant shift in how AI infrastructure gets built and financed. By financing land and power procurement upfront, NVIDIA removes months or years of lead time from the development process. For OpenAI, it also represents a step toward controlling its own infrastructure rather than relying entirely on cloud provider capacity. The facility's initial plans call for 4.25 gigawatts of computing capacity, with options to scale to 8 gigawatts total. To put that in perspective, 8 gigawatts exceeds the electrical output of most nuclear power plants.

What Are the Risks and Revenue Implications?

NVIDIA estimates the facility could house approximately 1.5 million GPUs across its full 8-gigawatt capacity. At current pricing, that represents $150 billion to $200 billion in potential GPU revenue per generation deployed through 2030.

The risk for NVIDIA is concentration. A significant portion of its revenue growth thesis now depends on a small number of massive data center projects reaching completion and operational status. If power availability, permitting, or construction delays slow the PORTS-Pike timeline, the GPU orders associated with the facility get pushed back too. This creates a dependency where NVIDIA's financial performance becomes tied to infrastructure development timelines beyond its direct control.

The data center deals will likely dominate discussion during NVIDIA's upcoming earnings calls. Analysts will scrutinize whether the infrastructure financing commitments translate to firm GPU orders or remain contingent on construction timelines. Revenue recognition depends on actual chip deliveries, not financing arrangements, which means NVIDIA must ensure these facilities reach operational status to realize the projected revenue.