Nvidia agrees to spend up to $105 billion on an Ohio data-center campus for OpenAI

Nvidia has agreed to invest up to $105 billion in a new data-center campus in Ohio that OpenAI is expected to lease, according to a financial filing cited by Yahoo Finance. The plan, if carried out at the top end of its range, would represent one of the largest industrial-scale bets Nvidia has made tied to specific AI demand.

The arrangement centers on a large data-center footprint in Ohio. Instead of limiting its commitments to selling graphics processing units (GPUs) and related systems, Nvidia’s agreement links capital spending directly to the build-out of compute capacity that OpenAI will use. Nvidia, which supplies much of the hardware used to train and run frontier AI models, has increasingly been pulled into the “infrastructure layer” of AI demand.

Yahoo Finance reported that the agreement calls for Nvidia to spend as much as $105 billion to support the data center campus, with OpenAI planned as the tenant via a lease. The report frames the news as another sign of expanding partnerships between the two companies as AI workloads grow and data-center capacity becomes a gating factor for deployment timelines.

While the filing referenced by Yahoo Finance sets a spending cap, the report did not provide granular detail in the information available here on the exact timing of construction, the size of the campus in megawatts, or the schedule of the lease payments. It also did not clarify whether the full $105 billion is contingent on milestones such as occupancy, power availability, equipment delivery, or performance terms.

The deal highlights a structural shift in the AI supply chain. GPUs are only one component of a data center, and the bottlenecks for large AI deployments often include power delivery, cooling, networking infrastructure, and rapid construction. By funding a purpose-built campus and tying it to a known customer via leasing, Nvidia can potentially reduce uncertainty for both its own customers and its own manufacturing and integration pipeline.

For Nvidia, the financial logic extends beyond revenue from hardware. Large data-center projects can shape the trajectory of long-term demand for Nvidia’s full stack, including data-center platforms, networking, and software that supports AI workloads. For OpenAI, securing compute capacity through a dedicated facility can help de-risk operations and support scaling as model usage grows.

As with many major AI and data-center announcements, the key unknowns are in the terms Nvidia and its partners have not disclosed in the information available here. The Yahoo Finance report points to a financial filing but does not provide, in the excerpted context, details such as project phasing, tenant exclusivity, caps or floors on the spending, or the degree to which the company’s obligation changes if construction delays or demand differs from expectations.