One company develops some of the world’s most influential artificial-intelligence models. The other supplies the computing platform needed to train and operate them. Together, NVIDIA and OpenAI are turning AI into an industrial project measured in gigawatts, data centres and billions of dollars.
The development of artificial intelligence is often described as a race to create the most capable model. Yet software is only one part of that competition. Every advanced AI system depends on a physical foundation: processors, high-speed networking, storage, cooling equipment and enormous supplies of electricity.
The relationship between NVIDIA and OpenAI connects these two layers of the industry.
OpenAI creates models and services used by consumers, developers and businesses. NVIDIA develops the accelerated-computing systems on which many of those models are trained and deployed. Their cooperation began years before ChatGPT became a global product, but it has now expanded into a much larger infrastructure alliance.
A Partnership Measured in Gigawatts
In September 2025, the companies announced a letter of intent under which OpenAI would deploy at least 10 gigawatts of NVIDIA systems. The planned infrastructure could involve millions of GPUs supporting future model training and inference—the process of running trained models for users.
NVIDIA also stated that it intended to invest up to $100 billion in OpenAI progressively as each gigawatt was deployed. The companies identified NVIDIA’s Vera Rubin platform as the technology expected to support the first gigawatt, targeted for the second half of 2026.
The language matters. The announcement described an intended strategic partnership whose investments would occur in stages, not an immediate transfer of $100 billion. Each phase depends on new computing capacity actually being developed.
Ten gigawatts is far more than a conventional order for computer chips. It implies a network of large data-centre campuses supported by power generation, substations, cooling systems and extensive construction.
At this scale, artificial intelligence becomes as much an energy and infrastructure business as a software business.
Why OpenAI Needs So Much Computing Power
Modern AI infrastructure performs two major tasks.
Training requires vast clusters of processors to analyse data and adjust a model’s parameters. Inference begins after the model is trained and consumes computing power whenever a user requests an answer, image, analysis or automated action.
The popularity of AI services means inference demand can grow continuously. A successful model does not reduce the need for computing; it can create additional demand as more people and organisations begin using it.
In February 2026, OpenAI said it was expanding its work with NVIDIA through three gigawatts of dedicated inference capacity and two gigawatts of training capacity based on Vera Rubin systems.
This builds on Hopper and Blackwell infrastructure already operating through providers including Microsoft, Oracle Cloud Infrastructure and CoreWeave.
For OpenAI, dependable access to processors can accelerate research, support new products and provide more reliable service to a growing user base. Close cooperation with NVIDIA may also allow its engineers to optimise software for new hardware before that hardware is deployed widely.
Vera Rubin and the Move Beyond Individual GPUs
NVIDIA’s role extends beyond manufacturing processors. The Vera Rubin platform combines GPUs, CPUs, networking, data-processing units and storage technology into rack-scale systems designed to operate together as a large AI supercomputer.
NVIDIA says the platform supports the complete AI cycle, including pretraining, post-training, reasoning and real-time agentic inference. The company announced in March 2026 that the platform’s seven principal chips had entered full production.
This integrated approach is strategically important. Constructing a huge AI cluster from unrelated components can create bottlenecks between computing, memory, networking and storage.
NVIDIA’s platform attempts to optimise the complete system instead of selling an isolated processor.
OpenAI provides another advantage: information about the workloads future frontier models may require. The companies have said they will coordinate their hardware and software roadmaps.
OpenAI can help identify the demands of new models, while NVIDIA can develop infrastructure designed around those demands.
The Ohio Project Shows the Physical Scale
The PORTS-Pike Technology Campus in Pike County, Ohio, demonstrates how the alliance is moving into physical infrastructure.
In August 2026, OpenAI announced that it had entered an agreement to secure approximately eight gigawatts of IT capacity at the site. NVIDIA was identified as the exclusive AI-compute infrastructure provider, working alongside OpenAI, SB Energy and the United States Department of Energy.
OpenAI said it would cover project-specific energy and infrastructure costs, use water responsibly and support opportunities for local workers and businesses.
The company also added $40 million to an existing $40 million community-benefits fund established by SB Energy.
Projects of this size can create construction and technical employment, but they also raise difficult questions. Communities want to know how data centres will affect electricity prices, water resources, land and the environment.
Developers must secure local trust as well as processors and financing.
What NVIDIA Gains
OpenAI’s expansion creates demand across NVIDIA’s entire platform. Large deployments require accelerators, networking equipment, CPUs, software and new generations of rack-scale systems.
NVIDIA’s financial results demonstrate the extraordinary strength of this market. For the second quarter of fiscal 2027, the company reported revenue of $96.2 billion, an increase of 106% from a year earlier. Data Center revenue reached $89 billion, rising 117% year over year.
By investing in AI developers and helping support data-centre construction, NVIDIA is doing more than waiting for customers to order chips. It is helping create the financial and physical conditions in which future orders can be placed.
The model can reinforce itself. Additional infrastructure allows OpenAI to offer more AI services, increased usage creates further computing demand and that demand supports sales of NVIDIA systems.
OpenAI Is Still Diversifying
NVIDIA is foundational to OpenAI’s growth, but the relationship is not completely exclusive.
OpenAI says its broader computing portfolio includes AWS, AMD, Broadcom, Cerebras, CoreWeave, Microsoft, Oracle, SB Energy and SoftBank.
These partners provide different combinations of processors, cloud infrastructure, networking, energy, capital and construction. Diversification can help OpenAI obtain capacity more quickly and reduce the operational risk of relying entirely on one supplier.
It also means NVIDIA must continue proving that each generation of its technology offers sufficient improvements in performance, efficiency and cost.
The Risks Behind the Ambition
The scale of the partnership creates significant risks.
Gigawatt-level campuses require enormous capital and may face delays involving power connections, permits, construction and equipment supply. Hardware can also become outdated quickly, so facilities must begin operating before newer systems change the economics.
There is a larger commercial question: can revenue from AI products justify the extraordinary cost of the infrastructure required to run them?
NVIDIA may participate as a supplier, technology partner and potential investor within the same ecosystem. This alignment can accelerate development, but it also makes the financial relationship more interconnected.
Investors and regulators are likely to examine whether demand is supported by sustainable customer revenue rather than infrastructure spending alone.
OpenAI must demonstrate that additional capacity produces products people and organisations will continue paying to use. NVIDIA must continue delivering efficiency gains that make those services economical.
A Defining Relationship in the AI Economy
NVIDIA and OpenAI represent complementary sides of the artificial-intelligence industry.
OpenAI creates models and applications that generate demand for computing. NVIDIA builds much of the platform that makes large-scale AI possible.
Their partnership shows that the next phase of AI will be shaped not only by algorithms, but also by power generation, construction, financing and industrial coordination.
If the planned infrastructure can be completed and converted into valuable services, the alliance may become one of the defining business relationships of the AI era.
If demand fails to justify the scale of investment, it will provide an equally important lesson about the limits of the industry’s infrastructure boom.
The outcome will influence more than two companies. It will help determine how quickly advanced AI develops, how much it costs to operate and who controls the physical foundation of the technology.
This article reflects publicly available information as of September 3, 2026. It is for informational purposes only and does not constitute financial or investment advice.
