NVIDIA Secures Land and Power for OpenAI Data Centers for Decades

On August 17, NVIDIA announced a move that takes the chipmaker much deeper into energy and infrastructure. Together with SB Energy, it secured capacity at the PORTS-Pike Technology Campus in Portsmouth, Ohio. NVIDIA uses the abbreviation LPS—land, power, and shell—for the combination of land, electrical capacity, and buildings. OpenAI is expected to be the first tenant and plans to build and operate a large data center there based on the NVIDIA DSX platform. The initial plan calls for 4.25 gigawatts of power. NVIDIA says that one generation of computing systems installed at the site could correspond to roughly 1.5 million of its GPUs and $150 billion to $200 billion in revenue. The location is expected to support several technology refreshes over twenty years. The company is also considering securing the remaining 3.75 gigawatts of available capacity. The scale is extraordinary even in the broader context of NVIDIA's cooperation with OpenAI. NVIDIA says OpenAI's existing and planned commitments represent roughly 12 gigawatts of its compute infrastructure by 2030. If the PORTS-Pike project expanded to include the remaining capacity, the figure could rise to about 16 gigawatts. NVIDIA estimates the value of the opportunity at roughly $600 billion in computing equipment. From a business perspective, the most important change is NVIDIA's role. The chip supplier is no longer simply waiting for customers to build a data center and order servers. It is starting to secure the prerequisites for the investment itself. At PORTS-Pike it is supporting roughly four gigawatts of infrastructure for twenty years. Guarantees are expected to be activated gradually as data centers enter operation between 2028 and 2030. The support applies to a defined portion of rent and energy payments and to residual value, not to all of OpenAI's obligations. The reason is the growing shortage of suitable locations. A modern AI center needs not just a power connection but enormous and long-term available electricity, permits, cooling, physical space, and networking infrastructure. For laboratories developing frontier models, twenty-year energy and lease contracts represent a different kind of financial commitment from simply buying servers. The project also increases the concentration of risk among several companies. OpenAI needs compute capacity, NVIDIA needs long-term demand for its systems, and energy partners need a customer for multi-gigawatt infrastructure. If AI demand grows as expected, this vertical coordination can accelerate construction. If growth slows, the long contracts and guarantees will become a meaningful financial exposure.

NVIDIA Secures Land and Power for OpenAI Data Centers for Decades

The full article is available to users registered at Hard Skills