Nvidia Partners Target 8 GW Installed Capacity by 2026

Nvidia’s network of specialized AI cloud partners is on track to hit roughly 8 gigawatts of installed data center capacity by the end of 2026. The projection came from Nvidia CFO Colette Kress, who also outlined a new financing model designed to shift the company’s revenue mix away from one-time GPU sales and toward recurring income.

Nvidia’s “neocloud” partners are specialized cloud providers built almost entirely around Nvidia GPU infrastructure. CoreWeave and Nebius are the most prominent names in this cohort.

CoreWeave reported 1.5 GW of active power as of Q2 2026, representing a jump of approximately 500 megawatts from the prior quarter. Nebius, another key neocloud partner, is also targeting significant expansion in contracted power capacity by year-end.

OpenAI is involved in a deal for a data center campus in Ohio designed to support up to 8 GW of IT capacity on its own.

The new financing model announced in early July 2026 introduces a revenue-sharing structure that ties Nvidia’s income to the ongoing usage of its products deployed in partner data centers. The company is also providing credit support to help partners finance their buildouts. Nvidia has already guaranteed deals totaling $6.3 billion with CoreWeave alone.

Nvidia has signed memoranda of understanding with Apollo and BlackRock, with the goal of raising over $500 billion in third-party capital for AI infrastructure projects. An agreement signed on August 17, 2026, further cemented Nvidia’s role as a credit backstop and ecosystem orchestrator for these projects.

The risk to watch is concentration. Nvidia’s $6.3 billion in guarantees to CoreWeave alone means the company’s financial health is increasingly tied to the fortunes of a relatively small number of fast-growing but unproven cloud operators. If AI workload demand plateaus or a major neocloud partner stumbles, those guarantees could turn from strategic assets into balance sheet liabilities in a hurry.