Nvidia CEO Jensen Huang just told investors that the future of AI computing isn’t about chips anymore. It’s about land, power, and buildings.
Speaking at Thursday’s Goldman Sachs Communacopia + Technology Conference, Huang said Nvidia remains confident it can deliver roughly 70% year-over-year revenue growth next year. The catch: unconstrained demand is growing at more than 100%, and the bottleneck is shifting from chips to data center capacity. “The power of the neoclouds is this,” Huang said. “They secure land, power, and shell for us that the CSPs have already exhausted.”
He specifically named two companies to prove the point: CoreWeave (CRWV) and Nebius Group (NBIS).
The Capacity Shift
The chip shortage was the story a few years ago. Now hyperscalers are pouring hundreds of billions of dollars into AI, and the limits are physical: land, electricity, and finished buildings. The largest cloud companies have exhausted their ability to secure those resources, Huang argued, which creates an opening for smaller players who specialize in securing them.
The Neocloud Role
Nvidia is betting that these companies become another distribution channel for its AI systems, giving it access to data center capacity that would otherwise be difficult to secure. The neoclouds, as Huang calls them, secure the land, power, and data center shells that the hyperscalers have exhausted.
CoreWeave’s Numbers
CoreWeave is the larger and earlier-scaled of the two publicly traded neoclouds. It built its AI-cloud business years before Nebius emerged from the Yandex restructuring. Its growth has been extraordinary.
According to its second-quarter earnings release, revenue reached $2.58 billion, up from $1.21 billion a year earlier. Its revenue backlog reached $104 billion.
Nvidia’s Investment
Nvidia invested $2 billion in CoreWeave in January, purchasing shares at $87.20 apiece. The companies also expanded their relationship to support more than 5 gigawatts of AI factories by 2030, including Nvidia CPUs, storage platforms, and multiple generations of Nvidia systems.
| Company | Revenue Q2 | Revenue Growth Y/Y | Backlog |
|---|---|---|---|
| CoreWeave | $2.58B | Up from $1.21B | $104B |
What Huang Said
Huang’s comments at the conference were particularly bullish for CoreWeave and Nebius. He said both are “doing fantastically.” That is more than a compliment. It explains why Nvidia is willing to put billions of dollars behind these companies.
The Hyperscaler Constraint
Huang’s central argument is that the hyperscalers—the large cloud companies—have already exhausted their ability to secure land, power, and data center shells. The neoclouds, by contrast, specialize in exactly that work. They secure the capacity that the hyperscalers cannot.
The Road Ahead
Huang’s forecast is optimistic. Nvidia expects roughly 70% year-over-year revenue growth next year, even as unconstrained demand grows at more than 100%. The question is whether the neoclouds can deliver the capacity Huang says they will.
The Strategy
Huang’s framing is clever. Instead of competing with the hyperscalers, Nvidia is partnering with companies that do what the hyperscalers can no longer do. The neoclouds are securing the land, power, and buildings that the hyperscalers have exhausted.
It is a vote of confidence, but it is also a statement of strategy. Nvidia is betting that these smaller companies can secure capacity where the hyperscalers cannot. Whether that bet pays off depends on whether CoreWeave and Nebius can actually deliver the capacity Huang says they will.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.