Nvidia invests $2 billion in CoreWeave to speed U.S. AI data-center buildout

Nvidia is putting $2 billion into AI infrastructure provider CoreWeave, deepening a relationship that has become emblematic of the new AI supply chain: chipmakers, specialized cloud operators, and the developers that rent computing power to train and deploy models.

The investment, priced at $87.20 per share, comes as CoreWeave expands data-center capacity in the United States. CoreWeave and Nvidia have said the partnership is designed to accelerate the buildout of more than 5 gigawatts of AI data-center capacity by 2030, a scale that underscores both the demand outlook and the growing constraints around energy and site development.

CoreWeave is part of a class of so-called neocloud companies that provide GPU-heavy infrastructure to technology firms building and operating AI systems. These providers have benefited from a wave of enterprise AI adoption and the push by model developers to secure dependable access to high-end accelerators. For Nvidia, the deal strengthens an ecosystem that extends beyond traditional hyperscalers.

The announcement also reflects the practical bottlenecks in the AI boom. It is no longer just about buying chips; it is about locating sites, obtaining permits, and, most critically, securing electricity and grid connections. By backing a major customer and partner, Nvidia is signaling that capacity expansion—and the infrastructure to support it—is a central competitive battleground.

CoreWeave’s shares rose in premarket trading after the news. Investors have been watching how rapidly the company can expand while managing capital needs, given the large upfront costs required to build data centers at this scale. For the broader market, the deal is another marker of how quickly AI investment is moving from software into the physical economy.