Nvidia invests 2 billion dollars in CoreWeave to scale 5 GW of AI data centers by 2030

Nvidia is leaning further into the infrastructure side of AI by investing 2 billion dollars in CoreWeave, a GPU-focused cloud provider that’s racing to add massive data center capacity over the next five years.

Nvidia and CoreWeave deepen “AI factory” buildout

Nvidia said it purchased CoreWeave Class A shares at 87.20 dollars per share as part of a deal aimed at accelerating CoreWeave’s plan to add more than 5 gigawatts of compute capacity by 2030. The two companies also plan to co-develop “AI factories,” effectively data centers optimized around Nvidia hardware and reference designs.

For marketers and e-commerce founders, the practical signal is supply: more compute capacity can translate into better availability (and potentially more predictable pricing) for training and inference workloads that sit behind creative generation, personalization, and customer support automation.

CoreWeave also plans to integrate Nvidia’s latest roadmap across its platform, including the Rubin architecture (positioned as the successor to Blackwell), BlueField systems, and Nvidia’s newer CPU line, Vera. The message is clear: CoreWeave is aligning product and procurement tightly with Nvidia’s stack, which can simplify performance tuning for customers but also increases dependence on a single vendor ecosystem.

Debt scrutiny, acquisitions, and hyperscaler demand

The investment lands as CoreWeave faces increased attention on its financing model. PitchBook data shows CoreWeave had 18.81 billion dollars in debt obligations as of September 2025 (PitchBook). CoreWeave has argued that using GPUs as collateral is a rational response to rapid shifts in compute supply and demand.

Operationally, the company has been broadening its developer-facing stack through acquisitions, including Weights and Biases (experiment tracking and MLOps), and it counts major customers such as OpenAI, Meta, and Microsoft.

Nvidia will also support CoreWeave with site selection needs like land and power procurement, and with packaging CoreWeave’s approach into Nvidia’s reference architectures for enterprise and cloud buyers.

The near-term takeaway: Nvidia is using capital and platform control to keep compute expansion moving, reinforcing a tight feedback loop between chip supply, cloud capacity, and enterprise adoption.