NVIDIA has deepened its partnership with CoreWeave with a $2 billion equity stake—valuing the cloud provider’s shares at $87.20 each—and a joint push to construct over 5 gigawatts of AI factories by 2030.
The collaboration merges CoreWeave’s AI-native software—including its SUNK platform and Mission Control—with NVIDIA’s accelerated computing stack, from Rubin GPUs to Vera CPUs and BlueField storage systems. NVIDIA’s financial backing will help CoreWeave secure land, power, and infrastructure at scale, while the companies aim to integrate CoreWeave’s software into NVIDIA’s broader cloud reference architectures for global enterprises.
Key components of the expanded partnership include
- Deployment of multiple NVIDIA hardware generations across CoreWeave’s platform, starting with Rubin and Vera architectures.
- Validation of CoreWeave’s AI-native software stack for interoperability with NVIDIA’s ecosystem.
- Accelerated procurement of AI factory resources, leveraging NVIDIA’s capital strength.
- Potential inclusion of CoreWeave’s tools in NVIDIA’s cloud partner and enterprise reference designs.
Jensen Huang, NVIDIA’s CEO, framed the move as a response to AI’s infrastructure demands, calling it the largest buildout in history. CoreWeave’s CEO, Michael Intrator, emphasized the need for unified software, hardware, and operational design—a philosophy now reinforced by the partnership.
CoreWeave, which went public in March 2025, operates as a specialized cloud provider for AI workloads, combining custom infrastructure with deep technical expertise. The collaboration positions both companies to dominate the AI factory market, where demand for accelerated computing is outpacing supply.
Next steps include testing CoreWeave’s software against NVIDIA’s architectures and scaling AI factory construction in parallel with customer demand. The partnership also hints at broader industry trends, where AI adoption is driving unprecedented investment in compute infrastructure.
