CoreWeave outreach angle

NVIDIA announced a strategic expansion with CoreWeave on January 26, 2026 that includes a $2.0 billion purchase of CoreWeave Class A common stock and a joint plan to accelerate the buildout of more than 5 gigawatts of AI factories by 2030. (nvidianews.nvidia.com) CoreWeave’s infrastructure agreements with OpenAI have been public and large-scale: the company’s original contract was disclosed as being “up to $11.9 billion,” and later reporting described an additional $6.5 billion expansion that brought some totals reported in press coverage to roughly $22.4 billion. (prnewswire.com) CoreWeave’s contracted footprint includes a 260 MW Denton expansion with Core Scientific that increased its total contracted HPC capacity with Core Scientific to about 590 MW across six sites, and the company has reported operating more than 250,000 GPUs across roughly 32 data centers and ~360 MW of active power. (businesswire.com) Operational and construction issues have punctuated that growth: CoreWeave’s CEO publicly addressed third‑party data‑center delays that coincided with a stock drop of around 16% on November 11, 2025, and third‑party trackers show hundreds of user‑reported incidents since 2022 alongside multiple scheduled maintenances in recent weeks. (cnbc.com) NVIDIA’s on‑prem stack includes DGX H100 systems built from eight H100 GPUs delivering up to ~32 petaFLOPS for training workloads, and a networking portfolio that spans Quantum InfiniBand and Spectrum‑X Ethernet intended for large AI fabrics. (infohub.delltechnologies.com) Galaxy’s Helios agreement to deliver roughly 260 MW of incremental critical IT load to CoreWeave, combined with CoreWeave’s Denton expansion and OpenAI commitments, concentrates hundreds of megawatts of GPU capacity with a small set of providers—an observable commercial fact that sales teams cite when framing owned‑infrastructure and multi‑supplier continuity solutions. (galaxy.com)