NVIDIA has announced a major $2 billion equity investment in CoreWeave, to become the second largest shareholder.

CoreWeave, under the expanded agreement, has agreed to deploy and operate more than 5GW of AI compute capacity by 2030, anchored by NVIDIA’s latest platforms, including the Rubin AI platform, Vera CPUs, and BlueField data processing units. Spanning hardware and software, CoreWeave will be among the first to use each new generation of NVIDIA technology.

Critics have labelled the deal as yet another example of circular deal financing and that Coreweave is now heavily indebted to Nvidia. In a previous filing with the SEC, CoreWeave reported that its customers are contractually require to use NVIDIA GPUs, and that they will use Nvidia equitpment to kit out future datacentre buildouts.

Nvidia’s Jensen Huang hit back at the accusations of circular financing and lambasted the criticism “as ridiculous”. In an interview he commented, “These are generational companies — the investments that we make is confidence in them.

“But it’s a small percentage of the amount of money that they ultimately have to go raise, and so the idea that it is circular is — it’s ridiculous.”

Michael Intrator, Co-founder and CEO of CoreWeave commented on Linkedin, “AI has crossed an important threshold. The question is no longer “what’s possible?” It’s “how do we make this work everywhere it needs to?

“This next chapter will play out across industries and in tangible settings, places like hospitals, manufacturing lines, supply chains, and financial systems. These are real systems, where AI has to run dependably, securely, and at scale.

“That’s the shift we’re seeing up close at CoreWeave, and it’s just one of the reasons we’re proud to announce a newly-expanded collaboration with NVIDIA. It’s clear that production — not possibility — will define the year ahead, and we’re eager to roll up our sleeves and get to work.”

Once a crypto-mining operation, CoreWeave has rapidly repositioned itself as a leading AI cloud provider, earning a reputation for quick deployments and early adoption of advanced GPU architectures. Now publicly listed, CoreWeave serves a growing roster of AI model developers, research institutions, and enterprises seeking alternatives to the dominant cloud incumbents.

Already a close NVIDIA partner, CoreWeave was among the first to offer NVIDIA GB200 Grace Blackwell systems at scale, with clients including IBM, Cohere, and Mistral AI. It has also rolled out RTX PRO 6000 Blackwell GPU instances, targeting both cost-conscious and enterprise AI workloads.

While the NVIDIA partnership is grabbing headlines, CoreWeave’s has also clinched deals with Poolside to support vast AI campuses in the US and has been tapped to power Runway’s next-generation AI video models, proving its versatility beyond just large language models. The market has taken notice, with CoreWeave’s shares climbing on news of the deepened NVIDIA relationship.

RELATED STORIES

Nvidia crushes forecasts with $57bn quarter as AI chips sell out

CoreWeave boosts major AI partnerships as revenue surges in Q3