In a move that further tightens the bond between the world’s most valuable chipmaker and the infrastructure powering the artificial intelligence (AI) boom, NVIDIA Corp. has invested $2 billion in CoreWeave, establishing NVIDIA as the second-largest shareholder in the neocloud provider.
The announcement, made early Monday, sent CoreWeave’s shares climbing as much as 12% in active trading. Under the terms of the agreement, NVIDIA purchased Class A common stock at $87.20 per share, a modest discount from Friday’s closing price of $92.98.
The capital infusion is earmarked for a massive infrastructure expansion, underscoring the frantic pace of the global AI build-out. CoreWeave aims to operate more than 5 gigawatts of AI-dedicated data centers by 2030. To put that scale into perspective, 5 gigawatts is roughly equivalent to the annual power consumption of 4 million U.S. households.
“This deal allows us to accelerate our build,” CoreWeave CEO Mike Intrator said in an interview with CNBC. He noted the expansion would help the company diversify its client base and reduce dependency on individual large-scale contracts as it scales into new capacity.
NVIDIA CEO Jensen Huang, appearing with Intrator, emphasized that while $2 billion is a significant sum, it represents only a fraction of the total capital required to reach the 5-gigawatt goal. “The demand is just extraordinary,” Huang said. “We’re in the beginning of the AI infrastructure build-out.”
The partnership goes beyond financing. The two companies said they will align hardware and software strategies, specifically testing CoreWeave’s Mission Control platform. The resource-scheduling software is being evaluated for potential integration into NVIDIA broader ecosystem, which could streamline how developers deploy massive AI workloads across distributed networks.
“We are watching the AI infrastructure stack consolidate around capital, compute, and control,” said Mitch Ashley, vice president and practice lead, Software Lifecycle Engineering, at The Futurum Group. “NVIDIA’s expanded investment in CoreWeave reinforces that access to high-end accelerators is a strategic alignment decision. When the model provider, the chip vendor, and the infrastructure operator tighten their relationship, they shape who can scale AI workloads and how fast.”
“These kinds of partnership pressures reshape the market. Enterprises and AI-native companies will have to decide whether they want flexibility across clouds or preferential access to tightly integrated stacks optimized for performance and availability,” Ashley added. “As AI moves into production at scale, infrastructure choices increasingly define competitive outcomes, not just cost profiles.”
CoreWeave has undergone a dramatic evolution since its founding in 2017 as Atlantic Crypto, a bitcoin mining firm. After the 2018 crypto crash, the company pivoted to GPU-accelerated cloud computing, eventually becoming a primary partner for NVIDIA. CoreWeave’s business model revolves around building AI factories packed with NVIDIA’s high-end graphics processing units (GPUs), which are then rented out to tech giants and startups.
The investment follows a series of blockbuster deals for CoreWeave, including a $14.2 billion infrastructure agreement with Meta Platforms Inc. and a $22.4 billion contract expansion with OpenAI. However, the company’s stock has faced volatility recently as investors weigh its rapid growth against the high levels of debt used to finance multibillion-dollar projects.
Despite concerns, NVIDIA’s doubled-down commitment suggests a long-term bet on CoreWeave’s ability to execute. This latest deal adds to a previous $6.3 billion agreement where NVIDIA committed to purchasing residual unsold capacity from CoreWeave through 2032, ensuring the provider has a guaranteed safety net as it builds at a pace Huang described as “historically unprecedented.”
