Nvidia has invested $2 billion in CoreWeave, bringing its total stake in the AI cloud provider to 11% and making it the second-largest shareholder as competition from Google’s tensor processing units and AMD’s graphics chips intensifies.
The investment, announced Monday, extends a partnership that already includes $6.3 billion in committed cloud computing purchases and positions CoreWeave to build 5 gigawatts of AI data centerwhat both companies call "AI factories" by 2030. That infrastructure is critical as OpenAI, Anthropic, and other AI labs race to secure computing capacity for increasingly resource-intensive models.
The deal brings Nvidia’s total CoreWeave holdings to more than $5.3 billion and comes as the AI chipmaker faces mounting pressure from Big Tech companies developing custom silicon to reduce dependence on its processors.
Ogbonda Chivumnovu
Nvidia purchased CoreWeave Class A common stock at $87.20 per share. The stock jumped 12% in premarket trading Monday before closing up nearly 6%. The investment makes Nvidia CoreWeave’s second-largest shareholder behind Magnetar Capital and values the company at $52 billion.
“AI is entering its next frontier and driving the largest infrastructure buildout in human history,” Nvidia CEO Jensen Huang said in a statement. “Together, we’re racing to meet extraordinary demand for NVIDIA AI factories the foundation of the AI industrial revolution.”
CoreWeave CEO Michael Intrator told CNBC the funding would “accelerate our build, which will lead to continued diversification and reducing dependency on any particular client as we scale into this additional data center capacity.”
Why Nvidia is investing billions in AI infrastructure providers
The timing reflects mounting competitive pressure. Nvidia faces challenges from Big Tech companies building their own AI processors to reduce dependence on its chips. Google’s tensor processing units are being adopted by Anthropic and other AI labs. OpenAI is working with chip designer Broadcom to develop custom AI accelerators while simultaneously buying graphics processing units from AMD.
By investing heavily in “neoclouds” like CoreWeave specialized providers that rent out Nvidia-powered infrastructure Huang is bypassing traditional hyperscalers like Amazon Web Services and Microsoft Azure, which are increasingly developing internal silicon. The strategy gives Nvidia priority access to customers and locks in demand for its next-generation chips.
This isn’t Nvidia’s first bet on CoreWeave. The company already committed to buying at least $6.3 billion worth of CoreWeave’s cloud computing capacity through April 2032 in a deal disclosed last September. Before Monday’s announcement, Nvidia held a 6.3% stake in CoreWeave, valued at $3.3 billion.
CoreWeave’s aggressive expansion raises investor concerns
CoreWeave has positioned itself as critical AI infrastructure for major tech companies.
CoreWeave has come a long way since its days as a crypto miner. After the 2018 market crash forced a pivot, they switched to renting out Nvidia GPUs—a move that eventually led to a Nasdaq debut in March 2025. They’ve locked in some massive wins since then, including deals with OpenAI and Meta worth over $36 billion combined.
But it’s a high-stakes game: despite a $52 billion valuation, the company is still unprofitable because they’re pouring every spare dollar into building new data centers. The company’s projected capital expenditures could exceed $30 billion in 2026 spending between $2.35 and $2.77 for every dollar of revenue earned. Its stock has been volatile, doubling since its IPO but losing roughly half its value since peaking in June.
Is Nvidia’s investment strategy circular financing?
Nvidia’s approach has drawn scrutiny from analysts who worry about circular financing patterns. The company has poured billions into AI startups and infrastructure providers including up to $100 billion committed to OpenAI over several years and a $200 million stake in AI video startup Synthesia announced the same day as the CoreWeave deal many of which then use those funds to purchase Nvidia chips.
A CoreWeave spokesperson told Reuters the $2 billion “will not be used to purchase Nvidia processors” but instead will go toward land acquisition, power procurement, research and development, and workforce expansion. Still, the optics of Nvidia funding the very companies buying its products have fueled concerns about inflated AI valuations and potential bubble dynamics.
Huang acknowledged the scale of capital still needed. “We’ve invested $2 billion into CoreWeave, but recognize that the amount of funding that needs to be raised yet to support that five gigawatts is really quite significant,” he told CNBC. “We’re investing a small percentage of the amount that ultimately has to go and be provided.”
As part of the expanded partnership, CoreWeave will be among the first to deploy Nvidia’s upcoming Rubin computing platform and Vera central processing units. The Vera CPUs mark Nvidia’s entry into direct competition with Intel and AMD in the server processor market, a space traditionally dominated by those companies.
“We’re in the beginning of the AI infrastructure build-out, and the demand is just extraordinary,” Huang said.
Whether that demand justifies the spending and whether CoreWeave can execute on its ambitious buildout plans remains the open question for investors trying to separate genuine AI infrastructure needs from speculative excess.
Damilare Odedina
