Good morning. For finance leaders trying to understand how the AI infrastructure race is being funded, CoreWeave just offered a revealing case study.
In the span of a few days, the Nvidia-backed startup locked in tens of billions in customer commitments and layered on multiple forms of debt. This underscores a defining feature of today’s AI boom: growth is being financed as aggressively as it is being engineered.
CoreWeave, which provides cloud-based GPUs powered by Nvidia chips, announced on Thursday a $21 billion deal with Meta running from 2027 to 2032. That brings Meta’s total commitment to more than $35 billion, which could be viewed as a significant vote of confidence in sustained demand for AI compute.
At the same time, the company raised roughly $3.5 billion in convertible senior notes, a hybrid instrument that blends debt with an equity upside. Investors collect interest, but can convert into shares if CoreWeave’s valuation rises. It’s a structure that limits near-term cash strain while effectively betting that future equity will be worth more.
And that wasn’t all. CoreWeave CFO Nitin Agrawal said in a LinkedIn post on Friday that the company also:
—Upsized a high-yield bond offering due to heavy demand. —Secured an $8.5 billion delayed draw term loan at investment-grade ratings. —Executed what it described as one of the largest dual-tranche raises of its kind.
Also on Friday, the company’s stock climbed as much as 13% after it announced a multi-year agreement with Anthropic. Taken together, it shows that capital is flowing freely to AI infrastructure—at least for now.
I asked Morningstar equity analyst Luke Yang what this shows for how AI-native companies are approaching capital formation right now. He pointed to three aspects:
—Cloud infrastructure companies will continue to use all kinds of tools available to get the funds necessary for capacity expansion. “For this Meta-CoreWeave deal specifically, we see the company leveraging delayed draw term loans, corporate bonds, and convertible notes. Neoclouds also use OEM financing, operating/financing leases, share issuance, etc. Growth is the top priority, and financing should not be a bottleneck for these companies’ growth.”
—Creditors are getting more comfortable with the business of neoclouds. “We see a sequential improvement in the interest rates and credit ratings of CoreWeave’s delayed draw term loans. The DDTL 4.0 facility is the first investment-grade facility with around a 6% interest rate. Going forward, we expect to see more neocloud companies securing investment-grade financing with take-or-pay deals from credible AI labs/hyperscalers.”
—The interest rate that neoclouds enjoy is critical to the viability of their business model. “Given the high leverage of these companies, even a 100- or 200-basis-point increase in overall interest burden can completely break the business model. The ability to borrow at investment-grade is very important for the preservation of equity holders’ value.”
The company’s recent moves highlight a defining dynamic of the AI era: infrastructure is being built at extraordinary speed, financed by equally aggressive capital strategies. The model appears to be working now because demand is strong, its customers are credible, and capital is available—and Morningstar maintains its $97 fair value estimate for CoreWeave, noting shares look fairly valued following recent gains. But it could all be tested when growth normalizes and the cost of all that financing comes due. Sheryl Estrada [email protected]
The post Tens of billions in days: CoreWeave shows how aggressively AI infrastructure is being funded appeared first on Fortune.
