CoreWeave is overhauling its financing approach by requiring customers to pay upfront and pursuing shorter contracts that allow premium pricing, according to the company's third-quarter earnings released Tuesday. The neocloud provider, which leases computing infrastructure for artificial intelligence workloads, says it's shifting away from traditional debt markets to fund expansion. The strategy aims to extract more value from existing capacity while reducing reliance on expensive unsecured borrowing, though success hinges on continued strong demand and pricing leverage.
The company posted revenue of $3.66 billion for the quarter, climbing 168% year-over-year, but reported a net loss of $709 million alongside $846 million in quarterly interest costs. CoreWeave expects to collect roughly $9 billion in advance payments and deposits from clients over the coming three months, including over $3 billion connected to new contract signings for infrastructure already constructed, which demands no further hardware investment. The company holds a backlog worth $132.5 billion. In August, CoreWeave secured a $2.6 billion loan facility—its first structured specifically for shorter-duration customer agreements—that permits borrowing in increments as capital needs arise. The firm raised prices approximately 25% in July and another 10% later in the quarter, with increases continuing into the current period.
"We will not use the unsecured debt market to fund our business at current yields, nor do we need to," CEO Michael Intrator said during Tuesday's earnings call. Instead, the company plans to lean on loans collateralized by individual customer contracts, tapping funds as required to build most of the infrastructure needed to fulfill its substantial order pipeline. Intrator explained the company is "allocating more capacity to short-term opportunities, where the pricing justifies it, selling capacity at more than $40 million per megawatt annually, while pricing for new infrastructure on four- and five-year contracts is moving to $20 million to $25 million per megawatt." The report notes that competitors Nebius and IREN have similarly relied on significant Microsoft advance payments to finance computing buildouts, suggesting the sector is trending toward requiring major technology firms to fund expansions upfront.
Advance payments could reduce CoreWeave's immediate borrowing requirements, but they also lock the company into delivering services at rates that may become less favorable if computing demand continues surging and prices rise further. While shorter-term agreements enable CoreWeave to charge elevated rates and adjust pricing more frequently, lenders assign greater value to longer-term commitments despite lower pricing, prioritizing predictable revenue streams and stability for loan repayment above all else. The report notes that although Wall Street responded favorably to the earnings, bond investors remain considerably more cautious—yields on certain CoreWeave bonds reached above 15% last month, reflecting investor concern about the debt required to finance data center construction before those assets generate returns. The strategy CoreWeave has adopted could enhance profitability but also introduces heightened risk surrounding contract renewals and pricing dynamics, illustrating how heavily the AI infrastructure expansion depends on sustained demand that remains unproven.
The company's pivot toward customer-funded growth and premium short-term contracts represents a bet that today's pricing power will persist even as it builds out capacity against a massive backlog. If compute demand weakens or pricing softens before CoreWeave delivers on prepaid commitments, the company could face margin pressure while servicing substantial debt loads. The approach signals a broader industry shift where cloud providers are asking hyperscale customers to shoulder more upfront capital risk rather than relying solely on traditional financing. This financing model reshapes the implicit bargain between infrastructure providers and their largest clients, transferring construction risk in exchange for locking in future capacity. Whether this proves sustainable depends less on CoreWeave's execution than on forces largely outside its control—the pace at which enterprises adopt AI and the duration of elevated compute pricing.