AivexaNewsSearch
AI news for builders and product teamsChecked every hour

CoreWeave turns to customer prepayments, shorter contracts to fund AI growth

Collected Nov 11, 2026

CoreWeave disclosed in its third-quarter earnings, reported Tuesday, that it is changing how it finances the buildout of AI computing capacity. The neocloud, which rents out GPU capacity behind AI workloads, said it is increasingly asking customers to pay in advance, borrowing against individual customer contracts, and pursuing shorter-term deals that can carry higher prices. CEO Michael Intrator told investors on Tuesday's earnings call that the company will not use the unsecured debt market at current yields and does not need to.

The quarter showed both the scale of the business and its cost. CoreWeave booked $3.66 billion in revenue, up 168% from a year earlier, but posted a $709 million net loss and $846 million in quarterly interest expense. Its backlog of contracted work stands at $132.5 billion, a figure that represents infrastructure the company still has to build and deliver.

The financing plan hinges on contract-backed lending. Rather than issuing general corporate debt, CoreWeave intends to borrow against specific customer agreements and draw funds as needed to finance construction. The company also expects roughly $9 billion in customer prepayments and deposits over the next three months. More than $3 billion of that is tied to signing new customer contracts for capacity already built, meaning it requires no additional hardware spending. In practice, that is close to free cash: the GPUs are installed and the customer pays before consuming the service.

Prepayment is becoming a pattern across the sector. Rivals Nebius and IREN have likewise leaned on large Microsoft prepayments to fund computing capacity, a sign that AI cloud providers are increasingly asking large customers to pay ahead for buildouts. For buyers, an upfront commitment locks in capacity and, presumably, a price; for the provider, it converts future revenue into present capital.

CoreWeave is also shifting its contract mix toward shorter terms. It raised prices across its offerings by about 25% in July and by another 10% later in the quarter, and prices are still climbing this quarter in a way Intrator said more than offsets cost increases. The company is allocating more capacity to short-term opportunities where pricing justifies it, selling capacity at more than $40 million per megawatt annually. New infrastructure on four- and five-year contracts, by contrast, is moving toward $20 million to $25 million per megawatt.

That gap explains the strategy. A short-term deal at roughly double the annual per-megawatt rate lets CoreWeave capture spot-market scarcity, but it also means the revenue has to be re-won sooner and at whatever price prevails when the term ends. Upfront payments reduce immediate borrowing needs, yet they commit the company to delivering service at a price that may look cheap if compute demand keeps surging.

The lending side is built around that trade-off. In August, CoreWeave closed a $2.6 billion loan, its first designed specifically for shorter-term customer contracts. Known as DDTL 5.5, it allows the company to borrow in installments as needed rather than in a single draw. This suggests a deliberate attempt to match the duration of debt to the duration of the underlying contracts, so the company is not paying long-term fixed rates against revenue that reprices every few months.

Lenders, however, weigh predictable cash flow heavily. Longer contracts, even at lower prices, give a lender more confidence in repayment, so a book tilted toward short deals may support less leverage or demand better terms. That is a likely trade-off: CoreWeave gains pricing power and flexibility but may find its borrowing capacity and cost of capital tied more tightly to renewal risk.

Public markets are already pricing some of that risk. Wall Street reacted positively to the earnings, but bond investors have been less sanguine. Last month, yields on some CoreWeave bonds rose above 15%, a level that signals concern about the debt required to fund data center buildouts before those investments generate returns. For context, high-yield corporate debt typically trades far below that; a 15% yield implies investors see meaningful default or refinancing risk, or simply demand a large premium to hold the paper.

TechCrunch said it reached out to CoreWeave for more information about its loan facility.

Why it matters: For developers and product teams buying GPU capacity, the shift toward prepayment and short-term contracts could mean more upfront commitment and more volatile pricing on flexible capacity, while long-term reserved deals may stay comparatively cheaper. Inference and training teams that plan capacity years out should expect providers to push longer commitments and deposits, and should model renewal risk if they rely on spot-like pricing. The broader read, and this is inference rather than anything the company stated, is that the AI infrastructure buildout now depends on sustained demand and pricing power holding up; if either softens, financing structures built on contract-backed loans and prepayments come under pressure first.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

CoreWeave is increasingly asking customers to pay up front, borrowing against specific contracts, and even looking to short-term deals that could lead to higher prices, according to the company’s third-quarter earnings.