CoreWeave’s latest earnings laid out a new playbook for financing the AI infrastructure race: get customers to pay earlier, borrow against signed contracts and specialized hardware, and avoid the unsecured debt market while its yields remain punishing.

The company reported third-quarter revenue of $3.66 billion, up 168% year over year, but also a $709 million net loss and $846 million in quarterly interest expense. Against a $132.5 billion backlog, those figures illustrate both the scale of demand and the cost of building data-center capacity before it turns into cash.

On Tuesday, Chief Executive Michael Intrator drew a hard line on how CoreWeave plans to bridge that gap. “We will not use the unsecured debt market to fund our business at current yields, nor do we need to,” he said.¹

Instead, CoreWeave expects roughly $9 billion in customer prepayments and deposits over the next three months. More than $3 billion is tied to new contracts for capacity already built — money that does not require fresh hardware spending. It has also structured borrowing around particular customer contracts, including an August $2.6 billion facility designed for shorter-term deals and drawn down as needed.

The attraction is obvious. Shorter contracts allow CoreWeave to reset prices in a tight compute market: it raised prices about 25% in July and another 10% later in the quarter. Intrator said some short-term capacity is selling for more than $40 million per megawatt annually, compared with $20 million to $25 million for new four- and five-year contracts.

But lenders prefer the steadier cash flows of longer contracts, while prepayments can lock CoreWeave into prices that may look cheap if AI-compute demand keeps accelerating. Bond investors have already signaled unease, with yields on some CoreWeave bonds topping 15% last month. The company’s wager is that demand stays hot long enough for customer cash to finance the buildout — before debt costs catch up.