CoreWeave Shifts to Prepayments and Spot Pricing as Infrastructure Costs Climb
The GPU cloud provider is asking customers to pay upfront and pursuing shorter contracts at premium rates, betting that AI demand will justify riskier financing structures.
A New Financing Playbook
CoreWeave unveiled a financing strategy built on customer prepayments and contract-specific borrowing in its third-quarter results, released Tuesday. The GPU cloud provider reported revenue of $3.66 billion, a 168 per cent year-on-year increase, but also disclosed a $709 million net loss and $846 million in quarterly interest expense. With those figures in focus, chief executive Michael Intrator made clear the company would not tap unsecured debt markets at prevailing yields.
Instead, CoreWeave plans to draw on loans backed by individual customer contracts, pulling funds incrementally as infrastructure projects progress. The approach is designed to match borrowing more closely to committed revenue, reducing exposure to broader credit markets. The company holds a backlog of $132.5 billion in signed contracts, much of it still requiring capital deployment to deliver the compute capacity customers have reserved.
Prepayments Take Centre Stage
CoreWeave expects roughly $9 billion in customer prepayments and deposits over the next three months. More than $3 billion of that total is tied to new contracts for capacity the company has already constructed, meaning no additional hardware spending is required to recognise that revenue.
The shift toward upfront payments mirrors moves by competitors Nebius and IREN, both of which have secured large prepayments from Microsoft to fund their own infrastructure expansions. The pattern suggests that hyperscale customers are willing to commit capital early in exchange for guaranteed access to scarce GPU capacity, effectively acting as financiers for the buildouts that will serve their workloads.
For CoreWeave, prepayments ease immediate liquidity pressure and reduce reliance on external lenders. Yet they also lock in pricing at today's rates. If compute demand continues to accelerate and spot prices rise further, the company will be obliged to deliver services at terms agreed months or years earlier, potentially leaving margin on the table.
Shorter Contracts, Higher Prices
CoreWeave has also begun allocating more capacity to shorter-term agreements, where pricing power is greatest. The company raised prices by approximately 25 per cent in July, followed by another 10 per cent increase later in the quarter. Pricing continued to climb in the fourth quarter at a pace that, according to Intrator, more than offsets rising infrastructure and energy costs.
Short-term contracts now command rates exceeding $40 million per megawatt annually, nearly double the $20 million to $25 million per megawatt CoreWeave charges for four- and five-year commitments. The premium reflects both tight supply and customers' willingness to pay for immediate availability, particularly as training runs and inference deployments scale faster than anticipated.
In August, CoreWeave closed a $2.6 billion loan facility, designated DDTL 5.5, structured specifically to finance shorter-term customer agreements. The facility allows the company to borrow in tranches as needed, aligning debt drawdowns with contract milestones. This marks a departure from CoreWeave's earlier reliance on loans backed by longer-duration contracts, which lenders traditionally prefer because they offer more predictable cash flow.
The Trade-Off in Tenor
Shorter contracts deliver higher unit economics but introduce renewal risk. Customers on three- or six-month terms can renegotiate or walk away more frequently, leaving capacity idle if demand softens or if competitors undercut on price. Lenders, meanwhile, assign lower borrowing capacity to short-term revenue streams, even when those streams carry higher margins, because repayment depends on a series of renewals rather than a single, locked-in commitment.
CoreWeave's decision to pursue this mix reflects confidence that AI workloads will remain capacity-constrained for the foreseeable future. The company is effectively betting that customers will continue to compete for GPU access, sustaining both utilisation and pricing. If that assumption holds, the strategy could lift profitability without requiring additional equity dilution or unsecured borrowing.
Bond Markets Signal Caution
Equity investors responded positively to CoreWeave's third-quarter results, but the bond market has been less enthusiastic. Yields on some of the company's outstanding bonds climbed above 15 per cent in October, a level that reflects concern over the scale of capital required to complete data centre buildouts before those facilities generate cash.
The yield spread suggests that fixed-income investors see meaningful risk in CoreWeave's debt load relative to its current cash generation. The company's interest expense of $846 million in a single quarter underscores the cost of carrying that debt while infrastructure projects are still under construction. CoreWeave's financing model depends on converting backlog into live, revenue-generating capacity quickly enough to service obligations and fund the next wave of expansion.
Industry-Wide Pressure
CoreWeave's financing pivot is part of a broader recalibration across the AI infrastructure sector. The combination of elevated interest rates, long lead times for hardware procurement, and intense competition for power and data centre space has made traditional project finance more expensive and harder to secure. Companies that can secure customer prepayments or contract-backed lending gain a structural advantage, but only if they can deliver capacity on schedule and maintain pricing discipline.
The reliance on hyperscale prepayments also concentrates risk. A handful of large customers now effectively underwrite much of the industry's expansion. If any of those customers slow spending, delay projects, or renegotiate terms, the ripple effects could force providers to adjust their own capital plans or seek alternative financing at less favourable rates.
What Comes Next
CoreWeave's strategy rests on sustained AI demand and the assumption that compute scarcity will persist long enough to justify premium pricing on shorter contracts. If those conditions hold, the company's financing model could deliver higher returns without exposing it to the volatility of unsecured debt markets. If demand falters or new capacity floods the market, the risks inherent in shorter tenors and contract-specific borrowing will become more pronounced.
The company has not disclosed the full terms of its DDTL 5.5 facility or the covenants attached to its contract-backed loans. Those details will matter as CoreWeave works through its backlog and navigates the trade-offs between locking in long-term revenue and capturing spot market premiums. For now, the bet is that customers will continue paying up for access, and that prepayments can bridge the gap between construction and cash flow.



