Core Wave imposes upfront payments and short contracts: interest pressure reprices AI computing
Listen to this article
Read by Anchor
The quarterly financial figures of cloud computing company Core Wave reveal a decisive shift in the financing of the AI infrastructure boom. While third-quarter revenue jumped 168 percent to $3.66 billion, quarterly interest costs alone reached $846 million, pushing the company into a net loss of $709 million. This high financial burden prompted a radical review of the expansion model, as CEO Michael Intratour made it clear that the company would not turn to the prevailing unsecured debt market, but instead shift the cost of expansion onto customers’ pockets to finance the unfulfilled backlog of $132.5 billion.
The new plan relies on loans backed by specific customer contracts, and most importantly on large upfront payments. Core Wave expects to raise about $9 billion from customer payments and deposits over the next three months, more than $3 billion of which will be tied to new contracts for pre-built computing capacity, meaning immediate cash liquidity without additional spend on new hardware.This shift is not limited to a single company; it reflects an operational pattern that forces tech giants to pay data-center costs in advance, as competitors Nebius and Airin did by collecting large payments from Microsoft.
The bolder bet lies in moving to short-term contracts to exploit peak demand and raise prices. Core Wave raised its service prices by about 25 percent in July, then added another 10 percent later in the same quarter, while continuing to increase prices at a pace that offsets rising operating costs. The company sells its computing capacity in short-term contracts at over $40 million per megawatt per year, whereas prices in four- or five-year contracts fall to between $20 and $25 million per megawatt. To support this maneuver, the company closed a $2.6 billion credit facility in August specifically designed for borrowing on successive payments secured by these short-term contracts.
But this path carries a highly sensitive trade-off, as lenders always prefer stable, guaranteed cash flows from long-term contracts to repay their loans, even if the rates are lower. While equity markets welcomed revenue growth, bond investors expressed explicit concern; some of the company’s bonds yielded over 15 percent last month.This rise in borrowing costs reveals real fears of the debt accumulation required to build data centers before those investments generate sustainable returns, making the expansion model contingent on continued standard demand and current pricing power without any decline.
This financial shift in computing markets is clearly reflected in the decision makers and infrastructure managers in the Gulf, Egypt, and the Levant. Regional institutions and banks that build their models or lease advanced computing capacity no longer face a market that automatically cuts costs, but find themselves with two options: either commit to long-term contracts spanning years to secure prices near $20 million per megawatt, or pay a doubled price premium exceeding $40 million if they want short-term operational flexibility. For projects and data-center developers in the region, the fact that interest costs reach hundreds of millions offers a practical lesson that building and operating infrastructure requires upfront cash liquidity and avoiding overreliance on debt instruments amid global return volatility.
The cloud computing map for AI is reshaping its rules, as the era of cheap debt-driven expansion gives way to a model that links every new server to client-provided liquidity. I see this scene forcing institutions to scrutinise their computing budgets carefully, as continued technical development has become conditional on each entity’s ability to finance its consumption in advance before starting to process data arrays.