The latest signals out of the AI economy all point the same direction: the buildout has stopped being a pure spending race and become a financing test. Amazon, Alphabet and Microsoft each used earnings to disclose how they intend to pay for artificial intelligence over the long term, while CoreWeave is restructuring how it gets paid and OpenAI is reportedly scaling revenue fast. The common thread is that capital structure, not ambition, now separates the winners.
Three Hyperscalers, Three Balance Sheets
The most useful read on the quarter comes from SiliconANGLE's analysis of the three hyperscalers reporting within roughly 24 hours of one another. Its conclusion is that this earnings season revealed more about long-term AI strategy than any keynote or press release this year, and that beneath the headline numbers sit three very different approaches. That framing matters for the US market because these three companies are the anchor tenants of domestic AI capacity: they buy the chips, lease the power and set the terms that smaller players must live with. When their strategies diverge, the divergence shows up in everything downstream, from data-center construction schedules to the contract terms offered to AI startups. The takeaway from SiliconANGLE's piece is not which company is spending most, but that each has chosen a distinct way to fund and structure the spending.
CoreWeave Moves the Risk Upstream
CoreWeave's third-quarter earnings, as reported by TechCrunch, show the clearest version of the pattern. The company is increasingly asking customers to pay upfront, borrowing against specific contracts, and even looking to short-term deals that could lead to higher prices. Each of those moves shifts risk away from the balance sheet and toward the customer or the lender. Prepayment is effectively a customer-funded expansion; contract-specific borrowing narrows what lenders are exposed to; shorter deals preserve flexibility but, per TechCrunch, could come with higher prices. For US technology companies that rent capacity rather than build it, this changes the cost of planning. Budgeting for compute becomes less like signing a utility contract and more like managing a supply chain with prepayment terms and renegotiation risk. Price stability is the casualty.
OpenAI's Revenue Line and the Prepayment Logic
The case for prepayment gets stronger with OpenAI's numbers. TechCrunch reported that OpenAI's quarterly revenue crossed $10B for the first time, an almost 50% jump from the $6.7 billion recorded in the second quarter. A customer of that size and growth rate is exactly the counterparty a capacity provider wants to lock in early, which is why upfront payment arrangements are not simply a sign of vendor distress. They are a rational response to demand that is large and rising quickly. The same figures cut the other way too: if revenue keeps compounding at anything like that pace, the pressure to secure capacity ahead of demand grows, and the financing terms attached to that capacity become a competitive weapon rather than an administrative detail.
