Oracle ended fiscal 2026 with roughly 141,000 full-time employees, down from 162,000 a year earlier, a reduction of more than 21,000 jobs and close to 13 percent of its workforce. The company tied the cuts directly to its pivot toward artificial intelligence, stating in its filing that "the adoption and deployment of AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce." The reductions cost roughly $1.8 billion in restructuring charges. They also landed in the same fiscal year that Oracle poured record capital into the infrastructure those AI technologies run on.
The two facts belong to the same story. Oracle spent $55.7 billion on capital expenditures in fiscal 2026, up from $21.2 billion the year before, a 162 percent increase driven almost entirely by its AI cloud and data center buildout. Free cash flow swung to negative $23.7 billion. To fund the gap, the company raised roughly $43 billion in debt and $5 billion in equity during the year. The headcount cuts and the debt-financed expansion are two outputs of the same calculation: how to underwrite a capital program of this size without the cash flow to match it.
Oracle's remaining performance obligations, the contracted backlog it expects to convert to revenue, reached $638 billion, up from $138 billion a year earlier. The largest single driver is a five-year, $300 billion capacity agreement with OpenAI. Backlog of that scale is the justification for the capex, but it does not arrive as cash today. The data centers have to be built, powered, and cooled before the contracted revenue lands, and Oracle's own guidance points to roughly $70 billion in capex for fiscal 2027, plus another $20 billion to $25 billion it expects customers to reimburse. That is a company spending against future revenue, financing the difference, and trimming payroll to protect margin in the interval.
The financing strain is visible in the credit markets. Oracle's five-year credit default swaps widened to levels last seen during the global financial crisis, a signal that bond investors are repricing the risk of a balance sheet carrying tens of billions in new debt against data center assets that will not generate full returns for years. The private credit financing structures backing Oracle's data center expansion have absorbed much of this, but the cost of capital is rising even for the operators with the strongest contracted demand.
Oracle's expansion is one of the largest single sources of new liquid-cooling and thermal demand in the market. Capacity built for OpenAI-class training and inference workloads runs at rack densities that air cooling cannot serve, which means direct-to-chip and, in some configurations, immersion become procurement line items rather than future considerations. The buildout that the layoffs are meant to finance is the same buildout that fills cooling vendors' order books. That is the demand side, and it remains strong.
The cost discipline is the other side. A company cutting 21,000 jobs and running negative $23.7 billion in free cash flow will push every line of its construction budget for efficiency and price. Cooling sits inside that budget. Vendors should expect harder negotiation on unit economics, more pressure to demonstrate total-cost-of-ownership savings over a multi-year horizon, and closer scrutiny of energy performance because power and water are operating costs that compound across a fleet. The Oracle layoffs are an early, visible instance of a pattern that the broader capex surge across hyperscale operators will reinforce: spending is enormous, and the financing behind it is tightening.
The risk for cooling suppliers is that the financing strain does not stay contained at the hyperscale tier. When the operators with the deepest balance sheets are raising debt at crisis-era spreads and cutting headcount to defend margin, the smaller buyers further down the chain face worse terms still. That dynamic is already showing up in stalling neocloud and colocation deals where credit risk has started to gate construction. Cooling vendors planning capacity against this demand should underwrite their own pipelines for the possibility that some contracted projects slip as financing costs climb. The orders are real. The schedules behind them depend on capital that is no longer cheap.