The numbers that landed after Thursday’s close told two stories at once. Oracle reported first-quarter cloud infrastructure revenue that more than doubled from a year earlier, to $7.4 billion, while its capital spending and borrowing both climbed to levels that put the cost of the AI buildout in plain view.
Total revenue rose 30 percent to $19.35 billion, above the $19.14 billion analysts expected. Cloud revenue overall grew 62 percent to $11.6 billion, and non-GAAP earnings came in at $1.92 a share against a $1.74 consensus. The stock rose 6.77 percent to close at $163.29, after touching a gain of more than 7 percent in the immediate reaction.
The cloud infrastructure number is the one that matters. At $7.4 billion, up 121 percent from a year ago, it beat the $7.1 billion consensus and confirmed that Oracle’s decision to position itself as a supplier of AI compute is working.
The backlog is the forward-looking proof. Remaining performance obligations rose by $209 billion in a year to $664 billion, a figure driven by the multi-year AI contracts the company has been signing faster than it can build the data centers to serve them.
CEO Safra Catz said the quarter included more than $30 billion in new AI contracts, signed without drawing on additional Oracle capital. The line is aimed at investors worried about how the company finances its expansion: the customers, Catz argued, are paying for the buildout in advance.
That is the tension the quarter exposed. Oracle delivered 850 megawatts of data-center capacity and more than 300,000 GPUs in the quarter, nearly triple the prior quarter’s GPU deliveries. To do it, capital expenditure jumped to $28.5 billion from $8.5 billion a year earlier, and debt on the books reached $125 billion.
The financing came from several places at once. Oracle completed a $20 billion at-the-market equity offering during the quarter, and free cash flow was negative $5.4 billion. Customers are prepaying for capacity, and the company is borrowing and issuing equity to build it.
Analysts said the quarter is the clearest illustration yet of the capital intensity of the AI infrastructure business. The revenue is growing fast, but the bill for the growth is arriving faster, and the balance sheet now carries the kind of debt that only makes sense if the demand keeps coming.
The company’s answer is that the demand is contractual. The $664 billion backlog is not a forecast; it is obligations customers have already signed. The risk is not whether the orders exist but whether Oracle can build, power, and staff the facilities fast enough to recognize the revenue.
The GPU deliveries are the operational proof point. Shipping 300,000 GPUs in a quarter means the supply chain, the data centers, and the customer deployments are all moving, and Oracle framed the near-tripling as evidence that the bottleneck is easing.
The strategy traces to the company’s founder and chairman, Larry Ellison, who has spent years pushing Oracle toward becoming a cloud and AI infrastructure company after decades as a database and applications vendor. The first-quarter numbers show how far that repositioning has come, and what it costs.
The guidance was a quiet raise. Oracle lifted its full-year adjusted earnings per share outlook to $8.10 from $8.05, a small change that signals confidence in the trajectory without overpromising.
The market’s reaction was forgiving. The stock’s 6.77 percent gain suggests investors are willing to overlook the negative free cash flow and the debt in exchange for cloud growth and a backlog that keeps compounding.
Oracle is still a fraction of the size of the biggest clouds, but its growth rate is among the highest of the group, and its backlog of $664 billion is a measure of how much of the future AI buildout it has already booked. The question is whether it can fund the buildout without stretching the balance sheet further.
The risk Oracle is carrying is a function of its own success. The faster the cloud business grows, the more capital it consumes, and the more the company’s fortunes depend on the AI spending cycle staying hot for years.
The positioning is a sharp reversal for a company that came late to the cloud. Oracle spent its first four decades selling databases and enterprise applications, and its founder, Larry Ellison, spent years dismissing cloud computing before reversing course and rebuilding the company around Oracle Cloud Infrastructure. The AI buildout has given Oracle a second chance to matter in infrastructure, and the first-quarter numbers show it has converted the moment into contracts.
Part of that conversion is the OpenAI relationship. Oracle is one of the partners in Stargate, the data-center venture announced in early 2025 to build computing capacity for OpenAI, and the company has separately signed agreements to supply capacity to large AI customers. Those deals are the multi-year commitments now sitting inside the $664 billion backlog.
For now, the company is selling a story of demand exceeding supply, backlog exceeding capacity, and a balance sheet stretched to build fast enough. The quarter showed all three are true at once, and that the market is willing to pay for the first two while it waits to see the cost of the third.
