The slide Oracle put up in Las Vegas on Wednesday carried a figure that had been argued over for months: $225 billion of revenue in the fiscal year ending May 2030. By the close the shares were up 11%.
The target implies average annual growth of 35% for five years, from a company that grew at a fraction of that pace before the data center buildout began. The average analyst estimate was $224.2 billion, so the forecast beat consensus by less than $1 billion. The revenue line was the expected part. What moved the stock was the sentence that came after it.
Oracle promised to raise no more than $20 billion of new financing through fiscal 2030, split evenly between equity and debt. Management argued that supplier financing, hardware supplied by customers and prepayments would hold down the cash it needs up front. Free cash flow turns positive in the second half of fiscal 2028, the company said, and exceeds $30 billion the year after.
A company can promise growth or it can promise restraint. Oracle promised both in one afternoon, and for one session investors took the pairing at face value.
The arithmetic underneath is less comfortable. Capital spending doubled in fiscal 2026 to $55.7 billion, and the company expects another 65% increase this fiscal year, to $91.9 billion. Long-term debt stands at $117 billion, roughly half the revenue target just announced. Oracle’s rating sits at BBB-, the lowest investment grade rung, and the spread on its five-year credit default swaps is 277 basis points. That is what investors charge to insure bonds that the agencies still call safe.
The distance between those two pictures is the story. A forecast of $225 billion five years out is a statement about demand. A capex line of $91.9 billion is a statement about what has already been ordered, built and financed. The first can be revised in a boardroom. The second arrives as depreciation.
Oracle’s answer is that the customer pays first. Under the contracts it has signed, buyers commit capacity and fund hardware, sometimes before the racks are installed, which shifts the working capital burden off Oracle’s balance sheet and onto the companies renting the compute. That arrangement is real, and it is also the reason the cash flow forecast can be credible while the debt load keeps rising.
The product side of the pitch was familiar. Oracle said it would run Nvidia’s Vera Rubin platform and its Vera CPU on OCI, which keeps its cloud in step with the hardware roadmap of its most important supplier and its most important customer base. It also announced an expansion of data centers in the Middle East, where sovereign buyers have been signing multiyear capacity deals and where electricity is cheaper than in most American markets.
Both moves point at the same constraint. Oracle is not competing for the largest share of cloud revenue. It is competing for a specific kind of workload: training and inference for a handful of very large customers, sold on multiyear contracts, delivered from sites built close to power. That business is concentrated by design, and concentration is what makes a 35% growth rate arithmetically possible.
It is also what makes the target fragile. If a few anchor tenants slow their orders, the revenue curve bends and the debt stays. Microsoft reported the same day that its own capital spending had come in below expectations, and the reaction in the shares of the companies that build its capacity was immediate and negative. Oracle’s stock went the other way on the strength of a promise about 2030, which tells you the market is still willing to pay for a long runway.
The mechanics of getting there are less glamorous than the target. Oracle has to convert contracted backlog into billed revenue, which means racks delivered, power connected and customers actually consuming the capacity they reserved. Every one of those steps has slipped somewhere in the industry this year. Memory is short, substations take years, and the transformers that step down grid voltage are backordered.
The financing commitment is the part that will get tested first. Twenty billion dollars over four years is a small number for a company spending $91.9 billion in a single year, and it works only if the prepayments arrive on schedule. If they do not, Oracle has three options: raise more equity than it promised, borrow more than it promised, or slow the build. The first two show up in the next filing.
Management was asked about the tradeoff and gave the answer companies give when the numbers are still theoretical. The cash flow trough is 2028, management said, the recovery is 2029, and the commitment holds in between.
What the presentation did not settle is how much of the $225 billion already sits in signed contracts, and how much depends on demand that has not arrived. Oracle reports fiscal second-quarter results in December. That call will show whether the $91.9 billion is buying revenue or buying capacity, and the difference between those two will decide how long the 11% holds.
