Oracle's AI Cloud Is Booming. Its Cash Burn and Debt Just Spooked Wall Street.

Oracle just delivered one of the strongest quarters in its history, and investors responded by selling the stock as hard as they have in more than a year. That contradiction is the story of where the artificial-intelligence boom now stands.

The software-and-cloud company reported record fourth-quarter and full-year fiscal 2026 results on June 10, headlined by an order backlog that has swollen to $638 billion. A day later, on June 11, the shares slid about 12 percent, wiping roughly $72 billion off a market value that had stood near $579 billion. What unnerved the market was not the revenue. It was the bill coming due to produce it: a $23.7 billion free-cash-flow deficit for the year, capital spending running far ahead of plan, and a decision to raise tens of billions of dollars more in debt and equity to keep building.

Why It Matters

Oracle has reinvented itself as one of the central suppliers of computing power for AI, renting out data-center capacity to companies training and running large models. That pivot has worked spectacularly on paper. But it has also turned Oracle into a case study in the question hanging over the entire AI trade: whether the cash being poured into data centers, chips and power will earn an adequate return, and how companies without enormous existing profits will finance the buildout in the meantime.

The largest cloud builders — Amazon, Microsoft and Google’s parent, Alphabet — fund their AI spending out of mature, highly profitable businesses that throw off cash. Oracle does not have cash flows on that scale, which means its expansion is being paid for by borrowing and by burning through cash rather than generating it. For the full year, Oracle reported negative free cash flow of $23.7 billion, a sharp reversal from a deficit of less than $400 million the year before. The company is spending the money now and counting on the backlog to convert into revenue and profit later.

That is a different risk profile than the rest of Big Tech, and it explains why a record quarter could trigger a sell-off. The same dynamic has been visible across the sector as the buildout collides with the physical limits of power and infrastructure needed to run AI data centers. When the constraint shifts from demand to the cost and financing of capacity, investors start scrutinizing balance sheets as closely as growth rates.

The Backlog Against the Cash Burn

The bullish case is written in the backlog. Oracle’s remaining performance obligations — contracted business not yet recognized as revenue — reached $638 billion at the end of the quarter, up from $553 billion just three months earlier and more than triple the figure a year ago. Much of that surge came from a handful of very large AI contracts; the company has said roughly $75 billion of the backlog ties to deals where customers either prepaid for graphics-processing chips or supplied the chips themselves.

The underlying business grew fast, too. Fourth-quarter revenue rose 21 percent to $19.2 billion, with total cloud revenue up 47 percent to $9.9 billion. Oracle Cloud Infrastructure, the unit that rents raw computing capacity, jumped 93 percent to $5.8 billion. For the full fiscal year, revenue climbed 17 percent to a record $67.4 billion. By the standard measures, this is a company accelerating, not slowing.

The bearish case is written in the capital budget. To deliver on those contracts, Oracle spent $55.7 billion on capital expenditures in fiscal 2026 — up about 162 percent and well above its earlier $50 billion guidance. And the company signaled that the figure climbs from here, telling investors it expects net capital spending of roughly $70 billion in the current fiscal year as it races to bring data centers online for customers including OpenAI. Some analysts think even that may prove conservative: one note flagged that the total could approach $100 billion as the OpenAI-linked “Stargate” data-center project expands.

That is the tension at the heart of the report. A backlog of $638 billion is meaningless if the company cannot finance the construction required to fulfill it — and the construction is precisely what is draining the cash. Oracle is, in effect, asking investors to fund years of heavy spending against a promise of future returns, at a moment when its traditional software business faces pressure from the very AI tools its cloud is built to support.

The Market Reaction

The selling was decisive. Oracle’s roughly 12 percent drop marked its steepest one-day decline in more than a year and erased a chunk of the gains that had made it one of the AI era’s standout large-cap stocks. The trigger was less the earnings beat than the financing plan attached to it: Oracle said it would raise about $40 billion through a mix of debt and equity, including a previously flagged $20 billion stock sale, on top of the $43 billion in debt and $5 billion in equity it raised during fiscal 2026.

Analysts at Melius Research captured the unease, writing that “it is hard to know if Oracle can stick to this capex plan if incremental business arises from the likes of OpenAI and Anthropic,” and adding that “its competitors are unlikely to slow spending and could use Oracle’s spending moderation as the means to gain share.” In other words, Oracle may be trapped: pull back on spending and risk losing AI customers to deeper-pocketed rivals; keep spending and deepen the cash deficit and debt load.

The reaction also reflects a jumpier market backdrop. Investors have grown more discriminating about AI names after stretches of volatility tied to shifting expectations for Federal Reserve interest-rate cuts and the latest inflation readings. Higher-for-longer rates raise the cost of exactly the kind of debt-financed expansion Oracle is undertaking, which makes the borrowing plan land harder than it might have in an easier-money environment. The episode stands in contrast to a name like Nvidia, whose record results this year were powered by selling the chips that everyone else is buying — a far more comfortable position than borrowing to install them.

What Comes Next

The near-term signposts are concrete. Watch how Oracle structures and prices the $40 billion in new financing, since the terms it can command will say a lot about how credit markets view the bet. Watch whether the headline capital-spending number for the current year holds near $70 billion or drifts toward the higher estimates as data-center commitments firm up. And watch the conversion rate — how quickly that $638 billion backlog turns into recognized revenue and, eventually, into the cash flow needed to service the debt.

The broader question is whether Oracle’s predicament is specific to Oracle or a preview of strains to come across the AI economy, where the appetite to build has outrun the cash on hand to pay for it. For now, the market has rendered a verdict that would have seemed strange a year ago: a company can post record growth, win a backlog larger than some national economies, and still be punished for what that growth costs. Readers tracking the sector can follow our business and economy coverage as the financing question plays out.