Oracle Is Down 50% — the Whole Bull Case Now Rests on One Cash-Flow Test

Here is the setup Oracle investors woke up to this week. On Thursday the company reported a quarter in which its cloud-infrastructure business grew 121% in a single year. On Friday UBS raised its price target on the stock to $250 from $245 and kept its Buy rating, pointing at cloud growth. Yet on that same day Oracle's shares faded through the session and closed where the week's selloff had left them, around $150. A Wall Street target roughly two-thirds above the current price, on a business growing that fast, and the market still won't let the stock rally. That tension is the whole debate in miniature.

How a top-decile AI name became a 50% loser

The stock is not behaving like a growth darling. Oracle trades at roughly $150, down more than half from its 52-week high near $330 and down over 20% year to date. The decline has a specific origin rather than a general one. In September 2025 Oracle signed a reported $300 billion, five-year cloud-computing contract with OpenAI — one of the largest cloud deals ever written — and markets initially treated the AI buildout as a license to print money. The enthusiasm did not survive contact with the financing. To supply a customer that size, Oracle had to buy data centers and GPUs on a scale that blew up its cash flow and stretched its balance sheet, and in July S&P Global cut Oracle's credit rating to BBB-, one notch above junk, citing a widening free cash flow deficit and heavy reliance on OpenAI.

The market has not simply been moody. The bear case is specific, and it is the reason the stock is cheap. But the counter-evidence is equally specific, which is what makes this a contested setup rather than a falling knife.

The numbers that make the cheap multiple matter

Strip out the debt and measure the operating business, and the growth is extraordinary. First-quarter revenue rose 30% to $19.3 billion, with total cloud revenue up 62%. The engine is infrastructure: IaaS revenue of $7.4 billion grew 121%, while the slower software-applications line grew just 10% and the legacy software business was flat to down. This is not a narrative of future promise — the revenue is being recognized today.

The forward visibility is the part investors rarely see in growth stocks. Oracle's remaining performance obligations, its contracted-but-not-yet-recognized revenue, stood at $664 billion at quarter end, up $209 billion from a year earlier, and the company booked more than $30 billion of new AI cloud contracts in the quarter alone. On a non-GAAP basis management lifted its full-year earnings forecast to $8.10 a share. Do that arithmetic against the $150 price and Oracle trades at a forward non-GAAP P/E of roughly 18.5 times — in line with, or below, the broad market's multiple — while growing revenue at close to three times the market's pace and carrying a contracted backlog that keeps widening.

That is the valuation-disconnect model in its purest form. When a stock's forward multiple compresses toward the market's even as its growth runs at multiples of the market's, the burden of proof shifts to the bears. The bears, however, have a genuine and specific answer, and it cannot be waved away.

The bear fact that must be answered

The cheap multiple exists because the earnings underneath it are engineered around an enormous, single-customer, capital-hungry buildout. Oracle reported negative free cash flow of $5 billion in the quarter, on $28.5 billion of capital expenditure versus $8.5 billion a year earlier — and it guides to roughly $90 to $95 billion of capex for the fiscal year. Operating cash flow set a record at $23 billion, up 184%, but capex run far outpaced it. The company also sold $20 billion of new stock during the quarter to help fund the expansion, which is dilution on top of debt.

This is where the clean contrarian call gets complicated. Reported earnings can look fine while cash is bleeding, because non-GAAP profit adds back costs and inflates the picture, and because depreciating tens of billions of GPUs defers the pain while the cash already left the building. A high-growth company trading below the market's multiple is supposed to shift the burden of proof to the bears — but here the bears are not pointing at vague doom. They are pointing at negative free cash flow, a downgraded balance sheet, and a single customer, OpenAI, that must keep paying for the model to work.

The question that decides who is right

The entire dispute collapses into one testable bet: does Oracle's $664 billion backlog convert into cash-flow-positive economics as revenue finally catches up to the capex, or does the company have to keep spending $90 billion-plus a year just to sustain that growth — leaving shareholders with a chronically negative cash flow and a balance sheet one notch above junk?

UBS, and the aggregate analyst picture, are betting on the first. AInvest's consensus signal labels the stock Buy, consistent with UBS's raised target. That is useful as a read on where positioning sits, but it answers nothing structural; target prices lag sentiment as much as they lead it. S&P's downgrade, by contrast, is the bears' evidence that the structure itself leaks. Neither the analyst target nor the rating settles the question — only the cash-flow inflection does.

So the honest conclusion is not that Oracle is a bargain waiting to be found, and not that the market is right to leave it at $150. It is that this is a discounted bet on a single inflection, and the reader should know which kind of position they would be taking. If free cash flow turns positive while the backlog holds, then the selloff outran the business and the compressed multiple is the opportunity. If the deficit keeps widening even as revenue grows, the market's discount is earned, and Oracle is the capital-heavy, low-margin reseller its critics say it is. The one number to watch is not the next quarter's revenue beat or the next price target — it is whether the free cash flow line, quarter by quarter, stops bleeding.