Oracle's real test this week is not the beat. It is whether the backlog finally becomes revenue

Oracle reports fiscal first-quarter results after the close on Thursday, September 10, after the close of the market. Here is the uncomfortable starting point: the stock has been cut roughly in half from its 52-week high near $330, even as the company has reported record revenue, a record backlog, and quarterly beats. The market is not doubting the size of Oracle's AI commitments. It is doubting how fast those commitments turn into revenue and cash — and that is exactly the question this week's numbers get to answer.

The scoreboard: a beat that already stopped paying

Oracle's fiscal year ends in May, so this print covers the June-through-August quarter. Analysts consensus expects Q1 fiscal 2027 revenue of $19.13 billion and Oracle guided non-GAAP earnings between $1.72 and $1.76 a share, with total revenue expected to grow 27% to 29% and cloud revenue 58% to 64%.

The history behind those numbers explains why guidance this week is the load-bearing part. A year ago, Oracle's Q1 report set off a record 36% single-day rally when remaining performance obligations — the backlog of signed but not-yet-recognized contracts — jumped 359% to $455 billion, and OpenAI announced a roughly $300 billion, five-year cloud deal with the company. That was the moment the market anointed Oracle the purest AI-capex story and the stock ran to its high.

The problem came when expectation inflation ran ahead of the ramp. In June, Oracle reported fiscal Q4 revenue of $19.2 billion, up 21%, with cloud infrastructure up 93% — and the shares still fell about 7% after the report. The reason was the same scoreboard quirk every quarter now has: strong operating results against a hurdle the market had already priced, while cloud peers like Microsoft and Google were simultaneously raising their guidance. A good quarter stopped being enough.

The conversion gap is the whole debate

The core tension is a number, not a story. Oracle ended fiscal 2026 with $638 billion of remaining performance obligations, up $85 billion in a single quarter to the record $638 billion. That is roughly 33 quarters' worth of revenue at today's run rate. The market's entire question is whether that enormous backlog converts to recognized revenue on a visible schedule or just sits there growing larger than the business.

In the quarter alone, Prepaid and customer-supplied GPU portions of large AI contracts totaled $75 billion. That structure is clever — it reduces the cash Oracle must raise — but it also concentrates the numbers in a handful of giant customers, with OpenAI's $300 billion commitment as the anchor. A backlog that lives in a few enormous contracts is powerful and fragile at the same time.

Meanwhile the bill for building all that capacity is coming due on the other side of the ledger. Fiscal 2026 capital expenditures hit $55.7 billion, free cash flow was negative $23.7 billion, and net debt is roughly $98 billion. Oracle expects to raise approximately $40 billion through debt and equity, including a $20 billion at-the-market issuance. None of this is a surprise, and it is not automatically a failure — this is a buildout, and buildouts run cash-negative before they run cash-positive. But it does mean the free-cash-flow number has nowhere to go but up, because Oracle cannot finance a widening hole forever.

What this week actually has to confirm

So when Oracle reports, ignore the headline beat-or-miss and watch three things that determine whether the thesis holds.

First, cloud revenue landing at the top of the 58% to 64% guided range — or above it. Management's own five-year map has Oracle Cloud Infrastructure climbing from $18.1 billion in fiscal 2026 to roughly $32 billion in fiscal 2027, then $73 billion and $114 billion. This quarter is the first real proof point that the ramp is on schedule, not just forecast on a slide.

Second, whether the backlog keeps growing. The cleanest evidence that demand is real is an RPO that still expands even as revenue accelerates — the sign that signed customers are converting into satisfied ones who buy more. A flat or shrinking backlog while revenue grows fast is actually fine for cash flow, but it would tell you the giant AI contracts are one-time, not recurring.

Third, any signal about the cash trough. Oracle has already told investors fiscal 2027 revenue should be about $90 billion and non-GAAP EPS raised to $8.05. The market is not going to re-rate the stock higher just because Oracle restates those numbers; it needs to hear that capex peaks and free cash flow inflects within sight. The willingness to keep raising $40 billion a year without visible conversion is what turns a bullish buildout story into a balance-sheet worry.

That last point is the deeper one. The old question was whether Oracle could keep signing giant AI contracts. The new question is whether it can turn $638 billion of promises into steady revenue and eventually cash — before the financing costs and dilution erode the value of the backlog it is so proud of.

The most useful single checkpoint is cloud revenue growth: if it comes in above the guided range and the backlog holds, that is the first credible proof the construction phase is becoming a revenue phase, and the de-rated multiple starts to look interesting again. The signal that would break the narrative is the opposite — capex raised again while revenue conversion stalls, or guidance below what peers are already delivering. That combination would tell you Oracle is spending ahead of where the demand curve actually is. Thursday will tell us which story the numbers support.