Oracle's Growth Is Explosive. Its Margin and Cash Math Is the Whole Debate

Oracle just reported a quarter most software companies would frame and hang on the wall. Revenue rose 30% to $19.3 billion in the fiscal first quarter. Cloud infrastructure revenue — the AI datacenter business at the center of this entire story — more than doubled, climbing 121% to $7.4 billion. And the stock barely moved: even after popping at the open, Oracle sits up only about 1% and trades roughly half its high from a year ago.

That gap between a blowout quarter and a flat stock is where the real story lives. The market no longer needs any convincing that Oracle can sell AI computing power. Its backlog of signed contracts now stands at $664 billion. The question investors are actually arguing about is at what margin Oracle delivers all that capacity, and who ends up paying for it. Morgan Stanley, which nudged its price target to $210 while keeping Oracle at an Equal Weight rating, frames it in two parts: near-term IaaS execution has been delivered, but the gross margin and funding questions are unresolved.

The numbers that should have thrilled — and the ones that didn't

Start with the growth report card, because on those factors Oracle is at the top of its sector. Cloud revenue jumped 62% to $11.6 billion, with infrastructure (IaaS) the engine. Remaining performance obligations — the contracted revenue still to be recognized — grew $209 billion in the quarter to $664 billion, helped by more than $30 billion in new AI cloud contracts. Adjusted earnings came in at $1.92 a share, above the roughly $1.74 analysts expected, and the company raised its full-year adjusted EPS guidance to at least $8.10.

Now the part the market is weighing. That growth is being bought with an eroding margin and a heavy cash commitment. The structural reason deserves a plain explanation: Oracle's legacy software business carries fat margins but is shrinking, down 3% as customers shift to the cloud; its IaaS business is surging but carries structurally thinner margins because selling raw compute and datacenter capacity is a higher-cost, lower-margin business than selling software licenses. As the mix tilts hard toward infrastructure, the blended gross margin compresses — Oracle's gross margin holds around 64% today, and its gross profit is growing at roughly half the rate of its revenue. Management itself guided to a step-down in gross margin for fiscal 2027.

The one number that carries the whole debate

Bigger than the margin drift is the cash math. Oracle plans $90 billion to $95 billion of capital spending in fiscal 2027, up from $55.7 billion last year — a figure so large that the company expects customer repayments to cover up to $25 billion of it. In the quarter just reported, operating cash flow hit a record $23 billion, yet free cash flow was still negative at about $5 billion, because the infrastructure outlays outstrip what the business generates. Over the trailing twelve months, operating cash flow of roughly $47 billion sits against roughly $76 billion of capital spending. Oracle also sold $20 billion of stock this quarter and says it will keep raising money.

Here is the whole debate in one move a beginner can feel: a company can grow revenue 30% a year and still lose value, because investors are pricing capital intensity. They are asking whether a firm spending more on datacenters than it produces in cash can eventually turn all that contracted demand into shareholder profit — or whether it is financing its own future with ever more debt and dilution. That is why a $664 billion backlog and a 121% growth quarter produce a stock down about a fifth this year and roughly 50% below its high.

The sector-relative read

Put Oracle next to its software peers and a strange picture emerges. On enterprise value to EBITDA it is the cheapest of the group — about 15.7 times, versus roughly 49 times for ServiceNow and 18 times for Salesforce — while its growth is far faster than those names. On a growth-adjusted basis, Oracle is not obviously expensive. So this is not the old "too expensive to own" story. It is the opposite of the ideal setup a disciplined reader wants to see: valuation and growth look reasonable together, but the profitability and balance-sheet pillars are degrading at the same time.

The momentum tape tells the same story. Oracle is down about 21% year to date, still below its 200-day average, with a middling RSI around 54. The crowd that bid it toward $330 treated Oracle as an AI hyperscaler; the crowd trading it near $154 treats it as a capital-hungry builder on a years-long construction schedule. The aggregate signal from AInvest labels Oracle a Buy with strong fundamentals — a genuinely constructive read that sits in open tension with Morgan Stanley's caution. When a cautious sell-side note and an optimistic aggregate disagree this sharply, what that usually means is the two are pricing different outcomes for the same facts, not that one has a number the other missed.

What the factor stack says to do

The two mistakes here are the mirror images of each other, and both are worth naming. The first is treating a strong quarter as proof the margin problem is solved; it isn't, and management has said so. The second is abandoning a proven winner just because the price fell — Oracle still has top-quartile, accelerating growth and a backlog that guarantees years of visibility. The decisive variable that upgrades or downgrades this story is free cash flow. If this multi-year, roughly $95-billion-a-year buildout converts its contracted revenue into actual profit and cash rather than deeper debt, the eroding-margin fear recedes and the stock's low relative multiple becomes the opportunity. If free cash flow stays negative while margins keep slipping, the low multiple is earned, not generous.

That is the trade to watch: not another quarter of 120% IaaS growth — that part has already been delivered — but the first quarter where Oracle's cash flow turns positive while its gross margin stops falling. Until then, this is a hold-and-monitor name inside a disciplined process, not a buy to be chased on the AI narrative alone.