Everything About AI
NextFin News - Oracle's first-quarter fiscal 2027 results did something the market had stopped expecting: they turned the company's AI buildout from a financing worry into a growth engine. Cloud infrastructure revenue more than doubled, jumping 121% to $7.4 billion, while the backlog of contracted-but-unbilled work swelled to $664 billion. The stock, down 22% this year entering the report and 5.4% on the day, reversed course and jumped 7% in extended trading on Thursday. The question now is whether Oracle has crossed the point where its data centers earn more than they cost.
For investors who have watched Oracle give back nearly half of last September's post-earnings surge, the quarter offered a cleaner answer than the stock's 2026 drawdown suggested it would. The buildout is no longer a story about how much cash the company must raise. It is becoming a story about how fast it can convert a $664 billion order book into recognized revenue.
The Numbers That Changed the Narrative
For the quarter ended Aug. 31, Oracle reported revenue of $19.35 billion, up 30% from a year earlier and ahead of the roughly $19.13 billion analysts expected. Non-GAAP earnings per share came in at $1.92, beating estimates by 18 cents and rising 30% year over year. GAAP net income rose to $4.68 billion, or $1.56 per share, from $2.93 billion, or $1.01 per share, a year ago. Non-GAAP operating income increased 31% to $8.2 billion, an operating margin of roughly 42%.
The composition matters more than the headline. Total cloud revenue climbed 62% to $11.6 billion, topping the $11.51 billion consensus. Within that, infrastructure as a service — the business where Oracle rents out AI compute — grew 121% to $7.4 billion, beating the $7.09 billion estimate. CPU and GPU work alone contributed $6.5 billion, up 151%. By contrast, the legacy software category slipped 3% to $5.5 billion, below the $5.61 billion forecast. Infrastructure is now roughly 38% of total revenue and growing four times faster than the company as a whole. Infrastructure revenue growth has now accelerated for nine consecutive quarters, from 52% in the fourth quarter of fiscal 2025 to 121% in the first quarter of fiscal 2027.
Against peers, the pace is striking. Microsoft's Azure cloud business, the yardstick for hyperscale growth, expanded about 40% in its most recent quarter. Oracle's infrastructure unit is growing three times that rate off a smaller base — evidence that it is taking share in AI workloads rather than simply riding a rising tide. The trade-off is scale: Azure's absolute dollar additions still dwarf Oracle's, and Oracle's total cloud revenue of $11.6 billion remains a fraction of what the largest hyperscalers book in a single quarter.
The backlog print was the number that drew the most attention. Remaining performance obligations — contracted revenue not yet recognized, plus deferred revenue and uncollected invoices — reached $664 billion at quarter-end, well above the $630.6 billion consensus and up $26 billion from $638 billion at the end of fiscal 2026. That is more than eight times Oracle's last full-year revenue run rate of $67.4 billion. Management told analysts it expects about half of that backlog to convert to sales over the next 36 months.
On execution, the company said it delivered 850 megawatts of data center capacity during the quarter, containing more than 300,000 GPUs — nearly triple the volume delivered in the fourth quarter. It also closed more than $30 billion of additional AI contracts in the quarter without raising additional capital. AI infrastructure ran at 97.9% utilization, and of the GPUs that came up for renewal, the capacity was renewed or resold at a 20% premium to prior contracts, with the majority of those GPUs four years or older.
"New AI contracts won't have any impact on Oracle's plans to raise capital," the company said in its earnings statement.
Why the Buildout Fear Existed — and Why It Is Easing
The skepticism was not irrational. In fiscal 2026, Oracle burned $23.7 billion in free cash flow as capital expenditures jumped 162% to $55.7 billion. The company raised $43 billion in debt and $5 billion in equity that year, and told investors in June it expected to raise roughly $40 billion more in fiscal 2027. A business that must constantly go to capital markets to fund growth is vulnerable to financing conditions — and Oracle carries a lower credit rating than the hyperscalers it is challenging.
This quarter reframed that equation in three ways. First, the spending is producing revenue, not just promises: IaaS grew 121% while 850 megawatts of capacity came online in a single quarter. Second, the backlog is converting: RPO rose $26 billion sequentially even as the company recognized record cloud revenue. Third, the company explicitly decoupled its new contract flow from its financing plan, telling investors the fresh AI deals — more than $30 billion in the quarter alone — would not force changes to its capital-raising strategy.
The cash flow picture is improving at the margin, even if it is not yet comfortable. In the first quarter, operating cash flow was $23 billion against capital expenditures of $28 billion, producing a negative $5 billion in free cash flow. That is still negative, but measured against the $23.7 billion burned across all of fiscal 2026, the trajectory is bending. Customer prepayments and short-term financing arrangements are doing the work of narrowing the gap between the gross capex headline and the net cash outlay that actually matters to shareholders.
The market reaction tells the story of a repricing. Oracle closed at $152.94 on Thursday, down 5.38% on the day — the kind of pre-earnings risk-off move that has become familiar after the stock sold off following two of its previous three reports. Once the results landed, shares jumped 7% in extended trading. That intraday reversal is the signature of a market that had positioned for disappointment and found the opposite. Entering the print, one widely followed options flow monitor showed the busiest trade was selling October through December calls at the $150 strike, with implied volatility among the highest in the S&P 500 — positioning that priced a move, not a direction.
Cyclical Momentum Meets a Structural Shift
The right read is that two forces are at work, and they point in the same direction for now but carry different risks.
The cyclical leg is the AI infrastructure super-cycle itself. Demand for GPU capacity is running ahead of supply, utilization is near full at 97.9%, and renewal premiums of 20% show customers are paying up for scarce compute. That dynamic is mean-reverting by nature: every megawatt Oracle adds today is a megawatt a competitor will add tomorrow, and pricing power compresses as supply catches up. If the AI capex cycle peaks before Oracle has fully depreciated its data centers, the company is left holding expensive, potentially underutilized assets. This is the classic buildout cycle — over-order, over-build, over-supply, reset — and every infrastructure boom from fiber optics to solar panels has run some version of it.
The structural leg is deeper. Oracle is not merely riding a cycle; it is rewiring its business model. For decades it was a database and applications vendor that sold licenses and support. Now infrastructure — a recurring, capacity-rental business — is 38% of revenue and growing at triple-digit rates. That mix shift is permanent in one direction: once enterprise AI workloads are trained and deployed on Oracle Cloud Infrastructure, migration friction keeps them there. The $664 billion backlog is not a speculative pipeline; it is contracted revenue that has to be recognized over the next several years, with management expecting about half to convert within 36 months.
Here is the mechanism that matters: Oracle's problem was never demand. It was whether it could build capacity fast enough, and pay for it without diluting shareholders or choking on debt service. The quarter showed capacity additions accelerating — in June, management said it was targeting almost one gigawatt of capacity in the quarter, roughly the total for all of fiscal 2026, and delivered 850 megawatts in the first quarter. Customer prepayments and short-term financing narrow the gap between gross capex and net cash outlay. The negative $5 billion free cash flow in Q1 is a far cry from the $23.7 billion annual burn of fiscal 2026, and the company's statement that new AI contracts will not alter its capital-raising plans signals management believes funding is already secured or structured through those customer prepayments.
The valuation reflects the doubt that remains. At roughly $153, the stock trades near 20 times the $8.10 fiscal 2027 earnings guidance and about 5 times forward sales. That is a premium to the broader market but a discount to the fastest-growing hyperscalers — a pricing of Oracle as a credible challenger rather than a proven winner. The multiple will expand only if the conversion of backlog into cash keeps pace with the conversion of capacity into revenue.
The Counter-Thesis: Concentration and the Financing Wall
The strongest argument against the bullish read is concentration. Bank of America analysts have estimated that more than half of Oracle's remaining performance obligation traces back to a single customer, OpenAI, with whom Oracle is a partner in the Stargate infrastructure project. If one customer accounts for over 50% of a $664 billion backlog, the quality of that backlog depends heavily on one company's funding, one technology roadmap, and one relationship. A slowdown in OpenAI's capital raising, a shift in its infrastructure strategy, or a renegotiation of terms would hit Oracle's recognized revenue far harder than a diversified book would.
The second pillar of the bear case is the financing wall. Oracle's credit rating sits below the hyperscalers, and it must raise roughly $40 billion in fiscal 2027. If debt markets tighten or equity issuance becomes necessary at depressed prices, the cost of capital rises and the economics of each megawatt deteriorate. The legacy software business, which slipped 3% this quarter, also remains a drag — it is the cash engine that funds the transition, and further erosion there would tighten the constraint.
These are real risks, but the quarter offered partial answers to both. The backlog grew $26 billion sequentially even after recognizing record revenue, which suggests new contract wins are broadening beyond the existing base rather than simply rolling over one customer's commitments. And the company's explicit statement that new AI contracts will not alter its capital-raising plans signals management sees funding already secured or structured through customer prepayments. Still, neither answer is complete until it is confirmed by the next two quarters of disclosure.
What Would Prove the Bull Case Wrong
The bull case rests on one measurable proposition: that backlog conversion and capacity delivery will outpace the cash burn. It fails if either leg breaks. The falsifying signal is specific: if remaining performance obligations grow by less than $20 billion in either of the next two quarters while free cash flow remains more negative than negative $10 billion per quarter, the conversion story is broken — the backlog would be failing to keep up with the spending required to fulfill it. A second warning sign would be any disclosure that the OpenAI share of RPO is rising rather than falling.
Scenarios and What to Watch Next
Three signals will determine whether this is a durable turn or a one-quarter relief rally. First, capacity delivery: Oracle delivered 850 megawatts in the first quarter, and any sustained slowdown below that pace would starve the growth story of its engine. Second, the net cash capex figure — not the gross $28 billion headline. Watch whether customer prepayments and short-term financing keep narrowing the gap between operating cash flow of $23 billion and capital spending. Third, backlog concentration: any disclosure showing the OpenAI share of RPO falling would materially de-risk the bull case; a rise would do the opposite.
The base case is that Oracle converts roughly half of its $664 billion backlog over the next three years while keeping free cash flow modestly negative, funding the gap through a mix of debt, customer prepayments, and selective equity. In that path, the stock's current multiple is fair and modestly accretive as revenue compounds toward the $90 billion-plus guidance for fiscal 2027.
The upside case requires the cycle to last longer than most expect. If AI demand stays supply-constrained through 2027 and Oracle sustains triple-digit infrastructure growth while renewal premiums hold near 20%, the backlog could prove conservative and the stock would re-rate toward the hyperscaler peer group. The trigger is two consecutive quarters of RPO additions above $30 billion with free cash flow narrowing toward breakeven.
The downside case is the classic buildout bust. If AI spending peaks in 2027, utilization falls from 97.9%, and renewal premiums compress, Oracle would be left depreciating expensive capacity against softer demand while still servicing the debt raised to build it. The trigger is RPO growth slowing below $20 billion per quarter alongside a credit-spread widening on new Oracle issuance.
For guidance, Oracle now sees fiscal 2027 revenue of at least $90 billion and non-GAAP EPS of $8.10, above the LSEG consensus of $8.07 per share on $89.76 billion of revenue. For the second quarter, it called for revenue growth of 30% to 34%, cloud revenue growth of 65% to 71%, and non-GAAP EPS of $1.85 to $1.93, up 21% to 25%. Analysts surveyed by LSEG were looking for $1.89 per share on $21.20 billion in revenue.
Data as of Oracle's fiscal first-quarter 2027 earnings release, September 10, 2026.
Bottom Line
Oracle's quarter did not eliminate the risks of its AI buildout — the spending is still enormous, the financing still has to be executed, and the customer concentration is real. But it did answer the question the market was actually asking: the data centers are turning on, the backlog is converting, and the company believes it can fund the rest without changing course. The stock's 7% after-hours jump was the market admitting it had priced a breakdown that did not arrive. The buildout is still a bet on the duration of the AI cycle — but it is no longer a bet that Oracle cannot build.
Explore more exclusive insights at nextfin.ai.
