NextFin News - Larry Ellison, the co-founder who has been Oracle's public face for nearly half a century, is stepping out of the spotlight at the precise moment the company's artificial-intelligence gamble is being put to the test. Oracle's fiscal first-quarter earnings on Thursday delivered the numbers Ellison spent years betting the company would produce: cloud infrastructure revenue more than doubled, a record $664 billion backlog, and a stock that jumped 7% after hours. But the man who once dominated more than 150 earnings calls is no longer the one explaining them.
The contrast defines the story. While chief executives Mike Sicilia and Clay Magouyrk now carry the earnings calls, Ellison's machine is running on $125 billion of debt, negative free cash flow of $5.4 billion in the latest quarter, and a $500 billion Stargate data-center vision that has already lost its flagship Texas expansion. Oracle is no longer a founder-led software vendor. It is a capital-intensive infrastructure builder whose fate now depends on whether AI demand compounds fast enough to service a balance sheet Ellison helped load up — and whether the succession he engineered can survive the first real downturn.
The Numbers Behind the Handoff
Oracle's fiscal first quarter, which ended Aug. 31, read like a company in the middle of a transformation rather than a mature software house. Revenue grew 30% year over year to $19.35 billion, and adjusted earnings rose 30% to $1.92 per share, ahead of the $1.74 per share analysts expected. The engine was cloud infrastructure, where revenue jumped 121% to $7.4 billion — more than double the year-earlier figure and above the roughly $7.1 billion consensus. Total cloud revenue climbed 62% to $11.6 billion.
The backlog tells the story investors are celebrating. Remaining performance obligations — contracted but unrecognized revenue, deferred revenue, and uncollected invoices — rose to a record $664 billion, up $209 billion from the same quarter a year earlier and well above the $630.6 billion analysts had modeled. More than $30 billion of that came from new AI cloud contracts booked in the single quarter.
A staggering $209B YoY jump in backlog to $664B backed by $30B in fresh Q1 AI bookings proves OCI demand is compounding exponentially faster than AWS or Azure.
Nicholas Mugalli, chief executive of World Trade Securities, wrote on X after the report. The trajectory is steep: Oracle's backlog stood at $455 billion in the year-earlier quarter and $523 billion in the second quarter of fiscal 2026; it has now more than doubled in four quarters. Oracle's finance chief framed the quarter as a structural inflection, not a one-off.
That's an important sign of our shift to a full technology stack company, all the way from infrastructure to database to software.
Hilary Maxson, Oracle's chief financial officer, told reporters Thursday. She noted it was the first time in the company's history that revenue increased sequentially from the May quarter to the August quarter.
Management raised its full-year fiscal 2027 guidance to at least $90 billion in revenue and $8.10 in adjusted earnings per share, ahead of the $89.76 billion and $8.07 consensus. For the second quarter, Oracle called for revenue growth of 30% to 34% and adjusted earnings of $1.85 to $1.93 per share.
Yet the same report carried the bill for that growth. Total debt stands at $125 billion. Free cash flow was negative $5.4 billion, compared with negative $362 million a year earlier. The company completed a $20 billion at-the-market equity offering during the quarter and allocated another $20 billion to $25 billion for prepayments of essential components — a line item that did not exist in Oracle's capital structure a few years ago. Oracle also delivered 850 megawatts of data-center capacity in the quarter and said the $30 billion in new AI bookings would not change its capital-raising plans. In other words, the spending is not slowing down.
Why Ellison's Exit Is Deliberate — and Why It Matters Now
Ellison gave up the chief executive title in 2014, but he never gave up the microphone. He remained executive chairman and chief technology officer, and he was a constant presence on quarterly analyst calls — absent only for rare exceptions, such as the September 2013 call he skipped for an America's Cup yacht race. This year, that changed. The company's earnings calls have gone forward without his voice, and on the March 10 call he did not deliver an opening statement for the first time in memory, leaving that role to Sicilia and Magouyrk.
This is not an accident of scheduling. It is the final stage of a succession Ellison and longtime CEO Safra Catz designed more than a decade ago. When Ellison stepped down as CEO in 2014, the board's presiding director said Ellison wanted to "keep working full time and focus his energy on product engineering, technology development and strategy." Twelve years later, that is exactly the lane he occupies: long-range vision, product strategy, market positioning, and high-level partnerships, while two professional operators run the company in public.
The timing, however, is what makes the moment tense. Ellison is handing the microphone to Sicilia and Magouyrk not during a quiet patch but at the peak of Oracle's most aggressive expansion since the company was founded 49 years ago. He is the architect of the company's most audacious strategic bet — competing directly against Amazon, Microsoft, and Google in cloud infrastructure — and the one who cultivated the OpenAI relationship that turned Oracle Cloud Infrastructure from an also-ran into the fastest-growing hyperscale platform in the market.
There is a logic to the sequencing. A founder's voice carries the most weight when a strategy needs selling; an operator's voice carries more weight when a strategy needs executing. Oracle is past the selling phase. It is now in the building phase: pouring tens of billions into data centers, negotiating power and cooling contracts, managing component prepayments, and converting backlog into recognized revenue at margins that will determine whether this expansion creates value or destroys it. That is operating work, and it belongs to the CEOs.
But there is also risk in the sequencing. Ellison's personal credibility was the collateral behind Oracle's early AI promises. His absence from the earnings call means the market is now judging those promises against the operators' execution record, not the founder's vision. If the AI buildout hits a pothole — a customer delay, a margin miss, a financing squeeze — there is no longer a figure with Ellison's standing to reassure investors that the long-term thesis is intact.
The Second-Order Question: Who Funds the Buildout, and at What Margin?
The conventional read of Oracle's quarter is straightforward: explosive demand, a record backlog, guidance raised. That read is priced into the 7% after-hours jump. The second-order question is harder: how much of this growth is durable economics, and how much is a capital-intensive wave that must be financed at Oracle's weaker credit profile?
Oracle's credit position is the weak link in the AI infrastructure trade. The company has a weaker cash position and a lower credit rating than the hyperscalers it is competing against, and it is funding a capital-expenditure program that has ballooned from $21.2 billion in the prior year to $55.7 billion in fiscal 2026, with guidance for $70 billion of its own spending in fiscal 2027 plus another $20 billion to $25 billion it expects customers to repay. That is a fundamentally different financial model from Microsoft or Amazon, which fund data-center expansion largely from operating cash flow. Oracle's negative free cash flow of $5.4 billion in the quarter is the price of admission to the AI race — but it is also a claim on future earnings that must be serviced before shareholders see the benefit.
The margin structure matters as much as the funding. Oracle has said adjusted gross margins on its AI infrastructure business would land between 30% and 40% — healthy for a cloud provider, but well below the 70%-plus gross margins that made Oracle's legacy database and applications business a cash machine. As AI infrastructure becomes a larger share of revenue, Oracle's blended margin profile shifts toward the capital-intensive end of its stack. The CFO's "full technology stack" framing is accurate: Oracle is now selling everything from capacity to database software. The question is whether the stack's average economics improve or dilute as the infrastructure layer grows fastest.
Customer concentration adds another layer. A large share of Oracle's AI backlog is tied to a small set of frontier-model developers, with OpenAI the anchor tenant. That relationship has already shown strain. Oracle and OpenAI canceled plans to expand their flagship Stargate data-center campus in Abilene, Texas, after financing negotiations dragged and OpenAI's capacity forecasts shifted. The 600-megawatt expansion that the companies had planned in September 2025 was dropped, with the capacity expected to come from other sites. Reports indicate winter weather damaged liquid-cooling infrastructure at the site earlier this year, straining relations between Oracle and its data-center partner Crusoe — and that Meta has been in talks to pick up the excess capacity, with Nvidia reportedly paying a $150 million deposit to secure the site.
The Stargate episode is a concrete data point on the fragility of the anchor-customer model. When your backlog depends on a handful of customers whose own revenue models are unproven, a shift in their capacity forecasts is not a footnote — it is a direct hit to the revenue recognition schedule that justifies the debt. OpenAI and Oracle still plan to build 4.5 gigawatts of capacity at other sites, and the 1,000-acre Abilene campus remains under development with facilities already in service. But the canceled expansion is evidence that even the most publicized AI infrastructure partnerships are negotiable, revisable, and cancellable.
The market's counter to this concern is the backlog itself: $664 billion of contracted obligations does not vanish because one expansion site is renegotiated. That is true as far as it goes. But backlog is a measure of contracted value, not collected cash, and it is only as good as the customers' ability to pay. If AI monetization lags — if the frontier labs cannot turn compute capacity into revenue fast enough to cover their own bills — the renewal and expansion rates embedded in that $664 billion become the variable that matters, not the headline number.
Cyclical Wave or Structural Shift? The Call That Determines the Stock
Here is the judgment that separates the bull case from the bear case, and it requires separating two forces that the market is blending together. The AI data-center buildout is cyclical: it is a front-loaded capital-expenditure wave driven by a finite number of frontier-model developers racing to install capacity, and it will normalize. History offers a template — the fiber-optic overbuild of the late 1990s, the smartphone supply-chain capex cycle of the early 2010s, the cryptocurrency-mining hardware boom and bust. In each, demand was real, the technology was transformative, and the capital deployed ahead of proven returns still produced a painful reckoning for the most leveraged builders.
But the migration of enterprise databases and applications to the cloud is structural, and that is where Oracle's durable advantage sits. Ellison's early recognition that the database and applications businesses were more valuable together than apart — and his later decision to make that combined stack available on Amazon, Microsoft, and Google's own clouds — created a revenue stream that does not depend on the AI capex cycle. Enterprise customers are not going to rip out Oracle Database because AI spending slows. That installed base, now being monetized through multi-cloud distribution, is the floor under the business.
The correct reading, then, is not "AI is a bubble" or "AI is forever." It is that Oracle is running both a cyclical leg and a structural leg at the same time, and the market is currently paying for the structural story while underwriting the cyclical risk. The cyclical leg — the hyperscale data-center race — will revert toward normalized growth rates once the frontier labs have installed the capacity they need. The structural leg — the cloud migration of the enterprise stack — will compound for years regardless. Oracle's valuation depends on which leg investors decide is the company.
On that question, the evidence cuts both ways, and that is why the stock is down roughly 22% this year even after Thursday's beat, while the S&P 500 has gained about 11%. The market is giving Oracle credit for the growth but discounting it for the leverage, the margin mix, and the customer concentration. That discount is the market's way of saying it has not yet decided whether Oracle is a software company with an infrastructure side business, or an infrastructure company with a software heritage.
The Strongest Case Against This View
The bull case is not weak, and it deserves a direct answer. Morgan Stanley raised its price target on Oracle this week, and the stock rallied 5% on the news, on the view that OCI demand is compounding faster than Amazon Web Services or Azure and that the backlog de-risks years of revenue visibility. The numbers support the momentum: cloud infrastructure growth accelerated from 55% to 68% to 84% to 121% across the past four quarters — a compounding curve, not a plateau. Oracle's remaining performance obligations have risen from $455 billion in the year-earlier quarter to $664 billion. And management's confidence is such that it is raising guidance while spending $70 billion a year.
The bull argument, in its strongest form, is that Oracle has already won the second prize in the AI infrastructure race — behind Nvidia, ahead of everyone else — and that the market is mispricing a company that will compound at 30% revenue growth for years. If that is right, Ellison's retreat is simply the mark of a founder who knows the machine no longer needs his voice, and today's leverage will look modest against the earnings power of a $90 billion-plus revenue base.
The answer to the bull case is not that Oracle will fail. It is that the margin of safety is thinner than the backlog suggests. A $664 billion backlog with a handful of anchor customers, funded by $125 billion of debt and negative free cash flow, at 30%–40% AI gross margins, is a high-operating-leverage bet on AI demand continuing to compound. If OCI growth decelerates to 60% while still being called "explosive," the stock can still re-rate lower, because the valuation is underwriting compounding, not merely growth. Ellison's absence from the earnings call does not change the numbers — but it does remove the one person whose track record has repeatedly convinced the market to look past the numbers.
What to Watch: The Signals That Will Settle the Debate
Three signals will determine whether Oracle's AI transformation is a structural re-rating or a cyclical peak. First, OCI revenue growth: if it decelerates below 60% year over year for two consecutive quarters while remaining performance obligations growth stalls below 30%, the compounding thesis is broken and the cyclical leg is turning. Second, margins: if adjusted gross margins on AI infrastructure print below the 30% floor Oracle has guided, or if blended operating margins compress materially as infrastructure scales, the economics of the buildout are worse than modeled. Third, financing: if Oracle's credit spread widens materially or the company is forced back to the equity market beyond the $20 billion already raised, the market is telling Oracle that its capital structure, not its demand, is the constraint.
By time horizon, the picture splits. In the short term, momentum and the backlog support the stock — the quarter was a beat, guidance was raised, and the AI infrastructure trade remains the market's most crowded long. In the medium term, cash flow is the pressure point: negative free cash flow of $5.4 billion a quarter is sustainable only as long as capital markets remain open and AI demand keeps compounding. In the long term, the structural leg dominates: if Oracle becomes the default cloud home for enterprise databases and AI workloads, today's debt will look like a cheap bridge to a much larger business.
The base case is that Oracle grows into its balance sheet over the next two to three years as the enterprise stack migrates to the cloud, with the AI infrastructure layer providing above-market growth but below-software margins. The upside case is that AI demand compounds for years beyond current expectations, OCI becomes a true third hyperscaler, and Oracle's stock re-rates to a growth-infrastructure multiple. The downside case is that frontier-model monetization lags, anchor customers renegotiate or delay, and Oracle's leverage amplifies a cyclical downturn in AI capex — the scenario in which Ellison's absence from the microphone is felt most acutely.
Larry Ellison built Oracle by being the loudest voice in the room when the room needed convincing. He is now betting that the company he built no longer needs the convincing — that the backlog, the capacity, and the code speak for themselves. The next four quarters will test whether that bet was confidence or hubris. Oracle's AI future no longer has a narrator. It has numbers — and for the first time in 49 years, they will have to be enough.
Explore more exclusive insights at nextfin.ai.
