Article Directory
Larry Ellison's AI Gambit: Is Oracle's Half-Trillion Dollar Backlog Built on Sand?
*
You don’t lose $38 billion in two trading days by accident.
For Oracle’s founder and CTO, Larry Ellison, the mid-October stock slide was a brutal, if temporary, correction. After a euphoric, AI-fueled run that saw his net worth briefly touch $393 billion in September, a sudden chill hit the market. Oracle’s stock fell 7% on October 17th and another 5% on the 20th. It was a classic case of profit-taking, a moment of sobriety after a year-long party that left even the Magnificent Seven in its dust.
But to dismiss this as mere "buyer exhaustion" is to miss the point entirely. The pullback wasn't just a technical correction; it was the market’s first serious attempt to reconcile Oracle’s audacious, almost fantastical AI promises with the grim realities of execution. The core of Oracle’s bull case now rests on numbers so large they border on the abstract. The real question isn't whether the stock will bounce back to $300. It's whether the half-a-trillion-dollar foundation Oracle is building its future on is made of solid granite or loosely packed sand.
The Anatomy of an Astronomical Promise
Let’s be clear: the numbers Oracle presented are staggering. In its last quarterly report, the company announced its Remaining Performance Obligations (RPO)—essentially its backlog of contracted revenue—had exploded to roughly $455 billion. The growth was about 360%—to be more exact, 359% year-over-year. CEO Safra Catz then casually mentioned that the RPO would likely exceed half a trillion dollars in the coming months. To top it off, management raised its fiscal 2030 revenue guidance to $166 billion, implying a growth rate that JPMorgan’s analyst politely called "virtually unheard-of" for a company of this scale.
The centerpiece of this narrative is a series of massive AI cloud deals, most notably a reported $300 billion, five-year contract with OpenAI. Wall Street analysts like Jefferies’ Brent Thill called the quarter "truly historic," while Deutsche Bank’s Brad Zelnick labeled the results "truly awesome." The market, for most of 2025, agreed.
But a backlog of this magnitude is like a restaurant that decides to take reservations for every meal it will ever serve for the next decade. The booking numbers look phenomenal. You can issue a press release claiming you’re sold out until 2035. The problem is, you haven’t built the kitchens yet. You haven’t hired the chefs, secured the supply chains, or figured out how you’re going to pay for the literal power plants needed to keep the lights on. The promise is easy; the execution is everything.
And this is the part of the report that I find genuinely puzzling. Oracle has been remarkably vague on the financing details for the colossal data-center buildout required to service this backlog. Ellison talks of a new facility in Texas consuming 1.2 gigawatts of power and housing half a million Nvidia GPUs. That’s not just a data center; that’s a piece of national infrastructure. How is this being funded? What are the capital expenditure projections? And what is the actual, non-cancellable, guaranteed revenue within that half-trillion-dollar RPO figure? These are not trivial questions.
The Financial Gravity Problem
This brings us to the OpenAI deal, a contract that perfectly encapsulates the tension between Oracle’s narrative and numerical reality. OpenAI, a company that has raised somewhere between $60 and $70 billion in its lifetime and currently operates at a loss, has supposedly committed to spending $300 billion with Oracle.
Let that sink in.
How does a company with that financial profile secure a line of credit or make a firm commitment for an amount that is more than four times its total historical funding? When Oracle’s co-CEO was asked if OpenAI actually has the money, his answer was a simple, "Of course." That is not a sufficient answer for anyone doing serious due diligence. It raises immediate flags about the structure of the deal. Is it a flexible usage agreement? Is it contingent on OpenAI’s own future fundraising and profitability? The discrepancy is glaring.
This is why some skepticism is beginning to creep in. Redburn Partners initiated coverage with a "Sell" rating and a startlingly low $175 price target, calling the OpenAI deal "very risky." Morningstar’s Dave Sekera quipped that the 2030 targets give him "late-90s… vibes," a nod to the dot-com bubble’s habit of valuing promises over profits. The stock now trades at roughly 11 times sales (a steep multiple for a company of its maturity, though still below peers like Microsoft).
Ellison’s own rhetoric at the recent AI World conference only adds to this sense of a reality distortion field. He spoke of what some have called Larry Ellison's latest craze: Vectorizing all the customers, a plan to use their data—apparently without asking for consent—to predict their next purchase. He talked of building AI-powered ambulances and using synthetic biology to have crops sequester carbon from the atmosphere. These are grand, world-changing visions. But they are also a convenient distraction from the more prosaic, and far more difficult, questions about cash flow and capital expenditure. When the story becomes this grand, my analyst instincts tell me to look closer at the balance sheet.
The market has priced Oracle for perfect, flawless execution of an unprecedented industrial buildout. It has baked in the assumption that a half-trillion-dollar backlog will translate directly into revenue without any significant hitches, financing problems, or client defaults. That is a bold assumption. The recent $38 billion haircut for Ellison suggests that, finally, some investors are starting to question it.
The Execution Discount is Coming
My analysis suggests the market is valuing Oracle's story, not its operational risk. The current valuation reflects a belief that turning a half-trillion-dollar order book into realized profit is a certainty. It is not. The chasm between OpenAI's supposed $300 billion commitment and its known capital resources is a microcosm of the entire Oracle bull case: the numbers are breathtaking, but they strain credulity upon closer inspection. The stock's recent dip isn't the end of the story, as one analyst put it. It's the beginning of the market applying a long-overdue discount for execution risk. The real test for Oracle isn't securing the next big AI contract; it's surviving the brutal financial and logistical reality of fulfilling the promises it has already made.