Oracle Stock Is Up 60% This Year—Here's What Nobody Tells You

Oracle stock has added more than $200 billion in market value since January. Larry Ellison is richer than ever. And the entire rally rests on a pile of contracts that, as far as anyone outside Oracle can tell, may or may not exist. Let me be clear about what I'm not saying: I'm not saying Oracle is a fraud. I'm not saying the AI boom is fake. I'm saying that when a 47-year-old database company suddenly trades like a hypergrowth startup, the burden of proof sits with the people selling you the story—and right now, the story is doing a lot of heavy lifting. Here's what's actually happening. ## The Numbers That Started the Party In September, Oracle reported quarterly results that sent the stock up double digits in a single session. The headline number was the "remaining performance obligation"—RPO, in Wall Street's alphabet soup—which is essentially contracted revenue not yet recognized. Oracle said its RPO hit $455 billion, up more than 350% year over year. Read that again. A company that did roughly $53 billion in revenue last fiscal year reported future contracted work worth nearly nine times that. Almost all of the increase came from a handful of AI training deals, the kind where a company like OpenAI or xAI agrees to rent massive amounts of computing capacity for years. The market looked at that number and did what markets do: it extrapolated. If Oracle has a quarter-trillion dollars of AI work booked, it must be one of the winners of the AI era. Buy. ## What RPO Actually Is Here's the part that gets lost. RPO is not cash. It's not even a signed check. It's a disclosure of contracts management believes will convert to revenue over time—and the accounting rules around what counts as a "contract" are looser than most retail investors assume. More importantly, Oracle doesn't break it down. We don't know how much of that $455 billion comes from a single customer. We don't know the cancellation terms. We don't know how much depends on those customers actually raising the money to pay for it. And we don't know the margins—because building out AI data centers is a brutally capital-intensive business, and Oracle is spending tens of billions on GPUs, land, power, and debt to fulfill promises that may take a decade to pay off. One customer concentration is the quiet risk in every AI infrastructure story right now. If a meaningful chunk of that backlog traces back to one or two startups burning venture capital, then Oracle's backlog is really a bet on those startups' fundraising prospects, not on Oracle's own execution. ## The Circular Economy Problem There's a word for what's happening across the AI buildout: circularity. Chipmakers invest in AI labs. AI labs sign compute contracts with cloud providers. Cloud providers buy chips from the chipmakers. Money moves in a loop, and every lap makes the loop look bigger. Nvidia is an investor in some of the same companies buying Nvidia chips. Oracle is signing enormous contracts with AI labs, some of which are backed by the same players. None of this is illegal. It's not even necessarily bad. But it does mean that the "demand" showing up in these backlogs is partly financed by the same capital markets that are cheering the backlog. That's a feedback loop, not a foundation. If AI startup funding slows—and venture funding is nothing if not cyclical—the contracts don't vanish overnight, but the ability to pay for them can. And unlike a consumer subscription, you can't just stop paying for a data center lease without a lawyer and a lot of explaining. ## What Oracle Is Actually Good At None of this means Oracle is a bad company. It's not. Oracle's core database business is a genuine moat—switching costs in enterprise databases are enormous, and the company has spent decades making itself nearly impossible to remove. Its cloud infrastructure business is real and growing. The stock could keep climbing for entirely legitimate reasons. But there's a difference between a good company and a good stock at any price. Oracle is now trading at a valuation that assumes years of flawless execution in a business—AI data centers—where the company is a relative newcomer competing against Microsoft, Amazon, and Google, all of which have deeper pockets, more experience, and their own chip designs. Oracle is taking on enormous debt to build this out. Its capital expenditures are exploding. Free cash flow, the thing that actually matters, is under pressure. And the stock is priced as if none of that risk exists. ## Who Benefits From You Believing Follow the incentives. Every analyst upgrade, every bullish media segment, every "Oracle is the next Nvidia" take has someone on the other side of it. Investment banks earn fees from the debt Oracle is issuing. Fund managers need the trade to keep working. And the loudest voices rarely disclose how much of their own money is riding on the narrative they're selling. That doesn't make them wrong. It makes them interested parties. And you should read interested parties differently than you read disinterested ones. The honest answer to "is Oracle stock a buy?" is: nobody knows, and anyone who tells you otherwise with total confidence is either selling something or guessing. The AI infrastructure buildout may turn out to be the industrial revolution of our time. It may also turn out that a lot of the contracts underpinning today's valuations were signed by companies that won't exist in five years. Both things can be true at once, and the stock price doesn't tell you which one is winning. What I'd want, as a skeptical observer, is disclosure. Break out the backlog by customer. Show cancellation terms. Explain the margins. Until then, the $455 billion number is a headline, not a fact you can underwrite. ## The Bottom Line Oracle might be a great investment. It might be the most crowded trade in tech. The uncomfortable truth is that the information needed to tell the difference isn't public—and the people who have it aren't in a hurry to share. Treat the hype like what it is: a sales pitch with a stock ticker attached. Do your own math, size your position like you could be wrong, and remember that every bubble in history was, at the time, called a revolution. Some of them even were.