THE TELL. Five days after saying it needed no more capital, Oracle filed to raise $20 billion. That gap is the whole story.
THE PIVOT. For thirty years Oracle sold software licenses — a business with gross margins near ninety percent, almost no capital expenditure, and customers who couldn't leave. Then it bet the company on data centers for artificial intelligence, and became something Wall Street has never quite known how to price: a software company that suddenly has to build factories.
The results are genuinely spectacular on one side of the ledger. In its first fiscal quarter of 2027, Oracle's revenue grew 30%, with cloud infrastructure leading. Its remaining performance obligations — contracted work not yet billed — reached roughly $664 billion. That is not a company in decline. That's a company with more signed business than it can physically deliver.
THE BILL. And then there's what it costs to deliver it. Oracle's capital spending hit $55.66 billion for fiscal 2026, blowing past its own $50 billion target. That money goes into land, buildings, power and GPUs — assets that depreciate fast in a technology cycle that reinvents itself every eighteen months.
Which means the growth is real and the cash flow is not. Every dollar of new backlog requires Oracle to spend money before it collects any. On June 11, 2026, the stock fell 12% in a single session on exactly this: surging spending and a ballooning debt load. By August, credit raters had Oracle on the cusp of a junk-grade rating.
THE TELL. Here's the moment that made the whole story legible. At its Investor Day, Oracle told investors that a specific amount of additional capital was all it would need. Five days later, on November 2, 2026, it filed to raise $20 billion — $10 billion in new stock and $10 billion in debt.
Five days. That gap between the promise and the filing is the entire investment case against Oracle. It's not that the company is lying; it's that nobody — including Oracle — knows how much this buildout will cost, because the demand is arriving faster than the capacity can be planned. When a CFO has to revise the capital plan within a week, the market starts discounting everything the CFO says next.
THE TENANT PROBLEM. There's a second exposure, and it's bigger. A large share of Oracle's AI backlog is tied to OpenAI, under a compute agreement reported at roughly $300 billion over five years.
That's a magnificent contract and a terrifying concentration. When the Wall Street Journal reported in April 2026 that OpenAI had missed its internal targets for new users and revenue, AI stocks sold off across the board — Oracle and CoreWeave leading the drop. Oracle's backlog didn't change that day. Its customer's ability to pay it did.
This is the structural difference between selling software and being a landlord. A license customer who runs out of money stops renewing. A tenant who runs out of money leaves you holding a building financed with debt you raised on the assumption they'd stay.
WHY GOOD NEWS SELLS THE STOCK. Oracle keeps announcing wins and the shares keep falling, and the reason is arithmetic. Growth funded by equity issuance dilutes existing holders. Growth funded by debt adds interest and covenant risk. Growth funded by both — which is what's happening — means the company can grow 30% and still leave shareholders with less per share than they had before.
The market isn't rejecting AI. It's repricing who is carrying the risk. Nvidia gets paid when it ships the chip. Oracle gets paid when the customer's workload actually runs, for years, at a price that covers the depreciation. Those are not the same bet, and Oracle is on the harder side of it.
IS IT A DECLINE? No — and the framing in the video's title is the thing worth arguing with. A declining business loses customers. Oracle's backlog is growing faster than its ability to build. What's declining is the multiple: the premium investors used to pay for a high-margin software franchise, now replaced by a lower multiple for a capital-intensive infrastructure business with one enormous tenant.
That's a change of species, not a fall. Oracle traded a 90%-margin business with no capex for a 30%-growth business that eats cash. It might win that trade enormously — the backlog says the demand is there. But it is now a company whose health is measured in credit default swaps, and that is a very different life.
SPLASH'S READ: The Oracle story is the AI boom's leverage story in one balance sheet. Everyone talks about the chips; almost nobody talks about who borrows the money to house them. Oracle borrowed, and borrowed again, and then borrowed again five days after saying it wouldn't need to.
If AI demand holds, Oracle becomes the indispensable landlord of the compute economy and this period looks like the smartest bet in enterprise tech. If it wobbles, the $664 billion backlog turns into the most expensive promise in the industry — because the buildings don't disappear when the tenant does.
Either way, the company you knew is gone. It's not declining. It's levered.