Remember when tech companies used to brag about having fortress balance sheets and billions in cash? Yeah, those days feel like ancient history now.
Today, if you’re not mortgaging your future to build AI infrastructure, you’re apparently yesterday’s news. And few companies have embraced that philosophy quite like Oracle.
The software giant that Larry Ellison built is in the middle of one of the most aggressive — and expensive — transformations in tech history. And with earnings coming tomorrow, the market is holding its breath.
A Debt Load That Would Make Even Elon Pause
Let that sink in for a second: Oracle now carries more investment-grade debt than any other non-financial company in America. We’re talking $111.6 billion as of last quarter, and that’s after raising an eye-watering $18 billion in a single bond offering back in September.
Eighteen billion. In one go. That’s not just big — that’s historically massive for tech.
I’ve been following tech earnings for years, and I can’t remember the last time a company went this hard on leverage for growth. Even Amazon during its AWS buildout phase looks conservative by comparison.
“There is something inherently uncomfortable as a credit investor about the transformation… that is going to require an enormous amount of capital.”
Head of U.S. investment grade credit strategy at a major bank
What Exactly Are They Building?
The short answer: pretty much everything they can get their hands on.
Data centers in Texas. Data centers in New Mexico. Data centers in Wisconsin. Construction loans. Bond issuances. They’re even exploring whether suppliers might finance GPU purchases with the chips themselves as collateral — which honestly sounds like something you’d see in emerging markets, not from a 47-year-old tech titan.
And the catalyst for all this? A partnership that sent the stock soaring 36% in a single day back in September when details first emerged.
The Deal That Changed Everything
When Oracle announced it had secured what sources describe as roughly a $300 billion commitment for compute capacity over five years — starting in 2027 — the market initially went wild.
Think about that number. Three hundred billion. Over five years. That’s larger than the GDP of many countries.
Of course, the stock has since given back all of those gains and more, because Wall Street started doing the math on what it actually costs to deliver that kind of capacity.
- Land acquisition
- Power infrastructure (and AI training eats power like nothing else)
- Cooling systems
- The GPUs themselves — at $40,000+ each and climbing
- Fiber connectivity
- Construction labor in a market where everyone is building data centers
Suddenly that $300 billion commitment started looking less like free money and more like an anchor that could pull the entire company under if execution falters.
The Credit Market Is Already Voting
You know things are getting serious when credit default swaps start making headlines.
Oracle’s 5-year CDS recently hit multi-year highs. That’s basically insurance against the company defaulting, and when those premiums spike, it means the smart money in credit markets is getting nervous.
Even more telling? Analysts at major investment banks are now openly recommending clients buy protection. One called it having “no clear path to credit improvement.” Another noted they’re seeing “tourists” — investors who normally don’t touch CDS — piling in specifically to hedge AI-related names.
When the credit market starts treating you like a speculative bet rather than a blue-chip credit, that’s rarely a good sign.
But Wait — The Growth Story Is Real
Here’s where it gets complicated, because I’m not entirely bearish here.
The demand is undeniably there. Companies are falling over themselves to secure AI training capacity years in advance. If you have the infrastructure ready when that demand hits, you’ll print money.
Analysts expect remaining performance obligations — basically contracted but not-yet-recognized revenue — to top $500 billion this quarter. That’s up from about $90 billion a year ago. Fivefold growth in booked business is the kind of metric that makes growth investors salivate.
Revenue is expected to grow 15% this quarter, which would be Oracle’s fastest growth in decades.
The Cash Flow Question
But here’s the rub — and this is what keeps me up at night thinking about this name.
All that growth? It’s coming from the cloud infrastructure business, which has much lower margins than Oracle’s traditional database cash cow.
They’re essentially trading high-margin annuity revenue for lower-margin growth revenue, while simultaneously levering up the balance sheet to build the capacity needed to deliver that growth.
It’s a classic growth-at-all-costs playbook, but executed by a company that spent decades being the anti-growth-at-all-costs company.
“Oracle can handle the debt load. But they need more cash flow to raise more capital from here.”
DA Davidson analyst Gil Luria
Creative Financing or Desperation?
The options being floated now are… creative, to put it mildly.
- Off-balance-sheet financing vehicles
- Joint ventures with private credit funds (Meta just did $27B of these)
- Sovereign wealth fund investment
- Vendor financing with GPUs as collateral
- Potentially even equity issuance (though that would be painful at current prices)
Some of these make sense. Others feel like the kind of financial engineering you see right before things get interesting — and not in a good way.
What Tomorrow’s Earnings Need to Show
For the bull case to remain intact, management needs to deliver three things:
- Continued acceleration in cloud infrastructure revenue growth
- RPO growth that justifies the capex spending (preferably north of $500B)
- Some visibility into how they’re going to finance the next $60-90 billion in spending analysts expect over the next three years
Anything less, and we could be looking at another leg down.
But if they can thread that needle — if they can show the demand is real, the pipeline is robust, and they’ve got credible plans to finance the buildout without destroying the balance sheet — then this could be one of the great transformation stories in tech history.
I’ve learned never to bet against companies when they’re in the early innings of a genuine demand supercycle. And AI infrastructure spending looks a lot like a supercycle to me.
The question is whether Oracle can execute fast enough, and finance aggressively enough, to capture their share before the window starts closing.
Tomorrow, we’ll get the next piece of that puzzle.
Buckle up.
