Oracle at a 52-Week Low: A $638 Billion Backlog Against Negative Free Cash Flow

Few Stocks Split Opinion This Sharply

Oracle is trading near its 52-week low right now, and the bull case and the bear case are both extreme. When I run into a stock like this, I don't try to pick a side first. I put both arguments on the same table and ask which one is built out of numbers I can actually verify.

On one side, these are facts. Oracle just landed a ten-year Department of Defense deal worth up to $7 billion. Its cloud backlog is enormous. Wall Street's average price target is still north of $265, more than double where the stock trades today.

On the other side, these are also facts. Oracle plans to spend $90 to $95 billion on AI infrastructure this year alone. Free cash flow is negative. S&P downgraded its credit rating. And the company is looking to raise $40 billion in new debt and equity, which brings dilution and additional strain on cash flow at the same time.

Analysts love this company long term while the financials right now are genuinely under pressure. Both things are true.

The Bull Case: The Contracts Are Already Signed

First, the cloud infrastructure business quietly became a real alternative. Oracle Cloud now stands as a genuine option next to AWS and Azure, with infrastructure optimized for training large models. Companies like Nvidia and xAI are already using it. Their network architecture lets AI workloads run faster and cheaper than many competitors, which is why demand is surging.

Second, the backlog. Remaining performance obligations reached $638 billion, up 363% year over year. That isn't projected revenue, it's committed future revenue from multi-year commercial deals. Governments, regulated industries, and large corporations are all locking in. That kind of visibility is rare at any size.

Third, the software business is still a cash machine. Products like NetSuite and Fusion ERP sit at the center of how companies operate, so they don't get ripped out easily. Retention is extremely high, and as legacy database customers migrate to Oracle's cloud, margins improve. The Cerner healthcare integration opens another recurring revenue vertical on top of that. While everyone stares at the cloud infrastructure story, the software side is quietly printing money.

The Bear Case: Borrowing Before Earning

First, the spending outruns the business. Capital expenditures are projected at $90 to $95 billion for 2027. That level pushed free cash flow negative, and funding it requires raising $40 billion in debt and equity. So it isn't just spending more than you earn. It's borrowing and diluting to cover the gap. If these bets are wrong, that $40 billion is simply gone.

Second, the balance sheet is under duress. S&P downgraded Oracle to BBB. Credit default swap costs have climbed to around 200 basis points, which tells you the bond market is getting nervous. As older cheap debt gets refinanced at today's higher rates with a weaker rating, interest expense rises and eats directly into margins. The debt load is real and it's growing.

Third, the competitive setup. Oracle Cloud is growing fast but remains the smaller player next to AWS, Azure, and Google, all of which sit on far more cash. As AI hardware supply loosens, pricing wars become possible. Meanwhile Oracle's legacy database business, a high-margin cash cow for years, is structurally declining.

Putting the Three Companies Side by Side

Oracle's position gets much clearer next to Meta and Microsoft, the two companies I looked at alongside it.

Metric	Meta	Microsoft	Oracle
Market cap	$1.4T	$3.4T	$370B
Enterprise value	$1.5T	$3.6T	$540B
Net debt (approx.)	~$100B	~$200B	~$170B
Recent free cash flow	$41B	$67B	-$23B
Recent net margin	under 30%	40%	25%
Eight checks passed	6/8	6/8	3/8

The rows I keep coming back to are the third and fourth together. Oracle's roughly $170 billion of net debt is smaller in absolute terms than Microsoft's $200 billion. But Microsoft is nine times larger by market cap, earns $133 billion in net income, and generates positive cash flow. Oracle carries that debt on a $370 billion market cap with negative $23 billion of free cash flow. The same debt figure weighs completely differently.

On the eight checks, Oracle passes only three: high returns on capital, rising net income, rising revenue. The debt check fails almost automatically because it's measured against five-year free cash flow, which has been close to nothing, and the five-year price-to-free-cash-flow fails for the same reason. The five-year P/E sits at 33, and share count has already increased.

This isn't an unprofitable business. Returns on capital have been decent over five years and ran 11% last year. Margins are actually improving: 22% over ten years, 20% over five, 25% over the last one. Revenue growth is accelerating too, at 6% over ten years, 10% over five, and 10.5% over three. For a software business, the price-to-sales ratio is low, and it trades around 21 times earnings. Looked at in isolation, that combination clearly has upside.

The dividend is the part that bothers me. A company with negative cash flow, whose five-year cash flow is roughly break-even, is still paying one. Where does that money come from? Ultimately from raised capital or newly issued shares.

Analysts See a 4x. I Can't Model It That Way

The analyst estimates are spectacular. Earnings per share go from $7.60 to $33 over seven years, more than a 4x. Revenue goes from $68 billion to $340 billion. Of the three companies here, Oracle clearly has the most growth potential.

If you take those numbers at face value, the math is simple. $33 in EPS seven years out at even a 20 multiple is a $660 stock. From around $120 today, that's more than 5x. The upside claim isn't an exaggeration.

I still can't plug those numbers in. Growing revenue and profit four or five times in seven years requires 30 to 40% annual growth. Applying that assumption to a company that just got downgraded to BBB, needs to raise $40 billion, and runs negative free cash flow is a bridge too far for me. Analysts would laugh at my inputs. I'm using them anyway.

Revenue growth of 8%, 13%, and 18%. Net margins of 20%, 22%, and 24%. Exit multiples of 17, 20, and 23 times earnings ten years out. A 9% required return.

The output: a low price of $111, a midpoint of $200, and a high of $360.

Here's what I found interesting. Even with those deliberately conservative inputs, today's price near $130 still sits well below the $200 midpoint. On my own assumptions, there's still room.

How I'm Actually Reading This

Oracle has the most upside of the three companies I looked at, and the thinnest margin of safety.

Meta and Microsoft are spending enormous sums, but they fund it out of their own cash. Oracle funds it by borrowing. However attractive the valuation output looks, I think that distinction has to be settled before the valuation question, not after it.

My conclusion isn't "don't buy it." It's "the certainty I want isn't there yet." I'd rather wait for a more obvious buy. That wait could be wrong. If the backlog converts into revenue and cash flow turns positive, I'll end up paying more than today's price. That's a mistake I can live with. The other kind, where the $40 billion never comes back, is not.

I've written about who's actually profiting along the AI infrastructure chain in AI Economy Toll Booths: Oracle, Dynatrace, and Tenable, and about the broader big-tech capex landscape in MAG7 Earnings Breakdown — $725 Billion in AI Capex, Clear Winners and Losers.

FAQ

Q: If the backlog is $638 billion, why is cash flow negative?

A: The backlog is work contracted for the future, and fulfilling it means buying data centers and GPUs now. Revenue arrives later, spending arrives first. Whether that's healthy growth investment or overbuilding depends on how much of the backlog actually converts, which takes several more quarters to judge.

Q: How serious is the BBB downgrade?

A: BBB is still investment grade, so it isn't an immediate crisis signal. The problem is cost. Refinancing old low-rate debt at today's rates with a weaker rating raises interest expense, which comes straight out of net income. Credit default swap costs climbing to around 200 basis points tells you the bond market has already started pricing this in.

Q: Why does Wall Street's $265 target differ so much from your $200 midpoint?

A: Different assumptions. Analysts are underwriting 30 to 40% annual growth for seven years. I'm using 8 to 18% revenue growth and 20 to 24% net margins. Time will show who's closer. The point isn't to accept a single price target, it's to build the habit of checking which assumptions sit behind it.