Oracle Just Hit a Fresh 52-Week Low and Had Its Credit Cut Toward Junk. Has the AI-Capex Panic Overshot?

Shares of Oracle (NYSE: ORCL) touched a fresh 52-week low of $121.50 on Friday. The database and cloud infrastructure company now trades about 63% below its high of $345.72, and its market capitalization has shrunk to about $365 billion.

The new low wasn't even the month's worst news.

On July 9, S&P Global Ratings cut Oracle's credit rating from BBB to BBB-, leaving the company one notch above junk status. The driver was the enormous cost of the AI (artificial intelligence) infrastructure build-out Oracle has signed up for.

The market, in short, is now treating Oracle's AI opportunity as a balance-sheet problem. But with the stock trading at about 16 times the earnings management just guided for this fiscal year, it's worth asking whether the fear has traveled further than the facts warrant.

Image source: Getty Images.

What S&P is worried about

The numbers behind the downgrade are uncomfortable. Oracle spent $55.7 billion on capital expenditures in fiscal 2026 (the year ended May 31, 2026) as it raced to build data centers for AI customers. The company generated $32 billion in operating cash flow, up 54% year over year -- and still spent it all, posting free cash flow of negative $23.7 billion for the year.

S&P expects the gap to widen. The agency projects Oracle's fiscal 2027 capital expenditures will reach $90 billion to $95 billion, and it sees the company's free operating cash flow deficit widening to about $42 billion. Oracle already carried nearly $130 billion in borrowings at the end of fiscal 2026. And after issuing $5 billion of mandatory convertible preferred stock in February, the company plans another $20 billion equity issuance later this calendar year.

There's a concentration problem, too. S&P noted that roughly half of Oracle's $638 billion in remaining performance obligations (the contracted revenue Oracle has signed but not yet delivered) comes from a single customer: OpenAI. If the ChatGPT maker ever struggles to fund its commitments, Oracle could be left holding data centers built for demand that never arrives.

That, to me, is the sharpest risk on the list.

The strain shows up in guidance, too. Management expects revenue to climb about 34% this fiscal year, to $90 billion. But it guided for non-GAAP (adjusted) earnings per share of $8.05 -- about 18% growth once one-time investment gains are stripped from fiscal 2026's figure. That's healthy, but it's still barely half the pace of revenue, because depreciation and interest are climbing alongside the build-out.

Demand isn't the problem

Oracle's fiscal 2026 results, meanwhile, were excellent. Revenue rose 17% year over year to $67.4 billion, and growth accelerated over the year, with fiscal fourth-quarter revenue climbing 21%. The company's cloud infrastructure business, the part of Oracle actually selling AI computing capacity, grew 93% year over year in fiscal Q4 to $5.8 billion. And full-year net income under generally accepted accounting principles (GAAP) rose 37% to $17.1 billion.

The backlog, concentration risk aside, is extraordinary.

Remaining performance obligations ended the year at $638 billion, up 363% year over year and up $85 billion from the prior quarter alone. A year earlier, the figure was about $138 billion. Notably, about $75 billion of the recent large AI contracts involve customers prepaying for graphics processing units (GPUs) or supplying the chips themselves -- an arrangement that shifts some of the build-out's cost off Oracle's books.

Then there's the price. With shares near $127 as of this writing, Oracle trades at roughly 22 times earnings and about 16 times the adjusted earnings management just guided to for fiscal 2027. However, that's a multiple more commonly attached to slow-growing legacy software companies than to a business guiding for 34% revenue growth.

So, has the panic overshot? Partly, I think.

The fear itself is rational. Negative free cash flow, a credit rating one notch above junk, and half the backlog riding on one unprofitable customer are real risks. And the coming share issuance will dilute existing shareholders. But at today's valuation, an awful lot of failure is already priced in -- and if the backlog converts anywhere near schedule, earnings growth could reaccelerate once the heaviest spending is behind the company.

I'm not buying yet, because the thing that would make this stock work (confidence that OpenAI's commitments turn into cash) isn't something Oracle controls. For investors with a strong stomach who believe AI demand is durable, though, a small position arguably starts to make sense at this price. I'd consider changing my mind if free cash flow stops deteriorating ahead of schedule, or if OpenAI's funding keeps showing up quarter after quarter. Those two things matter more than the next rating action.

Daniel Sparks and his clients do not have positions in any of the stocks mentioned. The Motley Fool has positions in and recommends Oracle. The Motley Fool has a disclosure policy.