Oracle Is One Step From Junk—Can It Afford the AI Boom?

Bar Chart ^ | 07/22/2026 | Sam Quirke - MarketBeat

[SeekAndFind]

The past few weeks have seen the bond market start asking a question the stock market has mostly been happy to ignore: Who can actually afford the AI buildout?

On July 9, S&P Global Ratings gave its answer for one of the biggest spenders. It cut Oracle Corp.'s (NYSE: ORCL) long-term credit rating to BBB-, the lowest rung of investment grade and just one notch above junk.

The stock has since fallen to around $125, down nearly 30% over the past month and nearly 50% over the past year.

What makes the downgrade worth more than a passing glance is what it reveals about the wider group. Every major AI spender is pouring money into building out their supply capabilities, but they're all doing it from different financial positions, and the gap between them is widening fast.

The Downgrade Isn't Really About Oracle's Business

S&P's reasoning for the Oracle downgrade was blunt.

The agency admitted it had underestimated the scale of investment required for Oracle's AI ambitions, and now expects the company's free operating cash flow deficit to widen to roughly $42 billion in fiscal 2027, nearly double its earlier forecast.

The other red flag was customer concentration. S&P estimates that around half of Oracle's $638 billion in remaining performance obligations is tied to OpenAI alone, and it described that single relationship as a key credit risk.

The concern is understandable—Oracle has taken on long-term commitments on facilities and equipment to serve OpenAI's demand, so even if the latter's payments were to slow, Oracle's obligations would still need to be paid in full.

Where Each Spender Actually Sits

So what does that mean for Oracle's peers? Well, if you put the four biggest AI spenders side by side, then the divide becomes obvious.

The top two are Microsoft Corp. (NASDAQ: MSFT), which carries an AAA rating, and Alphabet Inc. (NASDAQ: GOOGL), which carries an AA+ rating. Next up is Amazon.com Inc. (NASDAQ: AMZN), still at the higher end of the scale at AA, while Oracle sits alone down at BBB-.

The uncomfortable answer to the question, though, is that none of them can fully fund this from existing cash generation anymore. All four have seen their free cash flow compressed dramatically by the scale of the spending, and all four have been active issuers in the bond market to help cover the gap. What separates them isn't whether they borrow, but how heavily they're leaning on it, and how much of the repayment depends on revenue that hasn't arrived yet.

The Headroom Gap

Microsoft and Alphabet started this cycle with the strongest balance sheets and the most cushion, which is why they still sit near the top of the ratings scale despite spending heavily. Amazon is only just holding on to its positive free cash flow. Still, its accelerating AWS growth is giving investors a reason to stay cautiously bullish on the spending for now.

Oracle is the only one, so far, that seems to have run out of room, and its move into negative free cash flow is a bright red flag. The leverage numbers underscore the point: Oracle sits several times above Amazon on debt-to-equity and nearly 20 times above Alphabet, which is actually running a net cash position.

More importantly, it has no rating cushion left, whereas its peers have several notches to go before their own credit standing comes under real pressure.

Why the Funding Gap Matters

This distinction appears in three places that directly impact shareholders. The first is increased interest expense. Oracle's higher risk profile means that every additional dollar of debt costs more than it would for its better-rated peers, which will hurt its profitability.

The second is a lower likelihood of buybacks. Companies generating strong free cash flow, like Alphabet, can keep buying back their shares while continuing to invest in growth, whereas negative free cash flow will make it far more difficult for Oracle to do the same.

The third is what happens if demand for AI cools. A company funding most of its capital expenditure from operations can simply spend less and wait. A company funding itself predominantly with debt still owes the money it's borrowed, regardless of whether the expected revenue arrives, and its valuation becomes far harder to defend when investors start to doubt it will.

So Who Can Actually Afford It?

Line the four up against the question, and a clear order emerges. Microsoft and Alphabet are best positioned to fund this from what the business itself generates, with the strongest balance sheets, the lowest leverage, and, in Alphabet's case, more cash than debt.

Amazon sits in the middle. It's spending more than anyone, and its cash generation is stretched thin as a result. Still, the accelerating growth at AWS suggests the money is meeting existing demand rather than demand it's hoping to create.

Oracle is the outlier on both halves of the question. It's leaning hardest on borrowing and on revenue that hasn't arrived yet, with a backlog that has to convert and a single customer accounting for around half of it. That combination is why it's the only one of the four with no rating cushion left.

---

TOPICS: Business/Economy; Computers/Internet; Society
KEYWORDS: ai; datacenter; debt; oracle

[eyedigress]

Oracle is out of cash. IF it can get it Wisconsin wants 7bil upfront to build its big data center.

Heads rolled and they are going forward. If you take the bet now it may payoff, or not.

[Wally_Kalbacken]

Wisconsin will drive away big data centers, which will end up in Texas or FL. Too bad, so sad for their taxpayers.

[sauropod]

Bkmk

[jroehl]

GPT Summary:

### Summary

The article argues that **the biggest question in AI is no longer whether companies should invest, but whether they can afford to.**

Its main focus is **Oracle**, whose long-term credit rating was downgraded by **S&P Global Ratings** to **BBB-**, just one step above junk status. The downgrade reflects concerns that Oracle is spending far more on AI infrastructure than previously expected.

### Key points

* **Oracle’s AI spending is straining its finances.**

* S&P now expects Oracle to run a **$42 billion free cash flow deficit** in fiscal 2027.
* Oracle has committed to expensive AI infrastructure before the expected revenue arrives.

* **Oracle is heavily dependent on OpenAI.**

* About **half of Oracle’s $638 billion backlog** is tied to OpenAI.
* If OpenAI’s demand slows, Oracle would still have to pay for the infrastructure it has committed to.

* **All major AI companies are borrowing to fund expansion.**

* Microsoft, Alphabet, Amazon, and Oracle have all seen free cash flow reduced because of massive AI investments.
* The difference is **how strong their balance sheets are**.

### Financial strength ranking

1. **Microsoft** – Strongest (AAA credit rating)
2. **Alphabet** – Very strong (AA+; more cash than debt)
3. **Amazon** – Still healthy (AA), but cash flow is becoming stretched.
4. **Oracle** – Weakest (BBB-), with little financial flexibility remaining.

### Why Oracle is different

Compared with its peers, Oracle:

* Has much higher leverage (more debt relative to equity).
* Has the weakest credit rating.
* Faces higher borrowing costs.
* Has less ability to continue stock buybacks.
* Is more vulnerable if AI demand fails to meet expectations.

### Bottom line

The article concludes that **Microsoft and Alphabet are best positioned to finance the AI race**, **Amazon remains in reasonably good shape**, while **Oracle is taking the greatest financial risk** because it is relying heavily on debt and future revenue—particularly from a single large customer—to justify its AI spending. If AI demand weakens, Oracle has the least financial cushion of the four companies.

[Jonty30]

I’m sure Larry will be ok.

[Myrddin]

Pocatello has an abandoned site from the Hoku polysilicon panel bankruptcy. An AI data center developer purchased the building. Their proposal is for a 100 MW facility. The rub is where does the power come from? Who pays for the build-out? Where does the water come from? Pocatello municipal water is principally from an underground aquifer pumped using electric pumps. Aquifer refresh depends on good Winter snow melt. The past few years have had little snow. Idaho requires a closed-loop cooling approach. The rub is that it needs periodic flushing. It only takes one sloppy flush to permanently contaminate the aquifer that supplies potable water to Pocatello/Chubbuck/Inkom. A pretty high risk issue. Short term power might be provided using natural gas turbines. They are NOISY. The site is against a mountain that will reflect the noise directly into Pocatello. The data center also creates a "heat island" that may adversely affect Winter snow.

Having cited many of the community objections, there is one more thing. The rumored tenant for this facility is Oracle. The focus of this thread shows Oracle to be a weak player. Really another strike against the proposed center.

The "benefits" of the buildout are short term jobs for the construction crews with a small support staff remaining behind. It won't be a source of jobs for current locals. The level of expertise will require outside hires. The revenue stream comes from outside customers buying AI services on the hardware in the building. The hardware/software owners are outside the community. Little or no cash flow accrues to Pocatello for hosting the data center. A small amount of property tax will be collected.

[RitchieAprile]

Nucular might be an option. ID is home to a couple of labs where a private power generating reactor could be built.

[Myrddin]

INL has broached the topic of building the SMR manufacturing facility in Pocatello (real jobs) and locating an operating SMR in Pocatello to provide power. That would address the power and noise issue. The risk of contamination of the aquifer and "heat island" issues would still be a concern.

[Capn Hayek]

Debt load and debt ratings are a metric I think a lot of people miss in their stock analysis - glad this was posted.

When Oracle was surging last fall, I’ll admit I had some FOMO - but I was just having trouble squaring the 300+ price with some of the underlying concerns (debt being one of them). Fortunately, I decided I was too late to the party anyway and stayed away from the spike.

Sometimes, the best stock “picks” are the ones you decide to take a pass on....

[kawhill]

The DROP DATABASE SQL statement permanently deletes an existing database, including all of its tables, views, data, indexes, and stored procedures. It’s a solution Larry.