Oracle’s $523 Billion Question Mark

If you thought Oracle was just that legacy database company your IT department complains about, well, it is but they’re also trying to change that. The tech giant has been rebranding itself as the heavy lifter of the AI revolution, building the massive data centers required to train the brains of tomorrow. But as their recent earnings prove, transforming a big boy like this is messy, expensive, and—judging by the market’s reaction—a bit terrifying for investors.

The Headline Numbers: A Miss, a Beat and an Asterisk or Two

First, let’s talk about what Oracle actually does today. Yes, they still sell software licenses and ERP systems (and quite successfully), but the growth part of the business—and the part driving the valuation(the stock is still up for the year even after the recent correction)—is Oracle Cloud Infrastructure (OCI). They are building the massive superclusters that companies like OpenAI, Meta, and NVIDIA need to train their models.

The Revenue Miss

For Q2, Oracle reported total revenue of $16.1 billion, up 13% in constant currency. While double-digit growth for a company of Oracle’s size is nothing to sneeze at, it technically missed the $16.2 billion the market expected. The market was looking for a little more top-line sizzle given the AI hype cycle we are currently riding and the massive backlog they talked about last quarter. Guess it’ll take a bit longer to get those data centers up and running and have them generate revenue from those new deals.

The EPS “Beat”

On the bottom line, things looked rosier—at first glance. Non-GAAP Earnings Per Share (EPS) came in at $2.26, up a staggering 51% year-over-year. That sounds like a blowout quarter, right?

Not so fast.

If you dig into the fine print, you’ll see that this number was heavily heavily heavily lifted by a one-time event: a $2.7 billion pre-tax gain from the sale of Oracle’s interest in Ampere, a chip company.

If you strip out that massive one-time windfall(which is worth about 90 cents before taxes), that impressive EPS beat starts to look a lot more like a miss. The core operating business didn’t generate that profit growth on its own; it had a little help from an asset sale. Pure operating income was only up 9% in constant currency lagging top line growth which means margins are being pressured a bit as well. That realization was the second strike against the stock price combined with the revenue miss.

The RPO Rocket Ship: $523 Billion and Counting

If the current earnings were a mixed bag, the future looks absolutely blindingly bright—at least on paper.

The star of the show was Remaining Performance Obligations (RPO). This was what sent the stock soaring last quarter driving the stock price from the 200s into the 300s pretty quickly. This is essentially the total value of contracts Oracle has signed but hasn’t fulfilled yet. That looked good on paper but I guess it didn’t have a ton of lasting power in the minds of investors.

• Total RPO: $523 billion.
• Growth: Up a mind-melting 438% year-over-year.

To put that in perspective, Oracle effectively has half a trillion dollars of future revenue locked in. This surge was driven by massive new commitments from AI royalty like OpenAI, Meta and NVIDIA, along with others who are desperate for GPU capacity.

This backlog suggests that demand for Oracle’s AI infrastructure is virtually insatiable. They aren’t just selling cloud space; they are selling the shovels for the biggest gold rush in tech history.

So Why Did the Stock Tank?

You have a $523 billion backlog. You have partnerships with the biggest names in tech. Why is the stock down so much in the last few months?

The answer lies in one word: Reality.

The market is suddenly realizing that turning $523 billion of contracts into actual profit is going to be wildly expensive and risky. Here is what spooked the street:

The CapEx Black Hole

To fulfill those contracts, Oracle has to build data centers. A lot of them. And they have to fill them with NVIDIA GPUs, which aren’t cheap. During the call, management admitted that Capital Expenditures (CapEx) for fiscal 2026 would be $15 billion higher than they previously forecasted.

Investors did the math and realized Oracle is going to be burning cash at an unprecedented rate to build out this infrastructure. This is a company that was previously generating $10B+ in free cash flow and will now likely be burning more than that for the next few years to fulfill those obligations.

The Debt Trap

Where does the money come from? Oracle is already sitting on a mountain of debt. The market is worried that to fund this Field of Dreams strategy (If we build it, they will train), Oracle will have to leverage itself to the hilt. High debt in a high-interest-rate environment is a red flag for conservative investors. Their interest expense was already up 22% versus last year and this is just the beginning of their investment cycle.

The $523 Billion Dollar Question Mark. Is It Real?

Finally, there is a nagging skepticism about the quality of that RPO. Are these contracts rock-solid? What if the AI bubble bursts? What if these startups promising to pay Oracle billions in 2028 don’t exist in 2028? The sheer scale of the backlog ($523B) is so large compared to current revenue ($16B) that it feels almost theoretical to some analysts. An example is the OpenAI deal that’s around $300B across 5 years from a company that’s maybe projected to hit ~$15B in ARR by the end of this year although that number is growing quickly. Where is that $300B going to come from?

The Analyst Inquisition: Inside the Q&A

The Q&A session with analysts was less of a victory lap and more of an interrogation.

The “How Will You Pay For It?” Question Analysts pressed hard on the balance sheet. Their questions were basically something akin to asking “You need to build 100 data centers. You don’t have the cash. How much debt are you going to take on?”

• The Answer: Management was defiant. They argued that the street models are wrong and that Oracle needs to raise substantially less than the $100 billion figure some were floating. They cited creative financing, noting that some customers are paying for the chips themselves, and some suppliers are leasing equipment rather than demanding cash up front. They reiterated a commitment to maintaining their investment-grade credit rating.

The “What If They Don’t Pay?” Question Another analyst asked a crucial risk question: “This infrastructure is specialized. If a big customer defaults, are you stuck with a billion-dollar paperweight?”

• The Answer: Oracle’s team insisted that their cloud is standard and fungible. If Customer A goes bust on Tuesday, they can wipe the servers and hand the capacity to Customer B by Wednesday. They claimed the demand is so high that any returned capacity would be snapped up instantly.

The “When Do We Get Paid?” Question There was significant concern about margins. Building data centers drags down profitability before the revenue starts flowing.

• The Answer: Management asked for patience. They explained that there is a lag—you spend the money on concrete and chips today, and the high-margin revenue flows tomorrow. They argued that once these AI factories are online, the margins will be in the juicy 30-40% range.

The Verdict: A Bumpy Ride to a New Oracle

So, where does this leave us?

Oracle is in the middle of a metamorphosis. They are shedding their skin as a slow-growth software utility and trying to emerge as the engine room of the AI economy. The problem is that this is a highly competitive industry that includes some of the bigger players like Microsoft, Amazon and Google but also smaller upstarts like Nebius and Coreweave.

The valuation today reflects a tug-of-war.

• The Bull Case: In five years, that $523 billion backlog converts into high-margin revenue, and Oracle becomes the most important infrastructure company in the world, eclipsing AWS and Azure in the AI niche. If that’s true then today’s valuation is a bargain and this becomes a $150B+/yr revenue juggernaut with margins that allow for further investment and free cash flows that allow them to easily pay down debt. This does mean that growth will have to spike in a year or two with the street projecting 50% growth in fiscal year 28 AND 29 and a $220B revenue year by 2030.
• The Bear Case: The CapEx crush destroys free cash flow for years, the debt load becomes unmanageable, and the AI demand cools off, leaving Oracle with empty data centers and angry bondholders.

The recent drop tells us that right now, investors are choosing to worry. But if Larry Ellison is right—and he has been right a few times in the last 40 years—this might be a great time to buy. However, you do have to buckle up; it’s going to be a bumpy ride while we wait to see if those investments pan out. For me, I’ll wait and see where this stock settles. This will be a story that will need a few years to play out which means it might also give more opportunities for investors to buy at a good price. In the 150s, this gets pretty interesting to me and that’s where I’d likely start building out a position.

Also whatever happened to that TikTok deal that Oracle was going to be involved in, is that still going to happen? I guess that’s a bonus too if it does.

Disclaimer: This article is for informational purposes only and does not constitute financial or investment advice. The author is not a registered investment advisor. All investment strategies and investments involve risk of loss. Nothing contained in this article should be construed as investment advice. Any reference to an investment’s past or potential performance is not, and should not be construed as, a recommendation or as a guarantee of any specific outcome or profit. Please consult with a professional financial advisor before making any investment decisions.