Is Oracle a Sinking Ship or a Buying Opportunity?

Oracle (NYSE:ORCL | ORCL Price Prediction) looked set to be the newest addition to the $1 trillion club just a few months back in September, where its valuation topped $877 billion. Things have only gone downhill for the stock as Wall Street began doubting the story.

ORCL stock has plummeted by over 40% from its peak after Q2 earnings. FY 2026 earnings disappointed investors. Analysts began piling concerns about how overextended data center spending was and whether it was even sustainable for Oracle to fund the growth. Investors baked in a valuation that would only make sense if Oracle managed to grow at a blistering pace for years with ever-growing margins.

But now that ORCL stock is down from its peak, should you buy the dip or jump ship? Let’s first take a look at what’s wrong here.

Why investors are dumping ORCL stock

Investors’ pet peeve with Oracle is that the funding is proving to be quite unreliable. News came out that Blue Owl Capital was withdrawing from its funding commitments, meaning that Oracle’s $10 billion Michigan data center was in jeopardy. Oracle denied this and said that the news reports were “incorrect”. Oracle’s Partner, Related Digital, said that Blue Owl did not withdraw unequivocally and that the company still had others willing to fund the project.

In any case, this has thrown sand into the gears. Q2 FY 2026 didn’t help, either.

It invited a fresh wave of criticism after Oracle reported $16.06 billion in revenue, up 14.22% year-over-year. This missed the estimates by $134.2 million. EPS did beat estimates by a wide margin, but this was due to a one-time boost from Oracle selling its Ampere stake.

The Wall Street Journal pointed to another “squishy number” that investors are paying more attention to as Oracle’s story looks more brittle. Remaining performance obligations (RPOs) have surged to $523 billion. What are these RPOs? Sales that management believes are probable. Half a trillion dollars of “probable” sales looked more or less guaranteed a few months back. But with that Blue Owl debacle and the AI buildout catching more flak, Wall Street is less willing to price that in.

Time to go bottom fishing with ORCL stock?

If those RPOs do convert to sales, the picture looks rosier than ever, but it’s still worth keeping in mind how quickly things can go wrong. For example, Oracle has a $300 billion contract with OpenAI, set to begin in 2027.

If the AI train continues full steam ahead, maybe it’s possible that OpenAI will be able to spend $300 billion on Oracle cloud compute resources over five years, as per the agreement. I don’t believe that “maybe” will translate into reality, as OpenAI does not have $300 billion.

Per Oracle, they expect to grow Cloud Infrastructure revenue “…77% to $18 billion this fiscal year—and then increase to $32 billion, $73 billion, $114 billion, and $144 billion over the subsequent four years.” The majority of Oracle’s promised revenue generation over the coming fiscal years is expected to be from OpenAI.

For this to happen, we need to assume that OpenAI has the money ready and that Oracle can finish its data centers in time. The first one is unthinkable, since OpenAI is falling behind other AI competitors and expects to be deep in the red until at least 2029. The latter argument is already crumbling, with Oracle delaying “some” data center projects for OpenAI to 2028.

What you’re actually betting on, and why I wouldn’t take the bet

Buying Oracle stock is now essentially a bet on OpenAI and its commitments. I do not believe that OpenAI will be able to buy $300 billion worth of compute over the agreement period. OpenAI won’t be on the hook and can cancel or downsize the agreement, with Oracle’s shareholders taking most of the blow.

Weather, extreme demand, and finicky data center funding will inevitably push deadlines, too.

If you do want to bet on AI and data centers, you’re better off buying Nvidia (NASDAQ:NVDA), Microsoft (NASDAQ:MSFT), or Alphabet (NASDAQ:GOOG, NASDAQ:GOOGL) instead. They aren’t as dependent on RPO figures from one startup.