© Kimberly White / Getty Images News via Getty Images
Oracle (NYSE:ORCL) looked like it was destined to be the next $1 trillion company last year before things made an abrupt turn. The stock has nearly halved since then, with the market cap now below $470 billion. This is despite the company being knee-deep in the burgeoning AI data center race and having an oversized influence over the U.S. government.
Larry Ellison went on the podium next to Trump last year and recently inked a deal with ByteDance, the parent company of TikTok, to seize partial control of the app’s U.S. arm. The company is pouring capital into AI-centric data centers, something that the market has rewarded the company for just a few months ago.
So, why did the stock suddenly make a turn for the worse, and is it now time to buy the dip? We’ll have to look a bit deeper.
Investors are no longer blind about AI
AI’s novelty is wearing off, and Wall Street is no longer writing a blank check to any company that claims it is involved in the industry. Oracle may be involved in data centers, but it is only influential in a narrower and more specialized way than the hype implies. It is not a top‑tier general‑purpose cloud like AWS, Azure, or Google Cloud.
Where it is making progress is building out data centers. Oracle’s “AI” is more of a bet on those data centers, and there’s a lot that can go wrong. The company does not have the capital on hand, so it is heavily reliant on taking on debt to build out this infrastructure.
That’s a catch many investors are not willing to take. You can take a look at CoreWeave (NASDAQ:CRWV) as an excellent example. CRWV stock has nearly halved from its peak due to concerns about whether or not its buildout is sustainable, let alone even profitable in the end.
Oracle is executing one of the most aggressive AI infrastructure buildouts that dwarfs even CoreWeave’s numbers. It raised its fiscal 2026 (ending May 2026) capital expenditure guidance to $50 billion, up from the $35 billion forecast just three months earlier in September 2025. The company spent $12 billion in Q2 FY 2026 alone, and some analysts project actual spending could reach $60 billion or more by fiscal 2028.
Why investors are selling ORCL stock
Oracle’s total debt stood at approximately $108.1 billion as of Q2 FY 2026, plus another $24 billion in lease liabilities already on the balance sheet. Oracle raised $38 billion in term loans in October 2025 (led by JPMorgan) for two Vantage-built data centers in Wisconsin and Texas, plus another $18 billion in investment-grade bond financing for a New Mexico data center.
The bigger surprise is off-balance-sheet commitments: Oracle added $148 billion in new lease commitments in Q3 alone, bringing total future lease obligations to $248 billion, with some leases extending up to 19 years. These commitments function economically as long-term fixed liabilities once they commence, pushing Oracle’s total debt and obligations to roughly $380 billion when combined.
Sure, the company’s Remaining Performance Obligations (RPO) reached $523 billion, but most of that is just OpenAI.
Why I won’t touch ORCL stock
The stock may have halved, but from a very high base. ORCL stock remains overvalued, and it will need to decline a lot more before taking on the risk makes sense, since much of that debt rests largely on the success of OpenAI. Oracle needs to backlog into actual revenue fast enough to service its debt and lease obligations.
But can it do that? Likely not.
The OpenAI contract constitutes $300 billion of the entire backlog. The problem is that OpenAI is targeting ~$13 billion in total revenue for 2025, meaning it would need to dedicate multiple times its entire current revenue just to pay Oracle. CEO Sam Altman projects revenue could reach $100 billion by 2027. Those projections assume explosive growth that must materialize almost immediately to avoid a cash crisis. It’s not happening unless OpenAI somehow trounces all the competition.
HSBC analysts calculated that OpenAI faces a potential funding shortfall of $207 billion by 2030. Thus, I believe there’s very little leeway for OpenAI to meet its commitments here.
Only in the best-case scenario do I see meaningful returns with ORCL stock, but the risk-reward ratio is not in your favor.