Summary of Key Points
Larry Ellison, the 82-year-old founder of Oracle, could have enjoyed his later years with stable businesses such as databases and enterprise software (generating billions in annual revenue and substantial profits). However, after the GPT revolution sparked the AI boom, he decided to bet the company's future on AI computing power. He invested heavily in building data centers and purchasing GPUs, signing multi-billion-dollar contracts for the future. This required a significant upfront cash outlay, which led to negative cash flow, soaring debt, a near-junk credit rating, and even layoffs and the downsizing of traditional businesses. This is his final battle: success would turn Oracle into a giant in AI infrastructure; failure could result in massive debts and idle data centers, tarnishing his reputation for decades.
1. Why Would an 82-Year-Old Man Give Up a Comfortable Retirement to Go All In on AI?
Ellison's life was once that of a "island lord"—he purchased the 360-square-kilometer island of Lanai, where over 3,000 people called him "boss," and he largely avoided new technological concepts after retiring as CEO. But the emergence of GPT changed everything. In June 2023, he described AI at a company meeting as "the first time humans have used fire," calling it a revolution even more significant than the internet. The reasons for his bet are threefold:
- Personality: He has always embraced high risks; in 1999, he invested in the internet, predicting that if it wasn't the future, it would be a disaster, and he indeed became the richest man.
- Determination: As a legend of the database era, he didn't want to be marginalized in the AI era.
- Opportunity: The demand for computing power in AI is exploding, and he believes that customers (like OpenAI) will eventually have the funds to pay, and Oracle will be able to withstand the challenge.
2. How Much Has Oracle Invested in AI?
Ellison's "all-in" strategy wasn't just talk; it involved substantial financial commitments:
- Future Contracts vs. Immediate Expenses: Oracle holds $638 billion in future contracts (with OpenAI accounting for $300 billion, to be fulfilled starting in 2027), but it must spend now on building data centers and purchasing GPUs. In fiscal year 2026, capital expenditures (for building data centers and equipment) reached $55.7 billion, 83% of its annual revenue—more than three times its operating cash flow ($32 billion).
- Soaring Debt: Interest-bearing debt increased from $92.6 billion to $129.5 billion (an increase of $37 billion in just over a year), with plans to raise another $40 billion. There are also long-term leases for $26 billion in data center rental agreements (ranging from 15 to 19 years); even if AI fails, the debt and rent payments must be made.
- Negative Cash Flow and Downgraded Credit Rating: Oracle has a negative free cash flow of $23.7 billion, and Standard & Poor's downgraded its credit rating to BBB-, just one step below junk status, making borrowing more expensive. Some institutions may refuse to buy its bonds.
- Layoffs and Business Downsizing: The number of employees was reduced from 162,000 to 141,000 (a 13% cut), with restructuring costs increasing fivefold. This isn't because traditional businesses are failing, but because all funds were invested in AI.
3. Could This Gamble Fail? What Are the Risks?
The potential "fiasco" doesn't refer to a drop in stock prices but rather the failure of contract fulfillment, leading to asset impairment, a junk rating, forced large-scale share issuance, or the need to stop building data centers. The risks stem from three major mismatches:
- Out-of-Sync Expenses and Revenue: Debt interest and rent payments are monthly obligations, while payment from OpenAI won't start until 2027.
- GPU Obsolescence and Long Leases: GPUs are updated frequently (within a few years), but lease agreements are for over a decade; if they become obsolete before the leases expire, the investment will be wasted.
- Customer Risk Concentration: OpenAI accounts for nearly half of Oracle's future contracts. If it struggles with financing or delays payments, Oracle's data centers will remain idle, increasing cash shortages and further damaging its rating.
However, there's no need to panic too much: Oracle still has stable traditional businesses (generating $17 billion annually), and customers have prepaid for GPUs, so the likelihood of a complete collapse is low. Nevertheless, there's a significant risk of partial failure.
4. If Oracle Fails, Will the AI Industry Follow?
A collapse would have cascading effects:
- Oracle: It would cut expenses, stop building data centers, experience asset impairment, and possibly issue additional shares.
- OpenAI: It would lose its current computing power sources and need to find alternatives (such as Microsoft or Amazon), potentially incurring additional costs.
- Industry Financing: Banks would re-evaluate the risks of AI data centers, increasing borrowing costs for other projects.
- Upstream and Downstream Companies: Chip manufacturers might experience order delays, and energy/c cooling projects could shrink. The capital market would reconsider large orders from all AI companies.
However, AI technology itself won't disappear; Oracle's data centers will be taken over by wealthy firms like Microsoft and Amazon, shifting the industry from widespread expansion to consolidation. This is essentially a test of whether AI can generate enough revenue to cover these investments. Time will be crucial in determining the outcome.
5. Ellison: A Natural Gambler in His Final Battle
This isn't Ellison's first gamble; he invested in the internet in 1999 and became the richest man in 2000. At 82, betting on AI is his final opportunity:
- Success: He could transform Oracle from a mid-tier cloud provider into an AI infrastructure giant, creating another legend in his old age.
- Failure: Massive debts and idle data centers would mar his reputation, but perhaps he doesn't care much—after all, he could have retired comfortably yet chose to pursue his passions until the end.
(Aside: At 73, he fell in love with a 27-year-old woman named Zhu Keren from Northeast China; some say this gave him extra motivation. Regardless of the truth, this man is indeed more adventurous than many young people.)
In Conclusion
Ellison is betting his life's worth—his wealth, reputation, and credibility—on the future of AI computing power. This gamble determines not only Oracle's fate but also whether the AI industry can become profitable. After all, the money spent now must be earned back through AI in the future. (As of September 2025, Oracle's stock price briefly made him the world's richest man again, but it has since halved. The gamble continues.)