Transcription
In September 2025, it looked like Larry Ellison was finally winning the AI race without even building a model of his own. His net worth was climbing so fast.
Oracle is having its biggest jump since 1992. What a blowout quarter for the software company. Maybe the stock of the year. He briefly jumped ahead of Elon Musk to become the richest person on the planet.
Mr. Ellison now has a net worth of $364 billion according to Bloomberg's Billionaires Index after the biggest one-day jump in wealth ever recorded. Oracle's stock was hitting all-time highs and the market was convinced that Larry had pulled off the ultimate pivot. But while everyone was staring at the stock charts, a massive financial trap was being set.
Oracle was pivoting from 40 years of high-profit software to become a landlord for the AI industry, and it was taking on a mountain of debt to do it. Now, with bond holders suing over an $18 billion bond offering and over $18 billion in total borrowings on the books, the bill for that ambition is finally coming due.
Oracle didn't start out as a company that worried about construction crews or electricity bills. It began in the late 1970s with a singular focus on the relational database, a piece of software that changed how businesses handled their information. Larry Ellison and his co-founders realized that data wasn't just something you stored away. It was something you needed to search and manage across an entire enterprise.
By the 1980s, Oracle had become the foundational layer of a lot of corporate computing, providing the essential software that ran everything from global bank ledgers to national government records. The company's growth was fueled by a high margin asset light model where the primary product was intellectual property. In the software world, asset light is the holy grail. It means you don't have to own a lot of physical stuff like factories, warehouses, or heavy machinery to make a fortune. You write the code once and you sell it a million times.
Once a company integrated Oracle's database into its operations, the cost and risk of moving that data to a competitor became so high that it was near impossible. It's what tech insiders call vendor lockin. and Oracle was the undisputed master of it. For decades, this created a business model where Oracle could collect massive licensing fees and recurring maintenance payments with little extra cost for each new customer. This Oracle tax was the envy of the tech world.
Larry Ellison used the massive cash flow from this database monopoly to buy up dozens of other software companies, expanding Oracle's reach into every corner of the corporate world. Even though Oracle became known as a company that was often difficult to work with, famous for aggressive sales tactics, and a reputation for being litigationheavy, its products were so essential that the world's biggest organizations couldn't function without them.
Everything started to change when cloud computing arrived. Instead of buying software and running it on their own servers, companies started renting everything from Amazon Web Services, Microsoft Azure, and Google Cloud. These companies were hyperscalers, meaning they built massive data centers and sold computing power like a utility. You didn't buy the filing cabinet anymore. You just paid for the space inside someone else's cabinet.
Oracle was slow to react to this shift and for a while it looked like the company might become a relic of the past. To survive, Oracle had to pivot. They had to stop being just a software company and start being a cloud infrastructure provider. This transition was a fundamental change in the company's financial identity.
To build a cloud, you can't just write code. You have to build physical warehouses for data. Oracle launched Oracle Cloud Infrastructure or OCI and began the incredibly expensive process of building out global data centers. They were moving from a model where they collected rent on code to a model where they were a physical landlord.
By 2024, the artificial intelligence boom gave Oracle a perfect story to sell to investors. AI models require massive amounts of specialized computing power. And Oracle positioned itself as the premier landlord for AI companies. They weren't just selling database licenses anymore. They were selling access to specialized chips called GPUs and the massive data centers needed to keep them running.
This pivot worked brilliantly for the stock price in 2025. Investors loved the idea that Oracle was the flexible alternative to the bigger clouds, willing to build customized high performance environments for AI training. By September 2025, the stock was hitting record after record, and Larry Ellison was reaping the rewards.
To put this boom in another perspective, when Oracle first went public on March 12th, 1986, the company's shares were offered at $15. If you bought a share earlier that day when they were $15 back, it would be worth roughly $61,894 by the beginning of 2026. Although, it's not quite that simple. The company's gone through so many stock splits. Basically, a way for a company to divide its existing shares into more pieces, which makes the price per share look lower, even though the value of your investment stays the same. that your one share is actually the same as owning 324 shares today. But that still means those early investors are sitting on a return of about 412,000%.
The surge in valuation was so intense that Larry Ellison's net worth briefly rivaled Elon Musk's as he climbed to the top of the world's wealth rankings in September 2025. Around this time, the company reported its fiscal first quarter 2026 results, and they looked great on the surface. Total revenue was up 12% to 14.9 billion, and cloud services revenue had jumped 28% to 7.2 billion. Saffra Katz, who was CEO at the time, told investors that as cloud services became a bigger part of the business, the company's profits would only grow faster.
As the hype continued into October 2025, Oracle doubled down. They announced a major deal with AMD to use their AI chips, a move aimed at giving customers more options and challenging Nvidia's dominance. They were also aggressively recruiting cloud workers for their new Nashville headquarters, positioning the company as a modern infrastructure powerhouse.
In September, Oracle promoted technical leaders Clay Maggor and Mike Sicilia to the roles of CEOs, while Saffricatz moved into an executive vice chair position. This was seen as a move to a leadership style that prioritized the engineering and operations needed for a massive physical buildout.
But the shift to the landlord model changed the company's risk profile in ways the market was only just starting to understand. In the old software model, Oracle didn't have to worry about construction delays or empty buildings. But in the new model, they were signing long-term leases for the space and power needed to house their cloud. This created a dynamic where Oracle took on the physical risks while the customers kept the upside of the AI models.
If the AI revolution stayed on track, the customers would be able to pay the rent. But if the market cooled, Oracle would still be on the hook for billions of dollars in specialized infrastructure that might not have a second life.
The first real signs of strain appeared in December 2025. On December 10th, Oracle released its second quarter earnings report, and the market wasn't impressed. Even though revenue was up 14% to $16.1 billion, it wasn't quite the blowout that investors expected. The stock fell 11% as the realization set in that the AI payoff wasn't happening as quickly as the spending was. And the miss contributed to Oracle's worst quarterly performance since the dot bubble burst in 2001.
The most concerning detail wasn't the revenue mess. It was a disclosure about Oracle's future lease commitments. To meet the projected demand for AI services, Oracle had increased its future operating lease commitments by almost 150%. The numbers jumped from 100 billion to 248 billion in just one quarter. These are massive off-balance sheet commitments, meaning they are future obligations Oracle has agreed to pay for data center space and power. For a company that used to be asset light, this was a massive non-cancellable bet on the long-term stability of a very young industry.
Analysts began to worry about Oracle's dependency on a very small group of AI customers, specifically OpenAI. Brent Thre, for example, pointed out that Oracle's success was now directly tied to how well Open AAI was doing financially.
The success of Oracle is going to be directly linked to how well OpenAI's financials are. If OpenAI or other major customers struggled, Oracle would be the one stuck with the physical costs of the data centers. The sentiment on Wall Street shifted from excitement about AI potential to deep concern about capital intensity and balance sheet risk. Their landlord strategy had created a lopsided risk profile.
In a normal software deal, Oracle has very little downside if a customer leaves. But in the AI landlord model, Oracle is fronting the money for someone else's experiments. If the AI models are successful, the customers get the massive upside. If they aren't, Oracle is the one with the debt and the specialized buildings. Oracle has essentially become an intermediary that is taking on the physical risk for a technology it doesn't even own.
The realization that Oracle was shouldering all the risk while its customers kept the potential upside was a major wake-up call for the market. It showed that Oracle had traded its software fortress for a position as a highly leveraged utility provider with almost no room for error. The market's focus moved away from Larry Ellison's net worth and toward the billions of dollars in commitments the company was making to stay relevant in a race it didn't fully control.
If you want to know who else is losing control, read my newsletter. I go over distressed debt, leveraged finance, bankruptcies, so you can keep on top of what's happening. Click the link in the description.
When you look at the actual balance sheet Oracle reported for the period ending November 30th, 2025, the weight of this pivot becomes clear. Oracle's total borrowings reached approximately 108.1 billion. This included 8.1 billion in short-term notes and a massive $100 billion in non-current borrowings. To manage this debt, the company had 19.2 2 billion in cash and cash equivalents on hand. While that's a lot of cash, it's a small fraction of the company's total debt load. This high level of leverage means that Oracle has very little breathing room if its cloud revenue doesn't grow fast enough to cover the interest payments.
In addition to the borrowings, the current operating lease liabilities are nearly 18.7 billion. Oracle has truly become a highly debtladen infrastructure enterprise. On the profit side, things looked better on paper. Oracle's net income for the 6 months ending November 30th, 2025 was $9 billion compared to $6.1 billion for the same period the year before. That's a sharp 48% increase in profit. But the market looked past these numbers because they don't reflect the company's capital intensity or its balance sheet risk. Investors are now focused on the fact that despite the rising profits, Oracle is struggling to convert those investments into a healthier balance sheet or improved free cash flow.
By late December 2025 and early January 2026, the growth story began to fall apart. Reports surfaced that some of the specialized data centers Oracle was building specifically for OpenAI were being delayed from their original 2027 completion dates into 2028. These delays are critical because Oracle's debt and lease payments don't stop just because construction is running late. While Oracle issued a statement saying there'd been no delays, the market's trust had been damaged.
The pressure continued in early January 2026. Two longtime board members, Naomi Seligman and George Conrads, announced they were leaving the company. This happened right as the company was facing its worst stock performance in decades and mounting questions from the debt market. Around the same time, it was revealed that investor Michael Bur, who is famous for his big shorts on the housing market and being played by Christian Bale in the movie The Big Short, had taken out an options bet against Oracle shares. When someone like Bur bets against a company with this much debt, the market takes it seriously.
The situation reached a breaking point on January 14th, 2026 when bond holders filed a lawsuit against Oracle. The lawsuit was tied to an $18 billion bond offering from September 2025, right when the AI hype was at its peak. The bond holders alleged that Oracle had failed to adequately disclose the risks and the potential for losses tied to its massive AI buildout. They claimed they had been misled about the stability and timelines of the data center projects, leading to significant drops in the value of the debt they held.
This lawsuit put the focus squarely on the $18 billion in bonds Oracle had issued to fund its transformation, and it highlighted the fundamental conflict of the pivot strategy. Management and early shareholders captured the upside when the stock hit record highs in September. But as the physical reality of the buildout set in, it was the bond holders and the company's balance sheet that were left to face the downside. The bond issuance was meant to fuel an AI revolution. But it instead became the center of a legal battle over transparency.
Oracle's move from a high margin software monopoly to a high-risk infrastructure firm is now the defining story of the company's current era. By early 2026, the company was no longer being judged by its database dominance, but by its ability to manage a 174.5 billion mountain of liabilities. The Oracle tax that had funded Larry Ellison's life for 40 years was being overwhelmed by the costs of the very technology it was trying to host.
The challenge for the company is now a matter of finance and physical reality. They've built specialized data centers that have very few other uses. They've signed $248 billion in lease commitments that they have to pay whether their customers stay or not. and they have over 108 billion in total borrowings that are now being questioned by their own lenders.
In early 2026, Oracle's leadership faces a reality that is very different from the one they sold to investors just months before. The technical challenges of building specialized data centers for OpenAI and others are being compounded by the legal challenges from the people who funded them. Due to their collapse in valuation, the stock price is still significantly below its September 2025 highs, reflecting the market's concern over the company's leverage and its ability to generate cash.
As 2026 progresses, the legal and financial pressure will likely force a degree of discipline that Oracle has largely ignored during its AI buildout. The shift from growth at all costs back to capital preservation is a difficult transition for any company. But for Oracle, it is now a matter of necessity. The billionaire founder may have won the first round of the AI trade, but the company's future now depends on whether it can survive the bills he left behind.
If you want to see more about businesses that forgot who they were and crashed because of it, click the link to my videos above.