Following Elon Musk’s shocking ” U-turn,” billionaire Larry Ellison immediately poured $300 billion into OpenAI. Is this the revenge of the century or a misstep that will bring down the Oracle empire?

According to The Verge, if you want to know whether the artificial intelligence (AI) craze is truly sustainable or just a bubble about to burst, look at Oracle. Not OpenAI or Anthropic, but this long-standing database giant is the most accurate measure.

Oracle has poured all its resources into transforming itself into AI, but with a strategy that is completely different from the rest of Silicon Valley. They are not directly building large language models (LLMs) or simply providing cloud services.

This is a software-as-a-service (SaaS) company that is betting its entire future on a specific AI scenario. Meanwhile, its core traditional business is gradually declining over time.

Larry Ellison and the mindset of a competitive “prophet”.

Larry Ellison, the founder of Oracle and nicknamed “Bad Doggy,” is one of the most eccentric figures in the tech world. Even in his 80s, he still possesses a huge ego and a competitive spirit capable of powering any data center.

Ellison is known for making bold promises that leave engineers racing against time. In 1996, he promoted the “network computer”—a compact device that ran online applications, the precursor to cloud computing.

Despite the correct vision, the product failed miserably. This made Ellison stubborn, even mocking “cloud” as meaningless in 2011, leading to the loss of market leadership to Amazon and Microsoft.

Now, Ellison doesn’t want to repeat that mistake. By February 2025, he told former British Prime Minister Tony Blair: “AI is far greater than the Industrial Revolution, electricity, and everything before it combined.”

Oracle is abandoning its high-profit, slow-growing database business to jump into cloud infrastructure. Investor Paul Kedrosky commented, “The traditional, slow-growing, high-profit company – it makes him feel old and outdated.”

Venture alliance with OpenAI: “All or nothing.”

The OpenAI deal with Oracle came about like a twist of fate – after Elon Musk unexpectedly changed his mind about the xAI project. This move was extremely costly: Oracle borrowed up to $43 billion in fiscal year 2026, and spent $39 billion on capital expenditures, causing its free cash flow to become negative.

Of Oracle’s remaining $553 billion in contractual obligations, over $300 billion belongs to OpenAI. Significantly, Oracle doesn’t build its own data centers but instead hires companies like Crusoe to do it for it, often on risky 15-year long-term contracts.

This alignment caused Oracle to lose its autonomy. Sam Altman acknowledged the pressure: “I expect the atmosphere out there to be uncomfortable for a while.” The glamorous Stargate project was essentially shelved as key leaders migrated to Meta.

OpenAI projects spending up to $665 billion by 2030, despite having just raised $122 billion. CFO Sarah Friar has warned internally about the possibility of defaulting on computing contracts if revenue doesn’t grow fast enough.

A comparison of valuations reveals the level of risk: OpenAI is valued at 28 times its 2026 revenue, while Nvidia – the company that actually makes a profit – is only 12 times. Oracle is betting on a partner with an extremely precarious financial foundation.

Risks abound: From supply chains to geopolitics

Oracle’s journey to build its AI infrastructure is facing numerous practical obstacles. Currently, 11 US states are considering temporary bans on data centers due to fierce public opposition regarding energy consumption.

The “Project Jupiter” project in New Mexico is accused of emitting more greenhouse gases than both cities combined. In Wisconsin, round-the-clock construction has caused such outrage that protests have erupted demanding the project be cancelled.

Geopolitical tensions are adding fuel to the fire. The blockade of the Strait of Hormuz threatens the supply of helium—an essential component for chips. Soaring aluminum prices are significantly driving up the cost of server chassis and cooling systems.

The bond market reacted with considerable unease. John Lloyd of Janus Henderson Investors commented, “Oracle’s CDS has become a barometer of AI risk.” This risk is further amplified by OpenAI’s diversified operations, while Anthropic, with Claude Code, is gaining ground in the enterprise market.

Vision of a “Surveillance Empire” and Data Power

Ellison believes in “AI inference” on private data. Oracle holds sensitive data from the US Air Force, medical records, and has a $2 billion stake in TikTok US, as well as being ByteDance’s primary storage partner.

However, Oracle’s reputation regarding data is far from clean. In 2024, the company settled a class-action lawsuit alleging the unauthorized collection of data to compile “digital profiles” of users and sell them to advertisers.

Ellison’s vision is to consolidate all national data into a single database. He suggests AI will help the government monitor citizens 24/7. He even describes police continuously recording footage even when citizens request their cameras be turned off.

Oracle’s most formidable core capability is its ability to “lock down customers.” Once data is within the Oracle ecosystem, moving it is virtually impossible, forcing customers to remain loyal to their inference services.

Oracle is currently in its most intense phase of implementation. They are racing to turn theoretical numbers into actual data centers before funding runs out or their partner OpenAI collapses under financial pressure.

Oracle stock is no longer just a traditional software stock; it’s a risky derivative gamble. If Larry Ellison is right, he will dominate the surveillance era. If he’s wrong, it will be proof of the collapse of a bubble.