What Everyone Had Been Waiting for and Fearing: Oracle Just Triggered the Implosion of the AI Bubble
You have no idea how big the AI bubble is. The amount of resources poured into it is simply unimaginable. Researchers say people start losing their intuitive grip on numbers somewhere around a million. So, what chance do we have with billions — let alone trillions?
Okay, let’s compare it to something “normal.” Take our Sun. Now imagine its size equals the size of the economy of, say, one of the Baltic states. On that scale, the AI bubble is a star hundreds of times bigger.
If the Sun were that big, our planet would be a blazing hell.
And now this massive star is about to blow.
The AI bubble will go the same way such a star does — and for the same reason.
The forces are different, but the mechanism is the same: it runs out of the fuel that keeps the monster from collapsing in on itself under its own insanely powerful gravity.
An imploding star crushes itself into a tiny ball, and the rebound instantly triggers a colossal explosion that lights up the entire galaxy. A bursting AI bubble threatens the entire American economy.
The gravity inside the bubble is investor expectations. So far, those expectations have been held in check by promises of massive profits to come.
It’s that promise investors are putting their capital behind. That money builds hyperscale data centers and buys the hardware.
AI companies keep telling the public that this infrastructure will let them ship ever more powerful models. Their AI will be able to handle more and more tasks, and its business value will grow fast.
That’s the basic setup that still keeps the forces in balance. But it can’t go on forever. Sooner or later, promises have to turn into real revenue. Otherwise, implosion is inevitable.
What’s Happening to the Stabilizing Layer
Unfortunately for OpenAI, Anthropic, and the rest of the AI crowd, so-called “AI” can’t deliver that revenue — for objective reasons.
First, their margins. They’re nowhere near enough to cover the cost of building models, running operations, and paying back investment commitments.
Building a model costs hundreds of millions of dollars. Fine-tuning can run tens of millions more, and developer salaries eat up millions of dollars a year. And so on.
And then there’s the massive compute — the single biggest line item in AI labs’ obligations.
Second, as it turns out, businesses aren’t exactly eager to switch from older models to newer ones. The old models cover most of their needs, and switching is too complex and too risky to justify.
Third, there’s growing competition from open-weight models. The problem is obvious: why pay for expensive proprietary models when you can get by with much cheaper ones? This way, you control not just what you spend on the model but the model itself. You can run it on your own server, and nobody on “the other side” of the API can change a thing.
As a result, the stabilizing layer of the AI bubble keeps getting thinner.
Up to a point, that thinning is tolerable. But there comes a moment when everything changes — fast.
The First Domino
And now the balance has broken. The implosion was set off by Oracle.
Oracle is one-third of the infamous Stargate triumvirate — the ultimate symbol of the AI bubble. The project was unveiled at the White House in January 2025, and the roles were clear-cut: OpenAI is the customer that consumes the compute; SoftBank is the money partner; Oracle is the builder and tenant, turning desert, concrete, and Nvidia chips into working data centers.
In September 2025, Oracle and OpenAI signed a five-year, $300 billion contract, and Larry Ellison’s company’s stock shot into the stratosphere.
But now Oracle is in trouble. One reason is its heavy dependence on OpenAI, which currently accounts for roughly half of the company’s contracted future revenue ($638 billion).
In other words, a big chunk of Oracle’s future revenue is a promise from one single customer. A customer that doesn’t make money itself and lives off investors’ cash.
Capex Today, Profits Someday
Oracle has to build and equip the data centers and sign multi-year leases — all years before the first dollar from OpenAI shows up.
It doesn’t have the cash for that. In July, Oracle raised its fiscal 2027 capex forecast from $60 billion to $90–95 billion. S&P redid its math: this year’s free cash flow shortfall will hit nearly $42 billion (up from the $24 billion it expected before).
The gap is being filled with debt. The company’s loans and bonds have already reached $125 billion.
On July 9, 2026, S&P cut Oracle’s credit rating to BBB-. That’s one notch above junk. But that’s just the tip of the iceberg.
Phantom Debt
Under accounting rules, a lease on a data center that isn’t up and running yet doesn’t show up on the balance sheet as a liability. It lives in the footnotes.
For Oracle, those “footnotes” have already piled up to $288 billion (1, 2). That’s more than twice its entire official debt.
It’s worth pausing here to see how this works. Oracle doesn’t own these buildings. Developers build them through special purpose vehicles (SPVs) that borrow against a specific project. And what are the banks lending against?
Oracle’s lease. A building is worth exactly what a tenant is willing to pay for it. Oracle’s promise to pay rent is essentially the only guarantee holding the whole thing together.
Project Jupiter
Now let’s get to where the implosion actually starts.
Project Jupiter is Stargate’s flagship campus. Yet another monster growing inside the AI bubble.
Its capacity is about 2.5 GW, with planned investment of $165 billion. It’s owned by Stack Infrastructure (part of Blue Owl Capital) together with BorderPlex Digital Assets.
Construction is financed by an $18 billion loan from a syndicate of about 20 banks, including Sumitomo Mitsui, BNP Paribas, Goldman Sachs, and MUFG. And remember: Oracle is the anchor tenant, reselling that capacity to OpenAI.
For this whole layered scheme to work, the campus needs as much power as a major city — think Los Angeles. Hooking into the existing grid turned out to be a problem. The workaround: generate power on site with fuel cells running on natural gas.
The gas is supposed to come through a new pipeline spur that still has to be built. Most of the route crosses federal land, but about half a mile runs across state land. And that’s exactly where everything got stuck: on March 31, 2026, the New Mexico State Land Office denied the right-of-way, and on July 16 it rejected the second application. The pipeline, originally due to go online in August 2026, has been pushed back to February 2027.
On top of that, the air permit for the entire power system still hasn’t been issued: the state Environment Department’s decision isn’t expected before November 23.
Meanwhile, local opposition keeps gaining ground — as does the nationwide movement against hyperscale data centers (I wrote about it earlier here).
Force Majeure
And then, on September 24, 2026, Oracle sent Blue Owl a force majeure notice.
In it, Oracle states that it reserves the right to delay switching to full rent by up to three years if the campus isn’t up and running by 2028. Instead, it would keep paying the lower “development-stage” rate.
Force majeure is a legal move with real consequences for the other side. Here, Oracle is admitting that its flagship project may not make its deadline.
The fallout could be dramatic. Every previous warning about the AI bubble came from analysts, short sellers, and critics. This is the first one coming from inside the chain itself.
And the market reacted accordingly:
investors started dumping Oracle’s 30-year bonds (maturing in 2056), and their price dropped;
the cost of insuring against the company’s default (CDS) hit a record high.
But Oracle was already taking hits before that:
its stock had fallen roughly 60% from its September 2025 peak;
the $18 billion loan backing Jupiter was already trading at 89–91 cents on the dollar. In other words, lenders had started pricing in the chance they won’t get all their money back.
Larry’s Flip-Flop
On June 22, 2026, Larry Ellison adopted a plan to sell up to 50 million shares — about $7.5 billion — by October 24. Shareholders only found out on September 11, from the quarterly report. The very next day, a Saturday, Oracle rushed to announce that Ellison had backed out and “has no other plans to sell any of his Oracle stock.”
If this sequence of events strikes you as odd, you’re not alone. For almost three months, the most important man at the company was ready to sell billions of dollars’ worth of stock — and the people Oracle spent those same months persuading to bet on its AI strategy had no idea.
Twelve days later came the force majeure notice. And one day after that, on September 25, Oracle’s proxy statement revealed that Ellison had pledged 413 million shares as collateral for personal loans. That’s 67 million more than a year earlier (!).
And that, mind you, is roughly a third of Ellison’s own stake. Also note: since 2018, Oracle has banned pledging shares for all directors and employees. With one exception — Larry Ellison.
The board insists there’s no risk: these are term loans, not margin accounts, and Ellison has more than enough money to pay them off without selling a single share. What they didn’t spell out is that any loan backed by stock depends on the price of that stock.
And, as I already pointed out: 1) those shares have lost about 60% of their value in a year; 2) the company’s credit rating sits barely above junk.
A Stack of Promises
Now let’s look at the whole structure from the bottom up.
The banks get paid if the developer collects the rent.
The developer collects the rent if Oracle pays.
Oracle pays if OpenAI pays.
OpenAI pays if investors keep writing checks.
(If you skimmed the section “What’s Happening to the Stabilizing Layer,” now’s a good time to go back to it.)
The point is, all these “ifs” don’t add up to a pyramid of assets. It’s a stack of promises, each one propping up the next. And none of them starts paying off until the campus is up and running. And it’ll be up and running once a gas pipe gets laid across New Mexico state land. And if it doesn’t?
That’s what the stabilizing layer of the AI bubble looks like in real life: not a superintelligence granting us godlike powers, but a promise backed by a promise backed by a gas pipeline that doesn’t exist yet — and may never exist.
Oracle just happened to be the first to land on the spot where that layer is thinnest.
Conclusion
In a physics thought experiment, all it takes is tossing a pinch of iron into the furnace of a massive star teetering on the edge of explosion to trigger an instant implosion.
Oracle did just that. The problem with the AI bubble is precisely that its scale doesn’t match the reliability of the system it’s inflating around.
→ There’s no reliable technology that lives up to the hopes pinned on it.
→ There are no rational thinkers running the AI companies.
→ There are no competent (or honest?) media among the most influential ones, capable of telling the public what’s really going on.
What there is: hype beyond every conceivable limit and, as a result, a monstrous bubble devouring an endless amount of resources.
Yes, there are some experts who call things by their real names. But their views are unpopular in an atmosphere shaped by greed, incompetence, and shameless exploitation of the general public’s ignorance.
Oracle is a mirror of where we are. Those who have eyes, let them see: this is exactly how a collapse begins.
First, construction gets delayed.
Then contracts get renegotiated.
Lenders start demanding higher yields.
Investors want out.
Companies cut spending and cancel new projects.
And that, in turn, shrinks the future revenue of those who just sank tens and hundreds of billions of dollars into AI.
And the chain starts collapsing in on itself.
It all plays out just like the life of a massive star that can no longer hold back the inward pull of its own gravity. And that moment comes when it shines brighter than ever before.
It’s a sign. From a distance, it may look like proof of its invincible power. But deep inside, the fuel that holds it together is running out. The star collapses and, in a furious explosion, destroys everything around it within tens of light-years.
The AI industry has come right up to that line. The implosion will begin when the money starts flowing away from the center of the bubble. And that could happen very soon.
All we can do is hope (and some of us, perhaps, pray) for a miracle that lets us avoid a catastrophe on a planetary scale.
Okay, let’s compare it to something “normal.” Take our Sun. Now imagine its size equals the size of the economy of, say, one of the Baltic states. On that scale, the AI bubble is a star hundreds of times bigger.
If the Sun were that big, our planet would be a blazing hell.
And now this massive star is about to blow.
The AI bubble will go the same way such a star does — and for the same reason.
The forces are different, but the mechanism is the same: it runs out of the fuel that keeps the monster from collapsing in on itself under its own insanely powerful gravity.
An imploding star crushes itself into a tiny ball, and the rebound instantly triggers a colossal explosion that lights up the entire galaxy. A bursting AI bubble threatens the entire American economy.
The gravity inside the bubble is investor expectations. So far, those expectations have been held in check by promises of massive profits to come.
It’s that promise investors are putting their capital behind. That money builds hyperscale data centers and buys the hardware.
AI companies keep telling the public that this infrastructure will let them ship ever more powerful models. Their AI will be able to handle more and more tasks, and its business value will grow fast.
That’s the basic setup that still keeps the forces in balance. But it can’t go on forever. Sooner or later, promises have to turn into real revenue. Otherwise, implosion is inevitable.
What’s Happening to the Stabilizing Layer
Unfortunately for OpenAI, Anthropic, and the rest of the AI crowd, so-called “AI” can’t deliver that revenue — for objective reasons.
First, their margins. They’re nowhere near enough to cover the cost of building models, running operations, and paying back investment commitments.
Building a model costs hundreds of millions of dollars. Fine-tuning can run tens of millions more, and developer salaries eat up millions of dollars a year. And so on.
And then there’s the massive compute — the single biggest line item in AI labs’ obligations.
Second, as it turns out, businesses aren’t exactly eager to switch from older models to newer ones. The old models cover most of their needs, and switching is too complex and too risky to justify.
Third, there’s growing competition from open-weight models. The problem is obvious: why pay for expensive proprietary models when you can get by with much cheaper ones? This way, you control not just what you spend on the model but the model itself. You can run it on your own server, and nobody on “the other side” of the API can change a thing.
As a result, the stabilizing layer of the AI bubble keeps getting thinner.
Up to a point, that thinning is tolerable. But there comes a moment when everything changes — fast.
The First Domino
And now the balance has broken. The implosion was set off by Oracle.
Oracle is one-third of the infamous Stargate triumvirate — the ultimate symbol of the AI bubble. The project was unveiled at the White House in January 2025, and the roles were clear-cut: OpenAI is the customer that consumes the compute; SoftBank is the money partner; Oracle is the builder and tenant, turning desert, concrete, and Nvidia chips into working data centers.
In September 2025, Oracle and OpenAI signed a five-year, $300 billion contract, and Larry Ellison’s company’s stock shot into the stratosphere.
But now Oracle is in trouble. One reason is its heavy dependence on OpenAI, which currently accounts for roughly half of the company’s contracted future revenue ($638 billion).
In other words, a big chunk of Oracle’s future revenue is a promise from one single customer. A customer that doesn’t make money itself and lives off investors’ cash.
Capex Today, Profits Someday
Oracle has to build and equip the data centers and sign multi-year leases — all years before the first dollar from OpenAI shows up.
It doesn’t have the cash for that. In July, Oracle raised its fiscal 2027 capex forecast from $60 billion to $90–95 billion. S&P redid its math: this year’s free cash flow shortfall will hit nearly $42 billion (up from the $24 billion it expected before).
The gap is being filled with debt. The company’s loans and bonds have already reached $125 billion.
On July 9, 2026, S&P cut Oracle’s credit rating to BBB-. That’s one notch above junk. But that’s just the tip of the iceberg.
Phantom Debt
Under accounting rules, a lease on a data center that isn’t up and running yet doesn’t show up on the balance sheet as a liability. It lives in the footnotes.
For Oracle, those “footnotes” have already piled up to $288 billion (1, 2). That’s more than twice its entire official debt.
It’s worth pausing here to see how this works. Oracle doesn’t own these buildings. Developers build them through special purpose vehicles (SPVs) that borrow against a specific project. And what are the banks lending against?
Oracle’s lease. A building is worth exactly what a tenant is willing to pay for it. Oracle’s promise to pay rent is essentially the only guarantee holding the whole thing together.
Project Jupiter
Now let’s get to where the implosion actually starts.
Project Jupiter is Stargate’s flagship campus. Yet another monster growing inside the AI bubble.
Its capacity is about 2.5 GW, with planned investment of $165 billion. It’s owned by Stack Infrastructure (part of Blue Owl Capital) together with BorderPlex Digital Assets.
Construction is financed by an $18 billion loan from a syndicate of about 20 banks, including Sumitomo Mitsui, BNP Paribas, Goldman Sachs, and MUFG. And remember: Oracle is the anchor tenant, reselling that capacity to OpenAI.
For this whole layered scheme to work, the campus needs as much power as a major city — think Los Angeles. Hooking into the existing grid turned out to be a problem. The workaround: generate power on site with fuel cells running on natural gas.
The gas is supposed to come through a new pipeline spur that still has to be built. Most of the route crosses federal land, but about half a mile runs across state land. And that’s exactly where everything got stuck: on March 31, 2026, the New Mexico State Land Office denied the right-of-way, and on July 16 it rejected the second application. The pipeline, originally due to go online in August 2026, has been pushed back to February 2027.
On top of that, the air permit for the entire power system still hasn’t been issued: the state Environment Department’s decision isn’t expected before November 23.
Meanwhile, local opposition keeps gaining ground — as does the nationwide movement against hyperscale data centers (I wrote about it earlier here).
Force Majeure
And then, on September 24, 2026, Oracle sent Blue Owl a force majeure notice.
In it, Oracle states that it reserves the right to delay switching to full rent by up to three years if the campus isn’t up and running by 2028. Instead, it would keep paying the lower “development-stage” rate.
Force majeure is a legal move with real consequences for the other side. Here, Oracle is admitting that its flagship project may not make its deadline.
The fallout could be dramatic. Every previous warning about the AI bubble came from analysts, short sellers, and critics. This is the first one coming from inside the chain itself.
And the market reacted accordingly:
investors started dumping Oracle’s 30-year bonds (maturing in 2056), and their price dropped;
the cost of insuring against the company’s default (CDS) hit a record high.
But Oracle was already taking hits before that:
its stock had fallen roughly 60% from its September 2025 peak;
the $18 billion loan backing Jupiter was already trading at 89–91 cents on the dollar. In other words, lenders had started pricing in the chance they won’t get all their money back.
Larry’s Flip-Flop
On June 22, 2026, Larry Ellison adopted a plan to sell up to 50 million shares — about $7.5 billion — by October 24. Shareholders only found out on September 11, from the quarterly report. The very next day, a Saturday, Oracle rushed to announce that Ellison had backed out and “has no other plans to sell any of his Oracle stock.”
If this sequence of events strikes you as odd, you’re not alone. For almost three months, the most important man at the company was ready to sell billions of dollars’ worth of stock — and the people Oracle spent those same months persuading to bet on its AI strategy had no idea.
Twelve days later came the force majeure notice. And one day after that, on September 25, Oracle’s proxy statement revealed that Ellison had pledged 413 million shares as collateral for personal loans. That’s 67 million more than a year earlier (!).
And that, mind you, is roughly a third of Ellison’s own stake. Also note: since 2018, Oracle has banned pledging shares for all directors and employees. With one exception — Larry Ellison.
The board insists there’s no risk: these are term loans, not margin accounts, and Ellison has more than enough money to pay them off without selling a single share. What they didn’t spell out is that any loan backed by stock depends on the price of that stock.
And, as I already pointed out: 1) those shares have lost about 60% of their value in a year; 2) the company’s credit rating sits barely above junk.
A Stack of Promises
Now let’s look at the whole structure from the bottom up.
The banks get paid if the developer collects the rent.
The developer collects the rent if Oracle pays.
Oracle pays if OpenAI pays.
OpenAI pays if investors keep writing checks.
(If you skimmed the section “What’s Happening to the Stabilizing Layer,” now’s a good time to go back to it.)
The point is, all these “ifs” don’t add up to a pyramid of assets. It’s a stack of promises, each one propping up the next. And none of them starts paying off until the campus is up and running. And it’ll be up and running once a gas pipe gets laid across New Mexico state land. And if it doesn’t?
That’s what the stabilizing layer of the AI bubble looks like in real life: not a superintelligence granting us godlike powers, but a promise backed by a promise backed by a gas pipeline that doesn’t exist yet — and may never exist.
Oracle just happened to be the first to land on the spot where that layer is thinnest.
Conclusion
In a physics thought experiment, all it takes is tossing a pinch of iron into the furnace of a massive star teetering on the edge of explosion to trigger an instant implosion.
Oracle did just that. The problem with the AI bubble is precisely that its scale doesn’t match the reliability of the system it’s inflating around.
→ There’s no reliable technology that lives up to the hopes pinned on it.
→ There are no rational thinkers running the AI companies.
→ There are no competent (or honest?) media among the most influential ones, capable of telling the public what’s really going on.
What there is: hype beyond every conceivable limit and, as a result, a monstrous bubble devouring an endless amount of resources.
Yes, there are some experts who call things by their real names. But their views are unpopular in an atmosphere shaped by greed, incompetence, and shameless exploitation of the general public’s ignorance.
Oracle is a mirror of where we are. Those who have eyes, let them see: this is exactly how a collapse begins.
First, construction gets delayed.
Then contracts get renegotiated.
Lenders start demanding higher yields.
Investors want out.
Companies cut spending and cancel new projects.
And that, in turn, shrinks the future revenue of those who just sank tens and hundreds of billions of dollars into AI.
And the chain starts collapsing in on itself.
It all plays out just like the life of a massive star that can no longer hold back the inward pull of its own gravity. And that moment comes when it shines brighter than ever before.
It’s a sign. From a distance, it may look like proof of its invincible power. But deep inside, the fuel that holds it together is running out. The star collapses and, in a furious explosion, destroys everything around it within tens of light-years.
The AI industry has come right up to that line. The implosion will begin when the money starts flowing away from the center of the bubble. And that could happen very soon.
All we can do is hope (and some of us, perhaps, pray) for a miracle that lets us avoid a catastrophe on a planetary scale.
