This isn't the article I thought I'd be posting today.
The dotcom bubble wiped out $5 trillion. What we are sitting in right now is mathematically three times larger.
The AI bubble is nothing like we've seen before.
That might sound obvious, but the scary part is that this isn't a bubble in the way most people think about bubbles.
Bubbles are obvious in hindsight. They inflate, they pop, and everyone claims they saw it coming.
This time is different, and I'm about to explain why.
In today's article, I'm going to write about what no other AI creator is willing to.
And that is, the risks and headwinds with the current state of AI.
Sit down, take notes, and be prepared for slight doomerism - sorry, I have to warn you.
How the "Dam" Was Built
For the rest of this article, I'll be referring to the AI bubble as a "dam." It's the clearest way I can explain what's actually happening here.
To understand why this bubble is different, you need to understand how the dam was actually constructed.
Unlike the dotcom era, when companies just burned cash on marketing and went bankrupt, this bubble has a structure. One that makes everyone's revenue look real.
It is called circular financing (you've probably seen this viral chart on X before):
Circular Financing
Here's a look at how this actually works:
NVIDIA invests in OpenAI. OpenAI takes that cash and buys cloud computing from Microsoft and Oracle. Microsoft and Oracle use that revenue to buy more Nvidia chips. NVIDIA books those sales as growth, their stock goes up, and they use the higher valuation to invest more in OpenAI.
The same dollar moves in a circle; every company records it as revenue. Every balance sheet looks like it is growing.
But here's the key:
No new value was actually created.
And this is just the tip of the circular financing iceberg.
These companies are taking on debt to buy equipment from Nvidia, and Nvidia is investing that cash in startups that then buy services from companies like Meta and Google.
Circular Financing in Action
This is one of the reasons AI now accounts for nearly 75% of the S&P 500's total gains.
Now, to address a counterargument:
"This kind of arrangement exists in other industries. Suppliers invest in customers all the time."
My rebuttal: The scale here is unprecedented. We are talking about hundreds of billions of dollars flowing through Fortune 500 companies.
The same dollar is being counted several times.
That is the dam - and THAT is why the AI bubble is different from anything we've ever seen before.
Hundreds of billions in circular financing, holding back the pressure of 17x more revenue than what actually exists. As long as everyone keeps building and reinforcing the structure, it holds.
But what happens when someone stops and something cracks?
Where the Cracks Are Forming
For two years, the dam held because everyone had the same incentive.
NVIDIA needed OpenAI to buy chips. OpenAI needed Microsoft for cloud credits. Microsoft needed OpenAI to justify its AI investment.
Every company needed every other company to keep spending, keep building, and keep reinforcing the wall. Nobody was going to walk away; the system was self-sustaining.
That changed in the last few weeks. And it happened in three places at once.
Crack #1: Nvidia
Jensen Huang just told reporters in Taipei that Nvidia's hundred-billion-dollar commitment to OpenAI was never actually a commitment (his words, not mine). This is the CEO of the company that supplies 90% of AI chips, telling the world he is keeping his options open with his biggest customer.
Crack #2: OpenAI
Two weeks before that, OpenAI announced a $10 billion partnership with Cerebras, a chip company that directly competes with Nvidia. The deal runs through to 2028. OpenAI is now actively building infrastructure that does not rely on Nvidia.
The two most important companies in the AI supply chain are hedging against each other.
Crack #3: Microsoft
Microsoft's latest earnings report showed $37.5 billion in AI infrastructure spending in a single quarter. More than the GDP of most countries, yet the stock dropped -10%.
Why?
Because Azure growth decelerated.
Microsoft's core product, which is supposed to monetize all that AI spending, is slowing rather than speeding up. The market is now genuinely concerned that the increases in expenditure will not translate into outsized growth.
Three "dam" Cracks
Three companies that were supposed to be reinforcing each other are now hedging against each other.
The $600B Question (insiders are exiting)
Sequoia Capital (an early investor in Google and Apple) recently published an analysis where they posed the $600 billion question:
How much revenue does the AI industry need to justify the current infrastructure spending?
The answer: $600 billion per year.
The actual revenue: roughly $35 billion.
The industry needs to 17x its revenue just to break even on what has already been spent.
For every dollar these companies spend on AI right now, they get back 6 cents in AI revenue.
The Revenue Gap
And the spending is only accelerating. Microsoft, Meta, Google, and Amazon have committed over $560 billion to AI infrastructure over the next two years.
Here is where it gets interesting: while this plays out in public, something else has been happening in private.
Peter Thiel's hedge fund liquidated its entire Nvidia position. SoftBank sold nearly $6 billion in Nvidia stock. Warren Buffett cut his Apple stake by three-quarters and is now sitting on $380 billion in cash (the largest pile in Berkshire's history, btw).
And Jensen Huang, the man who built Nvidia, sold $1B+ of his own stock over the past 12 months.
The person who knows more about Nvidia's future than anyone else on Earth looked at these prices and decided he would rather have cash.
Maybe it is diversification. Maybe it is estate planning. Maybe it means nothing.
I'm just bringing it to your attention that Buffett, SoftBank, and the CEO who built the dam are all reducing exposure at the same time.
The Bull Case
Before I leave you with doom, I owe you the other side. Because if I don't, this is just a panic piece, and that's not what my page is about.
The dotcom crash wiped out 80% of internet stocks. Amazon survived, Google was founded after the crash, and Apple went on to become the most valuable company in history. If you held through the wreckage, you came out the other side richer than when you started.
The same could happen here.
AI is not fake. The technology is real and powerful. The question was never whether AI changes everything; it will.
The question is whether the financial structure built around it can survive long enough for the revenue to meet expectations.
Maybe it does. Maybe agents and automation unlock enterprise spending at a scale no one has yet modelled.
Maybe 6 cents becomes 60 cents becomes $6. Maybe the dam holds, and the pressure slowly releases as revenue grows into the valuation.
There is also the practical problem with being bearish: being right and being early are the same as being wrong.
The timing is the one thing nobody can predict. And it's hands down the thing that matters most.
So, what do you actually do about all this information?
What I'm Personally Doing
I'm not a financial advisor, and I can only say what I'm personally doing with all this information.
Cash
I'm sitting heavier in cash than normal. When the dam breaks, cash is the highest ground you can be on. It gives you the ability to buy quality assets for pennies on the dollar when everyone else is being forced to sell. Optionality is the most underrated asset in a correction.
Gold
When trust in the structure breaks, people run to things that can't be inflated away. Central banks are already buying gold at record rates. I'm thinking of gold as "insurance" against the dam breaking.
Selective Equities
I'm not saying sell everything. I still hold index funds. But I'm being more selective about what I own.
I'm focusing on companies with: real profits, real cash flow, and real customers.
A.k.a., companies that don't need the circular loop to survive.
Invest in Yourself
This is the one I'm focused on more than any other.
Build skills that don't depreciate.
Think: Taste, AI skills, leadership, complex problem solving (things that make you the person who manages the machines rather than the person replaced by them).
For those of you interested, I recently wrote two articles that teach you how to position yourself in this current AI bubble.
1. How to Invest in AI
- Seven AI Skills to Build Now
Final Thoughts
This is not the article I planned to publish today.
I spent a long time sitting with this one before hitting the post button, because it goes against the optimistic tone I try to bring to AI Edge, but I think you deserve the full picture and not just the upside.
My next article covers how to make yourself irreplaceable in the AI era. The skills, the positioning, the mindset - all of it. It is the most practical follow-up to everything I just wrote.
Follow me @aiedge_, and it will be on your feed in two days.
Lastly, please Like/Repost this article so others can find it 💙
