Welcome back to PyUncut — the place where technology meets truth, and hype gets separated from reality.
Today’s episode dives into a question that has suddenly become unavoidable in Silicon Valley, on Wall Street, and honestly, in every company trying to bolt “AI” onto their business model:
Are we living through an AI revolution… or an AI bubble waiting to burst?
And more importantly — what happens if the hype collapses before the economics catch up?
Let’s break it down.
The Hype Machine: Nvidia, Investors & the “Super-Cycle” Narrative
If you want a visual of peak AI mania, imagine Nvidia’s CEO Jensen Huang on stage in that signature leather jacket, telling investors there’s nothing to worry about. Nvidia’s market value has exploded more than 300% in just two years. That level of growth makes tech CEOs say one of two things: “This is the beginning of a revolution,” or “Please don’t panic.”
Huang chose the second.
He told investors there’s been “a lot of talk about an AI bubble,” but from where he’s standing, everything is booming.
And he’s not alone.
David Sacks — yes, the VC and White House AI adviser — is calling this moment an “investment super-cycle.”
Ben Horowitz insists demand will be “absurdly huge” for years.
JPMorgan executive Mary Callahan Erdoes says calling this a bubble is “crazy.”
But here’s the thing:
When everyone who benefits from the spending tells you the spending must continue… you better look under the hood.
The Harsh Reality Check: Growth Is Slowing, Costs Are Exploding
MIT’s Paul Kedrosky — a venture capitalist and researcher — says the quiet part out loud:
AI’s improvement curve has slowed dramatically.
We’re pouring capital into a technological revolution as if the underlying breakthroughs are accelerating. They aren’t.
Meanwhile, the money being spent is staggering.
OpenAI — the company that kicked off the modern AI race — claims $20 billion in annual revenue.
But here’s the real shocker:
It plans to spend $1.4 trillion on data centers over the next eight years.
Yes. Trillion, with a T.
Is the revenue there to justify that level of spending?
Not even close. Research shows:
- Only 3% of people actually pay for AI tools
- Most companies aren’t seeing productivity gains from chatbots
- And much of the AI value being promised is still speculative
MIT economist Daron Acemoglu, a Nobel Prize winner, says a huge chunk of what we’re hearing right now is “exaggeration.”
In other words:
The business model isn’t keeping up with the infrastructure bills.
Big Tech Goes All-In — Even if It Means Debt, Debt, and More Debt
Amazon, Google, Meta, and Microsoft together are on track to spend $400 billion this year — mostly on data centers.
Think of it this way:
Every iPhone user on the planet would need to pay more than $250 just to fund the AI expansion underway. And as Kedrosky notes, “That’s not going to happen.”
So how is Big Tech paying for all this?
By taking on a mountain of debt.
Goldman Sachs analysts found hyperscaler companies have added $121 billion in new debt in just one year — more than triple the usual level.
And some of the financing tricks being used look eerily similar to the financial engineering that helped take down Enron.
Here’s how one common setup works:
- A tech company wants a new data center
- A private equity firm fronts the cash
- A “special purpose vehicle” buys the chips
- The tech company leases the data center
- The debt never appears on the tech company’s balance sheet
Meta did this with a $27 billion data center in Louisiana — the debt sits elsewhere, but Meta is still tied to payments if things go wrong.
Gil Luria, an analyst who tracks these deals, says plainly:
“It’s not something we should be leaning on to build our future.”
The Ghost of Tech Bubbles Past
Morgan Stanley estimates Big Tech will spend $3 trillion building AI infrastructure by 2028. But their own cash flow covers only half.
If AI demand even slows slightly, we’ll end up with:
- over-built data centers
- debt that can’t be repaid
- financial firms exposed
- and another crisis reminiscent of the dot-com bust
Twenty-five years ago, fiber-optic infrastructure was built for a future that arrived much slower than expected. The bubble popped. Billions evaporated.
Today’s AI spending looks dangerously similar.
As Luria puts it:
“If we spend hundreds of billions and don’t need the capacity… we’re talking about another financial crisis.”
The Most Concerning Part: Circular Investing
We need to talk about the most eyebrow-raising trend of all: circular money loops.
Here’s an example:
Nvidia invests $100 billion into OpenAI to build data centers.
OpenAI uses that money to buy Nvidia chips.
So Nvidia’s investment boosts Nvidia’s sales — artificially.
It’s like paying someone $10 so they can buy your product and calling it “demand.”
This is the kind of behavior we last saw in the dot-com era.
CoreWeave takes it even further.
CoreWeave — once a crypto miner — now builds AI data centers.
OpenAI rents capacity from CoreWeave.
OpenAI receives CoreWeave stock as part of the deal.
OpenAI can then use that stock to pay CoreWeave for the services it rents.
And Nvidia owns part of CoreWeave… AND guarantees it will buy any unused capacity until 2032.
This is not a market reflecting genuine demand.
This is a feedback loop designed to manufacture demand.
MIT economist Acemoglu calls it what it is:
“A house of cards.”
Investors Are Starting to Flinch
Some big names are quietly backing away.
Peter Thiel sold his entire Nvidia stake — around $100 million.
SoftBank unloaded nearly $6 billion worth of Nvidia shares.
And Michael Burry — famous for calling the 2008 housing collapse — is openly betting against Nvidia. Whether he’s right this time is unclear, but his reasoning is sharp:
He argues that true end-user demand is tiny, and that AI growth is being propped up by these circular funding deals.
“Almost all customers are funded by their dealers,” he wrote.
That’s… not healthy.
Even AI’s Biggest Champions Are Nervous
Here’s when you know the bubble talk isn’t just Twitter noise:
- OpenAI CEO Sam Altman admits investors are “overexcited.”
- Google CEO Sundar Pichai says there are clear “elements of irrationality.”
- He also notes: “No company is immune if the bubble bursts.”
This is from the people selling the dream.
When the most optimistic minds in AI start sounding cautious… maybe it’s time for the rest of the world to pause.
The Bottom Line: What Does This All Mean?
AI is real. AI is powerful. AI will reshape industries.
But right now?
The economics don’t match the enthusiasm.
We’re spending money at a pace that assumes:
- AI breakthroughs will continue rapidly
- corporate adoption will skyrocket
- paying customers will appear in huge numbers
- and everything built today will be needed tomorrow
But the data suggests:
- usage is limited
- productivity gains are unclear
- and debt is rising much faster than revenue
This doesn’t mean an AI winter is coming.
But it does mean we may be living through a period of dangerous overconfidence.
In every tech boom, the same question decides who survives and who falls:
Is the demand real…
or is the demand funded by the hype itself?
Right now, AI is walking that line.
And we’re all going to find out soon which side it lands on.