Is AI Really a Bubble? What History Actually Tells Us About Today’s Market

The Bubble Question Everyone Is Asking

I keep getting this question from listeners: is AI a bubble? And not just a casual question either. People are genuinely nervous. Some are comparing today’s AI spending frenzy to the lead-up to the global financial crisis. Others are throwing out the Enron comparison. I understand why people are asking. But I think most of those comparisons are missing the mark, and I want to walk you through exactly why.

Why the Enron Comparison Doesn’t Hold Up

I was one of the few people who called Enron before it collapsed. You can read my original piece on my website. The reason I was skeptical came down to one simple thing: I couldn’t understand what they were actually doing or how they were making money. They were trading bandwidth. They were planting hard assets in countries where people historically don’t pay their bills. Fortune magazine named them the most innovative company in America five years running, and I wrote a column saying I didn’t get it.

The market proceeded to go up for another year. I got called a fool. Until I wasn’t.

That’s how markets work. But here’s the key distinction with Enron:

• The business model was fundamentally incoherent
• There was no traceable cash flow tied to real economic activity
• The complexity was designed to obscure, not to create

With AI infrastructure spending, you can actually follow the money. You can see the chips being built, the data centers going up, the enterprise contracts being signed. That doesn’t mean valuations are perfect. It means it’s a different animal.

Why the Great Recession Comparison Also Falls Short

The 2008 financial crisis had a very specific architecture. It started with subprime mortgage products that were engineered under bad policy going back to the Clinton era, pushed aggressively by lenders like Countrywide, and then packaged into instruments that nobody truly understood. The entire system was built on the assumption that home prices would never fall nationally at the same time.

I spotted problems as early as 2006. Homes in my neighborhood that sold for $130,000 to $140,000 were relisting at $300,000 twelve months later with zero improvements. No pool. No renovation. Nothing. I kept asking: where is the value that was actually created here? There was no answer, because there wasn’t any.

The critical factor in 2008 was the complete absence of underlying cash flow. It was a game of musical chairs built on debt layered on top of debt.

• The mortgages had no real repayment foundation
• The banks were so far behind on risk management it was criminal
• When the music stopped, the dominos fell in sequence

AI spending today has its own risks, but the structural rot that defined 2008 is not the same thing.

What to Actually Watch For in AI

Does that mean investors should just pile in without thinking? Absolutely not. The right questions to be asking about AI right now are:

• Is the capital spending sustainable without a clear monetization timeline?
• Are companies building real products and services, or are they just raising money on hype?
• How is the financing structured, and who is holding the risk if growth stalls?

These are legitimate concerns. A sector can be genuinely transformative and still produce terrible stock returns if prices run too far ahead of fundamentals. That’s not a bubble in the Enron or 2008 sense. That’s just a valuation problem, and those correct themselves differently.

My Bottom Line

I’ve been doing this long enough to know that bad comparisons lead to bad decisions. Investors who panic-sold tech in 2003 missed the recovery. Investors who assumed 2008 was just a normal correction got wiped out. Context matters. The AI conversation deserves more precision than lazy historical analogies. Stay skeptical, stay informed, and always ask where the actual cash flow is coming from.