How to Survive (and Thrive) in an AI Bubble

Dear Reader,

Happy Thursday.

Last night was parent-teacher night at my daughter’s high school. I got to meet her teachers, and thank goodness—they seem like solid people.

Next week I’ll meet my son’s elementary school teachers. I actually enjoy this kind of thing. It’s good to stay involved.

But today, let’s talk about something a lot of readers have been asking me:

What do you make of this AI boom?

Intuitively, people know the market is overvalued. By many metrics, it’s more overvalued than the dot-com bubble.

So how do you stay sane in a time like this?

How do you stay invested without getting crushed when the music stops?

I’ll tell you something—it’s not easy.

What I’ve learned over 35 years is that investing is basically a battle against yourself. Against your own emotions.

The temptation to get swept up in the hype is powerful. And until you learn discipline, it’s very hard to resist.

So let’s put this AI boom in context.

First, yes—AI is going to destroy a lot of businesses, just like the internet did in the 1990s.

But here’s something interesting: when you look at U.S. GDP, it tells a different story.

From 1987 to 1997—the decade before Netscape went public—U.S. real GDP growth was almost identical to the decade after, 1997 to 2007.

Even including the financial crisis, the economy didn’t suddenly leap to a new level because of dot-com.

That’s a reminder: no matter the hype, the math doesn’t lie.

Right now, total S&P 500 pre-tax income is about $2.5 trillion. That number, and the speed it grows, ultimately dictates market value.

Can earnings really grow fast enough to match today’s soaring market cap? That’s the real question.

Now, one reason earnings look strong today is Trump’s “Big Beautiful Bill.”

It allows for full expensing of new factories, capital equipment, and R&D.

That’s fueling a major CapEx boom.

And it reminds me of the 1990s telecom buildout.

Remember the fiber boom? Companies laid miles and miles of cable. The suppliers—Lucent, Nortel—booked big revenues.

But the underlying returns weren’t really there.

That’s exactly what we’re seeing with AI chips today.

The “Mag Seven” are buying Nvidia and AMD chips as fast as they can.

The chipmakers book those sales as profits.

But the buyers—Amazon, Microsoft, Meta—are capitalizing those expenses over five, six, even seven years… even though the chips could be obsolete in two or three.

That accounting illusion creates a temporary boost to corporate earnings. But it’s not sustainable. It leads to painful write-downs later.

So yes, there’s a bubble here.

The only question is: are we in 1997, 1998, or 1999?

Now, don’t misunderstand me.

Right now, the market has a bullish bias. That’s just reality.

Anybody who says otherwise is selling you something you don’t want to buy.

Sure, there are always permabears. The Chicken Littles who spend 30 years screaming: “The sky is falling!”

Like broken clocks, they’re right once a decade.

But that’s not a strategy.

Here’s where I stand.

Yes, I’m paranoid enough for all of us. I’m keeping a close eye on this bubble.

But with interest rates heading lower, where else can investors go?

Yesterday I mentioned my own dilemma: what do I do with money in 4% money-market funds if they fall to 3%?

That’s basically the inflation rate. It’s just treading water.

So like it or not, the stock market is still the only game in town.

For better or worse, that’s where the money will flow.

The real question isn’t whether AI is a bubble—it is.

The real question is: are we at the beginning, the middle, or the end?

That’s what we’ll be watching closely over the next quarter, as hype gives way to hard questions about returns.

Until then, let’s keep making the smart plays.

Enjoy your Thursday.

“The Buck Stops Here,”