The $279 Billion Tell: The AI Trade Just Got A Lot Harder

Introduction

I have been getting the same question over and over for about six weeks now.

Some version of: what do I do with tech here?

It comes from readers who missed the whole thing and feel sick about it. It also comes from readers who own it and cannot decide whether to trim. It came again this week, right before Nvidia (NVDA) reported, from someone who framed it about as honestly as anyone has framed it to me. He said he knows AI is real. He just does not know whether that fact is worth anything to him at today's prices.

That is the right question. It is also the hardest one in the market right now.

And it's also the question that keeps me busy, as I'm not anti-tech. I'm just against chasing a trend after it has become obvious. That's why I haven't bought back Comfort Systems USA (FIX) after I sold it - just to give you one example.

And you can see the difficulty in the tape rather than just in the sentiment. Chip stocks spent July getting hit on fears about whether anyone will ever see a return on this spending. Nvidia went into its own print on a seven-day losing streak, the longest since 2022. Then the numbers landed, and the stock jumped roughly 8%.

That is clearly not a market that has made up its mind. That is a market stuck between two stories at high speed, which is exactly what happens when a trade gets crowded enough that the price contains the entire thesis and nothing else. And it's exactly why I think this environment is super tricky.

As many of you know, my problem with the AI complex has never been the technology. The technology is real. The demand is real. The money is real. My problem is more boring than that, and much harder to argue away.

When a megatrend becomes popular, the price stops paying you for being right about the trend. It only pays you for being right about the magnitude and the duration of the trend. Those are far harder calls, and the penalty for getting them slightly wrong is severe.

We are there now. And this quarter, five different companies handed us the evidence, which means I just spent the past few days working through all reports to write one of my most in-depth articles on AI ever (yes, really).

So, as we have a lot to discuss today, I would say we get right to it!