Has the AI Bubble Popped?

The AI industry is in shambles. The world seems to have lost faith, and people's expectations are rapidly reversing. One could say being bearish on AI is now trendy, and will only get trendier after yesterday’s midterm elections.

So, has the bubble popped? Let’s look at what’s happening to the money financing the AI buildout, whether the latest model breakthroughs are delivering the productivity gains we were promised, and what Anthropic’s amended S-1 tells us about the cost of turning those breakthroughs into a business.

This article will give you dozens of data points to form an objective view of the industry, plus a framework for understanding what AI is missing.

AI is transformative, but the approach is severely broken. Let’s dive in.

The Macro Picture is Ugly

One issue I have with most AI analysts is that they analyze the technology in a vacuum, undeterred by its surroundings. “If AI continues to progress technologically, how is it possible that the world doesn’t see it?” they’ll argue.

They’ll even describe outsiders as “normies”, as if AI’s issues are not real; it’s that the world simply isn’t smart enough to see its value.

Besides the underlying arrogance I don’t need to comment on, this is a very clear example of missing the forest for the trees.

These very smart individuals are extremely enthusiastic about the technology, but bored to death by the economics, which leads them to forget they are closely related. You can’t analyze AI just by looking at technological progress; the economics, boring things like debt yields, move the needle just as much.

And the macro picture is very ugly.

Has the Bubble Popped?

Saying the markets are bearish on AI is an understatement. In this newsletter, we’ve been cautioning that AI was at risk of getting ahead of its skis for more than a year.

And, finally, markets have become convinced too. We’re starting to see several examples that might suggest that the bubble is popping, if not imploding.

For starters, several market indicators are starting to follow a trend very similar to the early 2000s, at the peak of the dot-com bubble.

The first, from Evercore, shows that the number of companies in the S&P 500 with negative beta (companies that have trended inversely to the overall market over the last six months) has reached higher levels than during the dot-com crash: almost 30% of the index.

This signals incredible divergence, not seen in more than two decades, explains what Goldman Sachs trader Bobby Molavi pointed out in a client note: that the S&P 500 is only about 3% below its peak while about 60% of its stocks are down 20% or more, showing that while most of the market is holding up, many stocks, especially AI companies, are seriously struggling.

But an even more telling example of how south things are turning is Morgan Stanley's TMT momentum index, showing an incredible negative momentum of 62% in tech stocks.

Momentum is computed as the ratio of the…

Enjoyed the introduction? Subscribe or sign in to continue with this edition.