FREE NEWSLETTER
IN AN INTERVIEW a little while back, the technology investor Peter Thiel drew an uncomfortable comparison. Today’s frenzy around artificial intelligence, he said, parallels the tech stock bubble of the 1990s. To illustrate his point, Thiel pointed to Amazon.
By any measure, it’s been an extraordinary success. But, Thiel points out, it hasn’t been a straight line. At one point early on, Amazon shares lost more than 90% of their value.
“My suspicion is that that’s roughly where we are in AI. It’s correct as a technology, but extremely bubbly and crazed…”
Thiel explained that he doesn’t doubt the importance of artificial intelligence as a technology. What he’s questioning is how these technologies are being financed.
Of particular concern are financing deals in the AI ecosystem that are seemingly circular. Nvidia, for example, has invested as much as $100 billion into ChatGPT maker OpenAI, at the same time that OpenAI has committed to spending billions on Nvidia’s chips. Similarly, OpenAI signed an agreement with AMD, another chip maker, to buy tens of billions of dollars of its chips while also buying a stake in the company. Transactions like this call into question whether these companies can continue to generate earnings at the same rapid pace.
Compounding this concern, market valuations are elevated. On a price-to-earnings (P/E) basis, the S&P 500 is trading at 21 times estimated earnings. That’s quite a bit above the long-term average of 16 and thus represents a risk. If investors cool on AI, both earnings estimates and P/E multiples would likely drop at the same time, causing share prices to take two steps down.
How unusual is this situation, and how concerned should we be about it? It turns out these are questions economists have been studying—and struggling with—for years.
Probably the most well known research on the topic dates to the 1970s, when economist Hyman Minsky developed what he called the Financial Instability Hypothesis.
This is how Minsky described it: “A fundamental characteristic of our economy is that the financial system swings between robustness and fragility and these swings are an integral part of the process that generates business cycles.”
Booms and busts, in other words, are inevitable. Why? Paradoxically, Minsky said, financial stability causes financial instability. That’s because periods of financial stability lead people to become overconfident and to assume that the good times will last forever. But that overconfidence leads to complacence and to a lack of financial discipline, especially among lenders. That then causes debt levels to rise.
What happens next? Writing in Manias, Panics and Crashes, Charles Kindleberger explains that there’s typically a canary in the coal mine that causes investor sentiment to shift. Often, it’s the unexpected failure of a bank or other institution. That’s why it caught people’s attention in February when Blue Owl Capital, which operates private credit funds and has helped finance AI data centers, announced that it was halting redemptions from one of its funds.
Looking at more recent research, economist Bill Janeway agrees with Minsky on the causes of bubbles but argues that they’re not all bad. He talks about “productive bubbles.” As an example, he points to the market bubbles surrounding the development of the British railway system in the 1830s and 1840s. Much like the 1990s tech bubble in the United States, investors piled into railway stocks, causing prices to spike to irrational levels. Overbuilding ensued, and that led to a number of bankruptcies.
Despite the financial losses, Janeway believes the railway bubble was productive. That’s for the simple reason that, at the end of the day, the tracks were laid. Yes, there were excesses, but Janeway sees no alternative. Investor enthusiasm acts as a sort of subsidy for early-stage, uncertain technologies that the market wouldn’t otherwise finance. The evidence certainly supports Janeway’s argument. The market does a very poor job picking winners.
Janeway notes that essentially the same thing happened in the 1920s, when investors piled into companies working to build out the electricity grid in the U.S. There was massive over-investment, which led to bankruptcies. But in the end, electrification projects were completed much more quickly than they might have been otherwise.
The key lesson: When market bubbles roll around, we shouldn’t be surprised. They’re inevitable. And over the long term, they’re arguably a good thing, enabling technology to move forward. Nevertheless, when bubbles burst, it’s unnerving. And indeed, in Janeway’s view, the same thing will likely happen with AI stocks.
If Janeway is right, how can you prepare?
The solution, in my view, is straightforward: Instead of trying to guess when the AI—or any other—bubble might burst, investors should take the view that the market could drop at any time. Then structure your portfolio accordingly.
There’s more than one way to approach this, but in my view, it’s a simple two-step process: First, make sure you’re diversified at the asset class level, with enough stowed in short-term bonds or cash to carry you through a multi-year market downturn. Then go one level deeper, auditing your stock holdings for individual stocks or funds overly exposed to any one corner of the market.
And if you’re in a private fund—especially a private credit fund—I’d identify the nearest exit.
Adam M. Grossman is the founder of Mayport, a fixed-fee wealth management firm. Sign up for Adam’s Daily Ideas email, follow him on X @AdamMGrossman and check out his earlier articles.
