Is this wildly overvalued stock market doomed? Yes, but maybe not yet

Their conclusion: Buy stocks anyway, because the next year looks great.

The underlying problem is simple to understand, and hard to do much about. Stocks are super expensive on just about every measure. That historically has meant low returns in the long run. Hence the dire 10-year forecasts.

But history also suggests no link at all between nosebleed valuations like we have today and returns over the next year. Expensive stocks can always get more expensive, and often do.

The S&P 500 and Nasdaq are at record highs, and U.S. stocks are really expensive. Six of the Magnificent Seven—Apple, Amazon.com, Meta Platforms, Microsoft, Nvidia and Tesla—are more expensive still, though Alphabet lags behind a little.

The justifications the bulls trot out become less convincing the more stocks rise. Three are popular: AI, rising earnings and American exceptionalism.

Artificial intelligencehas supercharged the performance of the biggest stocks as the companies plow vast sums into data centers and specialized microchips. JPMorgan estimates that capital spending and research by just the Magnificent Seven will be $500 billion in the next year, with a total corporate AI spend of more than $1 trillion in the U.S., bigger than the defense budget.

Investors anticipate fat returns from this capital. History begs to differ. In the past, stocks of companies with higher capital spending lagged badly behind those that spent less.

This time there is a particular reason to worry. Either this spending will produce useful products, in which case there will be lots of competing AI models, and the monopoly-like profit margins investors anticipate won’t be possible. Or it won’t, in which case most of it will have to be written off, even if there are one or two high-margin winners.

“Investors are assuming all this capital spending will be worthwhile," said Peter Berezin, chief global strategist at BCA Research. He has turned negative on stocks because of this and his view that a recession is looming. He predicts the S&P 500 will drop to 4100 by the end of next year. That is far below the 6500 and 6666 forecasts from Goldman and BofA strategists, respectively.

Lots of companies are experimenting with generative AI, but if actual use cases don’t come fast enough—perhaps because of genAI’s continued problems with facts and logic—investors won’t get the earnings they expect. I don’t know how to predict whether genAI will be widely adopted, or how much customers will be willing to pay. But it wouldn’t surprise me if it took longer and they paid less than expected.

Earnings have risen fast and are predicted to rise even faster, with Wall Street analysts expecting 14% growth next year.

Michael Strobaek, chief investment officer of Swiss bank Lombard Odier, is clear.

“It’s not like we have some bubblelike behavior where the valuations of everything have gone up, it’s been essentially backed by earnings," he said. “I think it’s solidly footed in good earnings and good U.S. productivity."

He might be right. But rising earnings and productivity were a big part of what was driving the market in 2000, just as the bubble of the century was about to burst.

Just as then, a rise in valuation has accounted for about half the rise in the market. Just as then, gains have been fairly narrow, with two-thirds of stocks and seven out of 11 sectors underperforming the S&P so far this year.

The dot-com stocks weren’t profitable, unlike the Magnificent Seven. But in 2000 the bubble was “TMT," technology, media and telecom, and the latter two parts of it were solidly profitable and wildly expensive.

Telecom companies spent big to build out the internet, and media groups spent big to supply online content. Both were right that there would be huge demand. Both were wrong that the demand would arrive quickly enough to cover their costs, and their stocks plunged.

American exceptionalism has convinced investors around the world that the U.S. is the place to put their money. They rightly point to rising U.S. productivity since the pandemic and the dominance of innovative U.S. technology groups in the stock market.

The twin problems are that this is already recognized in prices, and that historically investors have tended to misjudge how long such dominance will last.

The U.S. market is at its highest valuation relative to the rest of the world on forward earnings since at least 1988, when data collection began. Again, if you think AI will deliver even more than expected, that President-elect Donald Trump will bring a new age of American prosperity and that the rest of the world is doomed by tariffs, state intervention and geopolitics, there should be a U.S. premium. But the gap is really, really big. U.S. stocks trade for 22.5 times forecast earnings, and those earnings are far and away at a record high. The rest of the world is at less than 14 times earnings, which are still lower than was forecast in 2008.

The angst in Europe about American innovativeness mirrors the fear in Washington about Japan conquering electronics and cars in the 1980s, or the embarrassment of U.S. businessmen using clunky Motorolas when Europeans had advanced Nokias in the mid-2000s. Even now Japan leads in robotics, Denmark in anti-obesity drugs and Taiwan in microchip manufacturing. The U.S. might keep its lead in innovation, but there is a risk of repeating the overconfidence that doomed investors in Japan and Nokia to decades of losses.

Goldman and BofA are both worried enough about valuations to recommend cheaper stocks, even as they expect the market as a whole to rise next year. It isn’t unreasonable. But if you think the market is wildly overvalued and the long-run outlook is awful, you could stop worrying and buy 10-year Treasurys at 4.2%.

Write to James Mackintosh at james.mackintosh@wsj.com