The 2020s will go down in history as one of the best times to be a stock market investor. After a brief dip in 2022, the benchmark S&P 500 (^GSPC -0.22%) has soared by an average of roughly 21% per year — double its 100-year average annual return of 10%.
That said, history shows us that significant drawdowns often follow periods of unusually elevated stock market gains as the market reverts to its long-running mean. Let’s discuss some of the challenges facing this bull market to try to figure out what might come next.
Today’s Change
Index Level
7,801.77
The problem with valuations
By now, most investors have probably heard about the cyclically adjusted price-to-earnings (CAPE) ratio. This metric compares the price of the S&P 500 with its average inflation-adjusted earnings over the past decade. The long time frame smooths out short-term fluctuations, providing a clearer picture of the market’s value relative to historical norms.
Right now, the CAPE ratio stands at 40.5, which is well above its average of 17.4. The other two major peaks occurred before the Great Depression in 1929 and during the dot-com bubble in 1999, when it hit its all-time high of 44.19. Both milestones were followed by substantial declines in equity prices over the subsequent years as the speculative bubbles deflated.
History might not repeat itself, but it certainly tends to rhyme. And the current generative AI bubble shows strong similarities with the dot-com bubble two decades ago. Both are being driven by a transformational new technology that may not pay off as quickly as its major backers expect.
Can the generative AI bubble burst?
On some level, the generative AI boom looks much safer than the dot-com bubble because it is being driven by stable, profitable companies instead of speculative start-ups. Valuations are also quite reasonable, with major players like Nvidia and Micron Technology boasting forward price-to-earnings (P/E) multiples of just 25 and 6, respectively. Their revenue is simply growing so fast that their stock prices can’t keep up.
But while these factors will probably limit the scale of a potential crash, they don’t make one impossible, because the revenue itself comes from what appears to be an increasingly unsustainable source.
Goldman Sachs estimates that hyperscalers (Nvidia and Micron’s clients) could spend $800 billion on AI-related capital expenditures (capex) in 2026 alone. And if current trends continue, spending could rise to $1.4 trillion by 2028.
The problem is that these funds could otherwise go back to these companies’ shareholders via buybacks or dividends. And it may only be a matter of time before investors start punishing big spenders and pressuring management teams to adopt a more conservative approach.
Image source: Getty Images.
A recent Wall Street Journal study found that American businesses and consumers would have to spend the equivalent of 8.8% of the country’s GDP annually on AI to justify the industry’s immense spending. This sounds like a tall order, which raises the possibility that many current data center investments will never pay off.
Cloud computing giant Oracle is a good example of what happens when shareholders lose their appetite for AI spending. Shares have fallen by 50% over the last 12 months alone. And other big spenders could also come under increasing scrutiny as they continue to pour hundreds of billions into a yet-unproven opportunity.
What should investors do next?
Timing the market is notoriously difficult because even if you correctly identify the problem, it can take months or even years for the rest of the market to react. Instead of selling everything, investors should focus on taking some profits off the table and diversifying their portfolios away from stocks overly exposed to generative AI-related spending.
