In 2023, more than 16.5% of Russell 3000 companies and more than 20% of S&P 500 companies (110) mentioned artificial intelligence or AI in their earnings calls. That shouldn’t come as a surprise: Businesses are falling all over themselves to deploy AI in their day-to-day operations, it’s been an unavoidable talking point on Wall Street, and it’s helped fuel this nascent bull market.
But with technology stocks now dominating the S&P 500 in a manner unseen since the dotcom crash of 2001, it does beg the question of whether we’re in an artificial intelligence bubble.
The trouble is, there’s not a clear-cut way to define a bubble until it’s popped. And even companies that were in a definitive bubble can go on to enormous success. Amazon (AMZN), for instance, traded at a split-adjusted 5.36 per share on December 10, 1999, and would trade at less than 10% of that value a mere two years later. If you “top-ticked” the dotcom bubble and bought Amazon at the peak, your shares would have returned 2,850% over the next 24 years. If you happened to get into shares at their September 2001 lows (around 0.30, again, split-adjusted), you’d have returned an astronomical 52,000%.
The dotcom bubble was, definitionally, a bubble. But patient investors in a well-run, forward-thinking company were handsomely rewarded just the same.
But for every investor riding high on their life-changing Amazon returns, there are dozens more that lost a pretty penny investing in Pets.com or companies like Yahoo!. Even so, the technology underpinning the bubble is now ubiquitous in our daily lives.
[text_ad]
Put another way, while some internet companies were definitely in an investing bubble, the technology of the internet was a game changer.
I think that’s likely to repeat itself for AI. Some artificial intelligence companies will no doubt turn out to be spectacular busts, but artificial intelligence is not, as a technology, a bubble.
That’s not to say there aren’t existential risks to artificial intelligence. Regulatory risk seems the most likely overhang, but there’s also risk that the currently most prevalent iteration of AI, large language models (LLMs), could fall victim to intellectual property lawsuits, such as those facing OpenAI today.
This brings us to something of a double-edged sword: If our base case is that artificial intelligence is not a bubble, but some companies touting it are, what are we to do as investors? After all, we invest in companies, not technologies.
Let’s use the framework economist Hyman P. Minsky outlined to identify the five stages of a credit cycle in his book, Stabilizing an Unstable Economy, as a roadmap of sorts to identify where we (or a company) may be in a potential artificial intelligence bubble.
5 Stages of an Artificial Intelligence Bubble
#1. Displacement
The displacement phase of a bubble is marked by the arrival of a paradigm shift; in the case of artificial intelligence, I would volunteer the arrival of ChatGPT in 2022. We wrote about it at the time, but it arguably took AI from a sci-fi concept to something that the average investor could envision as part of their daily lives. Large language models, like ChatGPT, have been in development for decades, with the foundation laid at MIT in 1966 by a program called “ELIZA.”
Rudimentary chatbots were widely available in the ‘90s and early aughts, with software like Dr. Sbaitso offering chatbot therapy for MS-DOS as early as 1992 and SmarterChild available as an AOL Instant Messenger tool in 2001.
For our purposes, let’s refer to the Displacement phase as, “Have you heard about…?” If friends, family, and the media are all asking if you’ve heard about some new AI tool, we’re probably still in the early innings.
#2. Boom
The Boom phase is marked by momentum in early (and obvious) investments in the new technology. If we pinned the Displacement phase to late 2022, artificial intelligence’s Boom phase probably began early in 2023 when the first investors started generating significant returns. One of the biggest beneficiaries of the Boom phase was Microsoft (MSFT), which was an early investor in OpenAI.
This likely applies to many investors who got into the Magnificent Seven tech stocks before that was a prevailing narrative last year. They were rewarded with 100%+ returns in 2023 and are still riding high early this year.
The Boom phase is also typically marked by fear of missing out (FOMO), where later investors chase returns in what was an “obvious” trade (in hindsight). I suspect we’re still in the Boom phase today.
#3. Euphoria
The Euphoria phase tends to coincide with high levels of overconfidence in a sector or an investment, with buyers justifying sky-high valuations with convenient narratives about how this time it’s different. We saw this phase before the dotcom bubble burst when company valuations were so divorced from their fundamentals that the fundamentals were thrown entirely out the window.
The pessimist in me suspects that we may be at or approaching those levels with some artificial intelligence names, but the clearest markers come from zooming out. The Euphoria phase becomes the most obvious in hindsight. (Of course such and such company couldn’t grow earnings at triple-digit percentages forever. Why would anyone think that an online retailer that needed high-speed internet could thrive when we all had dial-up?)
The prescription for the Euphoria phase of a possible artificial intelligence bubble is to zoom out and question whether you’re buying reasonable growth projections or hype. If it’s the latter, consider what your exit strategy is or tighten your stops, and be prepared for the disappointment of missing out on further upside gains.
#4. Profit-Taking
The Profit-Taking phase of a bubble is when smart money starts cashing out and heading for the exits. This may coincide with sell-side analysts capping their price targets (but there’s no guarantee there) but will also be visible in institutional sponsorship.
Where retail investors get burned in this phase is by the belief that, to borrow from the online parlance, an investment is “up only” or going “to the moon.” Institutional investors aren’t necessarily seeking life-changing returns; a home run is good enough, it doesn’t need to turn into a grand slam.
If you enjoyed a phenomenal run-up in an investment, consider taking some profits off the table. You can lock in a “win” while offering your remaining stake more breathing room.
#5. Panic
The Panic phase probably needs no introduction, you’ve likely seen it before. Stocks gap down and keep trading lower, investors sell as quickly as they can, and the biggest motivator to keep holding shares is the fallacy that they can “make it all back, if only x, y, or z happens…” If you find yourself caught up in the Panic phase, it’s smart to keep a few things in mind. First, why did you buy the stock in the first place? If those criteria have changed, you should reconsider the position. Second, are you holding and hoping for a rebound, or do you have a well-reasoned justification for disagreeing with the market about how much a stock is worth? Third, what’s the worst-case scenario? What happens if an investment goes to zero, or close to it?
The calculus is much different if you’re invested in a play-money flyer than if your retirement account is staked on it.
So, to answer the question, are we in an artificial intelligence bubble: We don’t know, nobody does. And we won’t know until after the fact. But bubbles are fueled by emotion and human nature, so zooming out and applying some analytical thinking can help you avoid getting caught up in the hype.
For our original coverage of large language models and ChatGPT, read, “Is Now the Time to Invest in Machine-Learning Stocks.”
You can find more information about artificial intelligence in our articles, “3 Under-the-Radar Artificial Intelligence Stocks to Capitalize on a Red-Hot Sector,” “The Artificial Intelligence (AI) Story Evolves,” and “Profit From the Artificial Intelligence (AI) Revolution With These Dividend Stocks.”
[author_ad]
