Have We Encountered an AI Bubble?
Will AI eventually burst its bubble? Though it’s unlikely, the rally is nonetheless precarious.
With the help of artificial intelligence (AI) expectations, Nvidia has risen to the top of the stock market. The American chipmaker and the seven IT behemoths that are purchasing its components to train their artificial intelligence models have exorbitant valuations. A possible inflationary bubble is causing some investors concern.
One point of reference for the dreams and nightmares of today are the well-known “Nifty Fifty” stocks from the 1970s. Additionally, these companies were blue-chip powerhouses with inflated market values. Whether or not they managed to find a way to defend them is an open question.
Jeremy Siegel, a professor at the Wharton School, famously stated in 1998 that they did. His research demonstrated that a portfolio including all of these stocks had a poorer total return than U.S. equities overall—but only when compared to the market’s all-time high in December 1972.
This was challenged four years later by Jeff Fesenmaier and Gary Smith of Pomona College. Due to the lack of a formal Nifty Fifty list, Siegel listed other stocks with P/E ratios below 25, including General Electric and Philip Morris. If the Nifty Fifty were to be defined as the fifty most expensive stocks, they would have performed incredibly poorly.
Fesenmaier and Smith coined the term “Terrific 24” to describe a set of equities that appear on both lists. After a very bad performance since the epidemic, they would today have $95,000 if investors had divided $1,000 evenly in December 1972. In contrast, $185,000 would have been available to investors had they instead purchased the S&P Composite 1500 index. Among these companies, just four—McDonald’s, Eli Lilly, Merck, and Texas Instruments—have performed better than the market during the past year.
But Siegel was still right. Out of the 24 Terrific states, only 4 have declared bankruptcy in the 51 years following December 1972. It was the demise of analog photography that brought down two of those companies, Polaroid and Eastman Kodak. All twelve of the remaining independent companies are still considered major blue chips, with the possible exception of Xerox and IFF.
Market expectations are lower for most of the hot stocks of 1972 now, which is good news for “new economy” companies. However, it is remarkable how these large corporations have managed to stay afloat.
In contrast, there were a plethora of dot-com businesses that went bankrupt during the late 1990s dot-com rally. Amazon, eBay, and Booking.com were among the successful ones. Some faded into oblivion; for example, IAC acquired Ask Jeeves. However, the majority of the seventy or so internet stocks that experienced a meteoric rise in 1999 were entirely destroyed.
Consider the 2021 blank-check frenzy instead. A number of revenue-light firms that went public sought to adapt pre-existing technology to pre-existing markets. One such example is air taxi manufacturers. Compared to its all-time high, Solactive’s De-Spac Index—which monitors formerly blank-check vehicles—is 90% lower.
The importance of not placing too much stock in the future profits of well-established companies should go without saying. Investing $1,000 would have yielded $248,000 if you had chosen the five cheapest companies, even in the Terrific 24.
This, however, is distinct from speculating on the existence of any business, be it the British South Sea Company’s monopoly revenues in the 18th century, online sales of pet food in the early 2000s, or hydrogen vehicles today. The real definition of a bubble is this: everything is lost the moment it pops.
Since the extent to which huge language models will be commercially utilized is uncertain, the AI rally can give the impression of being overly enthusiastic. Many new businesses are receiving funding from venture capital firms, the vast majority of which will eventually fail.
However, a genuine stock market bubble is difficult to discern. One reason is that IPO activity is still low. The fact that blue chip software companies like Alphabet are already pouring millions of dollars into semiconductors means that there will almost certainly be an economic rationale for using generative AI to improve user interfaces. In the last quarter of 2023, sales for Nvidia were three times higher than the same period in the previous year. Because of this, even if the stock has gained 240% in a year, its price-to-earnings ratio is 31, which is lower than the recent average.
The true danger is shown by Cisco’s situation. The 126 price-to-earnings ratio it had in 2000 was obviously exaggerated, but the earnings volatility was a major contributor. Selling networking equipment to companies that were in a bubble was the problem, not being in a bubble itself.
Suddenly, Cisco’s earnings growth plummeted as the digital rush slowed and the market became oversupplied. Since then, investors have begun to see the company as nothing special, and its stock price is still below where it was during its dot-com heyday.
Nvidia and other semiconductor firms like AMD, ASML, and Arm should be worried about this situation. Their customer base is significantly more financially stable than Cisco’s was in 2000, which works in their favor.
Even while huge sums of money are being spent and made to prepare for a company with very doubtful future prospects, the word bubble does not apply to blue-chip stocks. What is the best alternative name for this type of market? Reasons on a piece of mail.
