Han Dieperink: The bubble in artificial intelligence

November 26, 2024

^{This column was originally written in Dutch. This is an English translation.}

By Han Dieperink, written in a personal capacity

Artificial intelligence is going to change everything. Yet the companies that have sufficient resources to develop profitable products can be counted on one hand. Those stocks have already risen sharply. For instance, Nvidia shares rose 1,000% in the past two years.

Often such a rise in share prices is cited to indicate that there is a bubble. The rise in prices is then contrasted with even limited returns, also to indicate that bubbles would not create innovation and big productivity gains. Now, there have been more paradigm shifts in the past that turned out to be both a bubble and a big change at the same time. Take railways and the internet, for example. Moreover, soap bubbles often become visible only when they burst.

A bubble

The current situation is ideal for the emergence of a bubble in the field of artificial intelligence. The technology is so amazing that there is a real paradigm shift: the prerequisite for the creation of a bubble according to Charles P. Kindleberger in his book Manias, Panics and Crashes. The economy will never return to the way it was.

It is also important for the creation of a bubble that we do not really understand how artificial intelligence will work as an industry. That promise in the distant future makes it the perfect fuel for bubbles: the second condition according to Kindleberger. There must also be a vehicle through which as many investors as possible can participate in the boom.

Gradually, an increasingly broader group invests in this issue: Kindleberger's third condition. Finally, according to Kindleberger, a bubble does require liquidity. This too seems to be met now that interest rates are being cut by central banks. With that, there is still plenty of room for a higher valuation.

Expectations for these companies are still being revised upwards and if you plot valuation against the growth rate, you will see that these shares are about 40% cheaper than the market. With that, there is a chance that these megacaps will outperform next year too. Bubbles are known to tend to overshoot at the upside.

Widening of the market

The peak of a bubble is impossible to predict. Valuation plays no role. Basically, the bottom line is that as long as there is growth in sales and profitability, the market does not tend to part with it quickly. Relatively, the performance of the Magnificent Seven becomes less unique. The reason why this group of stocks has done so well over the past two years is because the rest of the stock market has performed virtually nothing. In the land of the blind, the one-eyed man is king.

That is now changing. This quarter and also the next few quarters will see profits rise at other companies as well. These are therefore gaining traction, especially since many companies outside the Magnificent Seven are sometimes valued as if there could be a recession tomorrow.

Another reason why there is a widening market is that almost all companies can benefit from artificial intelligence. A company does not have to develop the technology, but only apply it in the right way to become a winner. Given the speed of implementation, 2025 is likely to be a year where we will focus less on Nvidia and the Hyperscalers that enable AI, but more on the companies that positively surprise by applying artificial intelligence.

Furthermore, the focus is still very much on generative artificial intelligence, while breakthroughs on autonomous driving vehicles, for example, are close at hand. Moreover, artificial intelligence is ideally suited to replace heavy, boring and mind-numbing work. Combined with the tight labour market, several positive effects then arise. Much more is produced at much lower prices. That's a reason to speak of an Uber Goldilocks scenario.

The function of a bubble

There are bubbles that serve a useful function. A bubble ensures that all investments are pooled together in a short period of time. Without bubbles, it would take years to achieve the same. Thanks to the Tulip Mania, the Netherlands is still a major exporter of flower bulbs. Thanks to the dotcom hype, we now have unlimited mobile internet at a relatively low cost.

Artificial intelligence has the potential to change the world to a much greater extent. Yet every bubble takes many casualties. Of the dotcom companies of the first hour, 99% have disappeared. Now that is part of the process.

The way to prevent bursting bubbles from doing too much damage in portfolios is to diversify. Not only in portfolio, but also by gradually parting with such a bubble. Fortunately, this bubble was recognisable at a relatively early stage. It is not clear whether the peak has been reached, but gradually opportunities to invest funds more broadly are emerging.

More columns by Han Dieperink