AI: boom, bust or sustainable earnings bump?

Despite mixed tech earnings in the past couple of weeks, the NASDAQ 100 continues to be up 42.2% year-to-date, reaching a record-high of 15860 points at the end of July.^[1] Meanwhile, some of its constituents are trading at record-high price-earnings ratios, some of them well above 200,^[1] in our opinion, usually driven by excitement about artificial intelligence (AI). At such vertigo-inducing valuations, it is hard not to think of some parallels with tech enthusiasm at the millennium.^[2] ^[3]

Such superficial parallels can be misleading. Historically, high price-earnings ratios for tech stocks in novel areas are far from uncommon, given the substantial growth potential such firms may turn out to offer.^[4] Of course, high price-earnings ratios are only justified if and when the hoped-for earnings growth eventually materializes

Our Chart of the Week illustrates the price-earnings ratios and earnings per share of the tech-heavy NASDAQ 100 for the last 10 years. Earnings multiples are well above historical averages, suggesting expectations of strong future growth. However, the NASDAQ 100 multiple of 35 times earnings is still far below its record high of over 110 (!) during the peak of the dotcom rally.^[1] At the same time, earnings expectations have been increasing recently, suggesting that the current rally is mostly driven by expectations of rising earnings in the coming quarters.

Investing in the era of AI

Artificial intelligence is likely to prove one of the most consequential developments to emerge during our professional lives as asset managers.

What the NASDAQ 100 is already pricing in

_{* Earnings per share}

_{Sources: Bloomberg Finance L.P., DWS Investment GmbH as of 8/2/23}

A more suitable comparison to current times may be the experience of the Web 2.0 era during the mid-2000s. AI certainly has the potential to improve productivity growth. Recent research from McKinsey estimates that generative AI could add the equivalent of $2.6 trillion to $4.4 trillion annually across 16 global business functions.^[5] Yet, getting there could take a while. “The view that the world gets from the media”, AI veteran Stuart Russell argues, “bears very little relation to what really happens in the world’s research labs”. In other words, in AI, as in other areas, “A good idea – a real breakthrough – will often go unnoticed and may only later be understood as having provided the basis for a substantial advance in AI”.^[6]

Of course, even AI tools already developed might bring revolutionary changes to the productivity of numerous professions – once firms learn how to make the most of them. Artificial intelligence, like steam engines or electricity, is a general-purpose technology that will take time to be adopted across industries.^[7]

Which brings us back to those dizzying NASDAQ price-earnings ratios, currently seen on Wall Street. Justifying those on a case-by-case basis, with detailed assumptions about products, competition, use cases and the like remains of utmost importance. Currently, there are only a few, very experienced companies developing and deploying AI models profitably in their product suite. However, the majority of companies is just getting started with AI and eventual, future AI winners might not even been founded. “As with any other product, not all market participants will develop profitable products” warns DWS AI portfolio manager Felix Armbrust, highlighting that “identifying AI winners and losers requires experienced investors that have a realistic view on the topic”.