Equities: The AI fever pushes market to new extremes

Data on the US economy is still showing economic activity below trend growth but is also not showing recession dynamics, and earnings estimates have increased substantially, especially in Europe, since the Q1 earnings season started in mid-April. US financial conditions peaked in late March at levels that can still be characterised as loose given the economic backdrop and have fallen ever since to levels that are close to the loosest since March 2022, before the interest rate shock began to tighten financial conditions significantly. The previous rate cuts before year-end have been almost priced out. In other words, as markets enter Q3, the direction of forces are pointing more towards upside risks to inflation vs current market expectations and higher policy rates rather than the ‘back to low inflation and rates’ scenario.

While the AI hype has undoubtedly unleashed animal spirits and growing optimism around the boost to productivity, it has also created a growing risk to the US equity market. Valuations on the US equity market are back to the highest levels since April 2022 and with the free cash flow yield on the S&P 500 down to 3.9%, the market is beginning to look a bit more stretched. While it is too early to call for an overall bubble, the semiconductor industry is clearly showing bubble-like behaviour, and equity valuations on semiconductor stocks are the highest since 2010, measured on the 12-month forward EV/EBITDA metric.

Another risk that has come back, and something we thought would never happen again, is US equity market concentration. The rally this year has been carried by a narrow group of mega caps, as the emerging AI technology is expected to deliver the highest economic gains for the large technology stocks. The deeper issue is that the US equity market has moved to an index weight concentration we have never seen before, with the 10 largest stocks weighing 30.4% of the S&P 500 and the Herfindahl-Hirschman Index to a level 40% above the market concentration during the dot-com bubble peak. This makes the US equity market more fragile and sensitive to fewer risk factors. As a result, we have moved to a negative view on mega caps and overweight the long tail of equities vs mega caps.

The main question for global investors is still whether to reduce exposure to equities. Global equities measured by the MSCI All-Country World Index have a dividend yield of 2.3% and an estimated buyback yield of 1.2%. If we add expected real rate earnings growth of 2.2%, then the long-term expected annualised real-rate return on global equities is 5.7%. By comparison, global investment grade debt has a yield-to-worst of 3.8% but subtracting the 10-year inflation expectation of 2.5% takes the long-term expected return on investment grade bonds down to 1.2% annualised. In other words, if the investor wants to maximise long-term wealth, an overweight exposure to equities is still the most prudent.

The two faces of AI technology

New technologies create both positive and negative impacts on society, and AI is no different. A recent McKinsey report on generative AI suggests that the technology will add $2.6trn to $4.4trn annually, adding roughly the economy of the United Kingdom. McKinsey is predicting that 60-70% of work activities today can be automated with AI, enabling 0.1-0.6% points to labour productivity growth. Long-term predictions on technology are difficult, but the McKinsey report encapsulates the zeitgeist perfectly, as the AI technology is becoming our new hope for a better and richer future, a bit like the space age during the Cold War and the Internet two decades ago.

Our view is that AI-related stocks have entered a bubble phase marked by ‘peak of inflated expectations’, following the logic of Gardner’s hype cycle model. Soon, we will likely begin to see companies disappoint against those elevated expectations and take investors downhill to the trough of disillusionment before entering the slope of enlightenment. In order for our clients to get the best overview of the companies that have reacted the most to the AI-hyped rally, we have created a theme around AI consisting of 20 stocks. This list is already our most accessed equity research by far, underscoring the interest in AI. It is worth noting in the table of AI stocks that many of these stocks trade close to their price targets, indicating that equity analysts have difficulties justifying current equity valuations against the outlook.

The pessimists would argue that while AI will create economic gains in the longer run, the world is greatly extrapolating the current trend, creating a bubble in AI-related stocks. Google search volume in the US on ‘ChatGPT’ and ‘AI’ peaked in April and is already declining, suggesting that the initial hype is beginning to fade, although the hype has not yet ended in equity markets. The pessimists will also argue that generative AI will cause a flood of fake news, images and video, essentially polluting its own training data in the future, causing a natural stagnation in future systems, but even worse, that it will potentially break down trust in our information systems. This scenario could be a big comeback for traditional media as a trusted source of information.

The AI race between the US and China

Vladimir Putin said in 2017 that whoever becomes the leader in AI will become the ruler of the world. Russian leaders are experienced in hyperbolic language, so this prediction should naturally be discounted, but AI will likely play a crucial role in the great power competition of the future.

Reading articles about technology and AI from those years around 2017, it is clear that the world thought China was either leading the AI race or at least had the speed to overtake the US in a few years. Surprisingly, it turned out, that the US was leading everyone else, as AI systems such as OpenAI’s GPT-4 and Google’s Bard are crushing AI systems from China across many benchmark tests.

As we described in our previous Quarterly Outlook, the future will be dictated by what we call the fragmentation game, which is essentially a strategic geopolitical dynamic fragmenting the world into regions with a higher degree of independence and with national security interests driving policies around four pillars: defence, energy, technology, and commodities. The fragmentation game is mostly a game evolving around how the physical world operates and it is a game in which Europe and the US are aiming to reduce China’s role in their respective supply chains. While it causes headwinds for China, it creates tailwinds for other countries, which is well crystalised in our chart showing Chinese equity market performance vs those countries that are benefitting.

Inside the fragmentation game framework, semiconductors also play a crucial role because they are the very foundation for AI chips. The capital expenditures on semiconductors will create an investment boom in the US and Europe over the next decade, as the regions will increase domestic production to limit dependencies on Asia. This dynamic will benefit semiconductor equipment makers as their revenue figures are linked to capital expenditures on semiconductors. Regardless of the rollercoaster experience investors will have with AI stocks, one thing is certain: this technology will be an important technological battleground between the US and China, and many opportunities and threats will arise in the years to come.