AI “bubble”: where do we stand?
Like the Internet and mobile telephony in 1999/2000, the AI boom is poised to produce a potentially destabilizing valuation bubble. Where do we stand on this front, as the AI world suffers its first shock with the arrival of low-cost ?
According to economist Charles Kindelberger, famous for his “World History of Financial Speculation” (1978), a bubble is defined as “a sharp rise in the price of an asset whose initial increase prompts further rises and attracts new buyers interested in the profits of trading the asset rather than its earning capacity”. In his book, Kindelberger describes a multi-stage exponential trajectory. The first stage is that of a change, for example in technology, about which investors become increasingly optimistic, then euphoric, before a trigger shakes confidence, leading to a crisis and then panic.
From this point of view, the rise of artificial intelligence (AI) seems a serious candidate for the formation of a large-scale speculative bubble. Along with disinflation, which improves financial conditions, AI is one of the pillars of the global stock market rally that has been underway since autumn 2022. This theme is contributing to the rise of what Keynes called the “animal spirit” of investors. It conveys not only positive expectations for the companies most involved, but also hopes of productivity gains for the economy as a whole. As was the case in 1999/2000 with the Internet and mobile telephony, enthusiasm for AI is spreading to all equity markets, and in particular to growth stocks. In terms of PER, the valuation of this universe remains below its 2000 peak, but is at its highest since 2001 if we omit the parenthesis of the Covid year in 2020 (graph 1).
The Magnificient 7 American companies now account for 43% of the market capitalization of the MSCI World “Growth”, and 60% of the MSCI USA “Growth”. In terms of valuation, however, the rest of the growth stocks have tended to join the Magnificent 7, testifying to a spread of optimism, particularly in the USA, where growth stocks are on average valued at the same level as the Magnificent 7 (graph 2). To a lesser degree, this is reminiscent of the late 1990s, when all growth stocks, even the most traditional, were priced at 40 to 60 times earnings.
Another potential bubble marker is the acceleration of corporate investment in artificial intelligence. This acceleration is particularly visible in some very large companies, including of course the Magnificient 7 (graph 3). At over 12% of sales in 2024, their investment ratio has risen sharply over the past year, essentially due to the rise of AI. However, this surge is fairly recent, and the profitability associated with these investments seems satisfactory for the time being. The arrival of “low-cost” programs, such as China's Deepseek and others, undoubtedly calls into question certain aspects of this profitability, but also opens up new prospects for accelerating AI adoption (deflationary boom).
For US companies as a whole, the situation is very different from that of 1999/2000, which was characterized by a boom in external investment financing (financial gap) and a decline in profitability. Today, the aggregate level of the financial gap remains low, while profits are at their highest as a percentage of GDP (graph 4).
All in all, the AI theme ticks a number of bubble boxes, but not all of them. The AI narrative is both highly optimistic and widely disseminated, helping to inflate the valuation of growth stocks, and inducing the start of a boom in corporate capital spending. It is taking place in an environment of falling interest rates and US economic optimism, conducive to risk-taking that could prove excessive. However, the profitability of the main players remains very good, and leverage modest at this stage.
Despite the legitimate doubts raised by the emergence of a “low-cost” AI, we do not believe that the situation calls for any major changes in our positioning. The balance between global economic growth, inflation and interest rates remains favorable. What's more, our international team-weighted strategies enable us to guard against the risks associated with concentration and high valuations in certain parts of the market. On an equal-weighted basis, the global equity market today carries a PER close to 15, which is in line with the historical average (graph 5).
