The AI has been described as a new industrial revolution. It’s the gold rush, the new electricity, the transformative force that will thoroughly change this world we live in. But here’s the quiet whisper coming out of the financial engine room right now: The current AI-driven market bubble is up to seventeen times larger than the dot-com bubble was at the turn of the century.
The dot-com era left many in shock when it burst, and the AI boom could make a similar effect. In order to avoid groundless speculations, this phenomenon requires a serious and methodical analysis.
The Financial Forensics
According to Julien Garran, a research analyst at MacroStrategy Partnership, the AI market is seventeen times larger than the dot-com . An important thing to consider is that this isn’t a comparison of one index versus another. It’s a measure of market misallocation, e.g. the difference between the cash being poured into a sector and the actual revenue and profit being that’s generated in the short term.
The dot-com bubble’s culmination in 2000 was defined by companies with P/E (price-to-earnings) ratios in the hundreds – to simplify, that means that there was very little revenue and that the companies were overvalued. The market cap of the companies that are the main actors of the AI boom really dwarfs the relative scale of the 2000 dot-com crash, especially if we adjust for the overall size of the global economy. This applies both to the large publicly traded tech enterprises, and the privately valued, multi-hundred-billion-dollar AI start-ups.
There are several reasons why this could be worrisome. First of all, the market concentration is bigger than it used to be. At the height of the dot-com mania, the top ten stocks in S&P 500 accounted for less than 27% of the whole index. As of right now, that figure is close to 40%, with a small amount of players that are directly tied to AI in one way or another. This includes tech giants selling chips, software and cloud computing companies etc. Naturally, if this over-leveraged pillar loses its grip, the corrective drop will be steeper.
While the P/E ratios for the wider market are not on the catastrophic levels that NASDAQ saw twenty-five years ago, some key metrics like the Shiller Cyclically Adjusted P/E (CAPE) ratio are coming very close. CAPE is defined as the price divided by the average of ten years of earnings (moving average), adjusted for inflation. This may suggest that the market is pricing in a decade of relatively uninterrupted profit growth. Some would argue that this represents a completely unrealistic expectation for basically any technology, however revolutionary it may be.
Unlike the dot-com era, where several start-ups were fuelled by equity that was only newly public, the AI infrastructure as we know it today is currently being funded by the cash-rich balance sheets of successful companies like Microsoft or Alphabet. These funds committed to an industry whose return on investment is most probably years from being realized might be a cause for misallocation alarm. The companies are being valued at a very high price even though clear and scalable business models have not been established in the real world, outside of infrastructure sales.
Of course, hardly anyone would argue that the underlying technology (artificial intelligence) is not transformative, which is the crucial difference from the 2000 crash. But the investor exuberance is really running ahead of the technology’s adoption lifecycle, and the sheer amount of the capital involved in this whole market makes the previous tech correction look like a fender bender. When the burst comes, it will be felt as an economic shock in most of the world.
The EU’s Role
For Europe, this volatile, US-centric market is both a threat and a massive opportunity that could’ve been a calculated one.
In the past few years, the European Union has been focused on its role as the AI regulator, culminating in the AI Act. The recent announcements by the President of the European Commission, Ursula von der Leyen, signal a decisive shift from a purely regulatory stance to an industrial and scientific one.
The twin initiatives, the Apply AI Strategy and the AI in Science Strategy, are the EU’s response to the bubble in the making.
The former’s goal is to move AI out of the lab into industries such as manufacturing, automotive, healthcare, and energy fast. The EU tries to make a “buy domestic” approach with the announced Apply AI Alliance in order to strengthen its tech sovereignty. The EU now understands that they might not be able to build the AI infrastructure in the way the US or China did, but they acknowledge that they need to start applying the existing one if they want to remain remotely competitive.
The latter is more interesting, as it is a game changer in Europe. The EU wants to launch different initiatives, such as RAISE , that will pool its resources to build its own advanced AI models. This is a very straightforward bet on European research, which the EU will want to use to break the US and Asian domination in large-scale model development and become a key player on the world stage.
This European strategy could be regarded as risky. They want to regulate the market to protect citizens from risk, while simultaneously injecting significant amounts of public-private investment to avoid being economically outpaced by foreign companies.
If the AI bubble bursts, the global markets will be hit bad. However, if the EU’s plan becomes successful, the correction might finally afford European players the chance to catch up and build sustainable, regulation-compliant AI systems while their global competitors will need to take a few steps back and restructure their businesses until they financially recover. The overvaluation will end sooner or later, and the question is if the world (and the EU in particular) is ready for a crash that could be bigger than anything we’ve seen this century.
