The AI Boom Is America’s Bubble and Europe’s Problem
The concentration of market value around artificial intelligence is reshaping US capitalism. If the bubble bursts, the rest of the world may pay the higher price. A commentary by Moreno Bertoldi, and Marco Buti
With SpaceX’s listing, measured by market capitalisation, the ten largest US companies, in descending order NVIDIA, Alphabet, Apple, Microsoft, SpaceX, Amazon, Broadcom, Tesla, Meta and Micron, now all operate in the field of artificial intelligence. Never in the history of American capitalism has so much value been concentrated in a single sector.
These companies account for more than 40 per cent of the market capitalisation of the S&P 500, far above the 27 per cent peak reached in 2000, when the dotcom bubble was at its zenith. If current trends continue and the expected listings of OpenAI and Anthropic deliver the hoped-for results, the value of AI could rise even further.
This concentration reflects a systemic evolution in US capitalism, which has reconfigured itself around the “new triad” of large technology oligarchs, crypto-assets and fossil fuels, as we argued in the European University Institute’s Florence Report.
The central question is whether this situation is financially sustainable, or whether it is a speculative bubble destined to deflate, or worse, to burst.
Those who believe current valuations reflect these companies’ fundamentals point out that Big Tech profits have risen sharply and are expected to increase even faster in the future.
As a result, the ratio between share prices and future earnings would not be overvalued. It is no coincidence that some analysts are revising upwards the market values these companies could reach by the end of the year.
This optimistic view is far from universally shared.
While AI-related investment was initially financed mainly by the retained earnings of Silicon Valley companies, it is now being driven largely by private credit and equity capital, with investors chasing extraordinary future gains. The risk of a technology-finance bubble cannot therefore be excluded, since the profits that are ultimately generated are very likely not to satisfy the expectations of all creditors and shareholders.
Moreover, many investors treat AI as an undifferentiated asset class, paying little attention to the fundamentals of individual companies. This creates phases of euphoria followed by bouts of panic. That is often what happens during speculative bubbles.
It is always difficult to identify a bubble ex ante. Yet, as the English say, if it walks like a duck and quacks like a duck, then it is probably a duck.
The current AI boom could be tempered by a regulatory framework capable of reducing the risks inherent in this technology. Internalising those risks would allow a more realistic valuation of AI companies and help curb the irrational exuberance surrounding them.
Another option would be a monetary policy that, also in response to relatively high inflation, sought to deflate the speculative phenomenon gradually through a measured increase in interest rates.
Neither option, however, is currently on the cards. The Trump administration has no intention of moving towards tighter regulation. On the contrary, it is trying to encourage further investment in the sector, acting as cheerleader for the AI stock market boom. This is because it is essentially thanks to AI that the US economy, hit by self-inflicted shocks ranging from tariffs to the war in the Middle East, continues to grow.
In addition, regulatory capture by the “Tech brothers” prevents not only ex ante regulation, but also ex post correction. Nor is there much hope from the Federal Reserve. Its new chair, Kevin Warsh, persuaded Trump to appoint him precisely on the basis of miraculous projections about the impact of the AI boom on productivity and therefore, in a disinflationary direction, on inflation.
In a world in which financial markets remain globalised, the risk is that, if and when the bubble deflates or bursts, the highest price of adjustment will not be paid by those who generated it. After all, much of the infrastructure created by AI investment will not be destroyed and will continue to be used to support the growth of the US economy, hopefully in a more orderly and sustainable way.
By contrast, much of the rest of the world will suffer the consequences through trade and financial channels, without benefiting from the subsequent adjustment, since it lacks the infrastructure mentioned above. What little it does have may also be swept away by the correction, because in such situations only the strongest survive.
For this reason, the European Union must urgently reduce its dependence on US AI. The regulatory tools exist at European level. What is now needed is the mobilisation of resources for the necessary public and private investment.
Otherwise, in a perhaps not distant future, and not for the first time, Europe may once again be told: “It is our bubble, but it is your problem.”
A previous version of this article was published by Il Sole 24 Ore
IEP@BU does not express opinions of its own. The opinions expressed in this publication are those of the authors. Any errors or omissions are the responsibility of the authors.
