A Tale of Two Bubbles
While AI is driving significant investment, U.S. economic growth remains lower than during the 1990s Internet boom due to the drag of Trump’s contradictory policies. Should the AI financial bubble burst, the global economy would pay twice. A commentary by Moreno Bertoldi, and Marco Buti
In general perception, the Artificial Intelligence (AI) revolution is even more important than the Internet revolution. There are good reasons to think so: AI is a general-purpose technology with enormous potential, and its economic impact will ripple across the entire economy.
In terms of GDP, in the United States, investment in data centres alone this year has already surpassed total telecommunications investment at the peak of the Dot.Com bubble in 2000.
Unless it goes out of control with devastating effects far beyond the economy, AI has the potential of generating significant increases in growth and productivity rarely experienced in economic history.
However, if we compare the economic dynamics of the Internet revolution and the AI revolution, there is a puzzle: U.S. economic growth between 1995 (the year Internet became economically relevant) and 2000 (the peak of the Dot.Com bubble) was much stronger than that between 2023 (the year AI became economically relevant) and 2026.
The results in terms of total factor productivity growth (i.e., the impact of technological and organizational innovations on growth) and labour productivity are currently somewhat lower than during the Internet revolution.
However, this is far from elucidating the growth differential between the two periods (real GDP grew by more than 4% in the first period compared to the current 2.5%).
How can this discrepancy be explained?
Is it perhaps a manifestation of the paradox noted forty years ago by Nobel laureate in economics, Robert Solow: “you can see computers everywhere except in the productivity statistics”?
Or is it simply too early and we have just to wait for the effects of AI to trickle down the economy? In reality, there is no real puzzle: the discrepancy is essentially the result of the economic policies conducted in the two periods.
During Bill Clinton’s second presidency, fiscal policy remained cautious and used the leeway created by strong private sector growth to consolidate public finances: few remember this, but between 1998 and 2001, the United States had a federal budget surplus and public debt as a share of GDP was less than half of today’s (55% versus 125%): the U.S. met easily the Maastricht criteria!
Internationally, the United States, like the rest of the world, enjoyed the benefits generated by the “global liberal order” and the dividend of the “end of history” that they themselves had fostered.
The comparison with the Trump’s AI era is discomforting for the latter.
First, it should be noted that in the last two years of the Biden presidency, average growth was only slightly lower than that during the Internet take-off (1995–1996): 2.8% compared to 3.2%.
It is with Trump’s arrival at the White House that the gap widens.
Whilst the U.S. economy grows 2.2% on average in the 2025–2026 period, this is just half of the 1997–1998 period (4.4%), and this despite the strong acceleration of investments in the AI sector and an expansionary fiscal policy with the federal deficit exceeding 6% of GDP.
The underwhelming performance of the US economy is the result of the chaos generated by Trump’s tariffs, the conflict with Iran, the inability to lower the cost of living (the so-called affordability crisis), the looming fiscal crisis that risks, in time, to debase the dollar as the dominant global currency.
In short, while in the late 1990s the Internet and the rest of the economy moved in sinc, under Trump the real economy stagflates, and it is only thanks to massive investments linked to the AI revolution that its growth rate remains decent.
To be sure, economic policies in the 1990s were far from ideal. Serious mistakes were made then as well: from the failure to reduce inequalities—which would produce the rise of populism and ethno-nationalism—to financial deregulation, which would be among the main causes of the Great Financial Crisis of 2007–2008.
Nevertheless, there is no doubt that its results were clearly superior to those generated by an incoherent and contradictory hodgepodge of measures inspired by short-sighted economic nationalism.
Furthermore, the policies run by the Trump administration —from unchecked deregulation to the promotion of cryptocurrencies, from tax reforms favouring the wealthiest classes to welfare cuts— seem aimed at reproducing on a larger scale the mistakes made during the 1990s.
All of this is compounded by the denial of any form of public regulation to mitigate AI-related risks, and even the refusal to encourage some form of self-regulation as suggested by leaders of the technological revolution.
Finally, the Internet revolution metastasized into the Dot.Com bubble. Its burst in 2001 caused significant losses for many investors, but a strong Main Street helped absorb the vagaries of Wall Street.
What ensued was a brief recession in the United States and some European countries, and a slowdown in the global economy in 2001-02.
Still, the burst of the bubble did not trigger a systemic crisis and the global recovery that followed was relatively strong.
We do not yet know for certain whether there is a financial bubble in AI, but many economic and financial indicators point in that direction. Being economically more pervasive, if the AI bubble were to burst, its impact would certainly be more severe than that of the Dot.Com bubble—not only for the United States, but also for the rest of the world.
Because of the financial and supply chain transmission mechanisms, the latter would find themselves in the position of having to pay a very high price for a bubble that Trumpian policies helped inflate without having benefited from the acceleration of growth that these types of bubbles initially produce – and that we did not see this time because of the international turmoil we are in.
In short, the European Union and other economies would pay double for the global chaos that Trump has generated.
A previous version of this article was published by Il Sole 24 Ore
IEP Bocconi 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.
