Is AI a bubble that might burst?

It is hard to imagine that anyone could have missed the fact that such economic crashes occur from time to time – crashes that could have terrible consequences for individuals and society. In recent history, many vividly remember the IT bubble that burst twenty-five years ago and, not least, the financial crisis of 2008, when the financial system almost suffered a cardiac arrest and could only be saved by massive injections of money on a comprehensive and global scale. Even today, we are still feeling the consequences in the form of sharply increased debt for both governments and individuals, financial uncertainty and increased economic nationalism.

To even attempt to answer the question of whether there is a risk of an AI bubble and perhaps even a crash, we must first ask ourselves why such bubbles and crashes occur in the first place. Is it purely a matter of chance, and do they strike like a bolt from the blue? It practically goes without saying that no two crashes are alike and that there are always specific time-related factors that determine what form bubbles and crashes take. But we can learn something from history to identify certain similarities that can hopefully be recognised and prevented. But memory is often very short, and there are strong economic and psychological reasons why we often prefer to ignore the possibility that a crash may be just around the corner. As Kenneth Rogoff and Carmen Reinhart (two American economists) put it, we like to think that things are “different this time”, even when stock prices are skyrocketing.

Today’s economists have learned that humans are social beings and perhaps not as rational as one might think. Markets are not impersonal entities that act free from what the great 20th-century economist John Maynard Keynes called “animal spirits”. People make choices – in markets and elsewhere – based on a rationality that is limited by tradition, social status and excessive optimism or pessimism. Significant fluctuations in markets for stocks and other assets are usually an expression of strong emotional reactions from participants who are either in a good (“bull”) or bad (“bear”) mood. But the fact that bubbles and crashes are difficult to predict with any certainty does not mean that we should ignore the possibility of interpreting the warning signs that exist.

With this in mind, we can return to the current situation and concerns that the enormous attention AI technology has received in such a short time may be a sign that a bubble is about to burst. What is certain, in any case, is that the world’s ten leading AI companies have increased their market value by USD 1 billion in 2025 alone. In the USA in particular, where the so-called “Big Five” AI companies – Google, Microsoft, Amazon, Meta and Apple – are based, they have contributed to perhaps around one percent of the country's GDP in recent years. At the same time, these companies have accounted for the largest part of the stock market rise of around ten percent on the major stock exchanges in the United States during the same period, including Nasdaq, which lists the major technology companies.

Is this worrying? Not necessarily. Many believe that the major AI companies are generating such substantial profits that this could justify a strong upturn in the stock market. This is certainly true, unlike during the IT bubble in 2000, when earnings were extremely weak despite extensive stock market enthusiasm. However, even more people still harbour doubts about the sustainability of the upturn in the long term. One of the lessons learned from other technological advances in history – the introduction of railways, cars, household appliances and, of course, IT – is that enthusiasm levels off when it becomes apparent that the new technology takes a long time to gain acceptance and is not as profitable in the short term as hoped.

When it comes to AI, there is also an additional problem that can send shivers down the spines of investors. The basis for AI’s income boost is largely built on just a small number of companies, maybe five, seven or a few more. Almost all of them are based in the United States. Many of them are characterised by a lack of transparency about how much money they make and how. Added to this is the fact that the political volatility during President Trump’s second term is extensive, which damages companies’ ability to make plans and predict the future.

In other words, it is not easy to answer the question posed. The conditions for a bubble undoubtedly exist – but are they strong enough?

Lars Magnusson, Professor of Economic History, specialising in social and labour history at the Department of Economic History