Speculative bubbles can take years to inflate, often in full view of everyone. Is the surge in stock market values linked to artificial intelligence (AI) one of them? If so, will it cause much damage when it bursts? We may not know until it is too late.
For a moment we thought that the rumors of a stock market bubble had just been deflated. Symbol of the AI revolution, the electronic chip giant Nvidia reported high-sounding financial results on Wednesday evening seeming to shut the mouths of the birds of doom who described its sector as a house of cards on the verge of collapsing.
Largely unknown to the general public three years ago, the company recorded a net profit of almost US$32 billion last quarter, an increase of 65% compared to last year. It plans to sell at least 350 billion worth of its processors over the next 14 months.
“We talk a lot about the AI bubble,” said his boss, Jensen Huang, who was heard using the b-word for the first time. But “from where we are, we see something different”.
Nvidia’s impressive results were clearly not enough to reassure the markets.
Down 1.2% over the past three weeks, the company’s stock initially rose, before taking a nosedive on Thursday. The six other giants of the technology sector – Apple, Microsoft, Alphabet (Google), Meta, Amazon and Tesla – which together form what is nicknamed the “Magnificent Seven”, have followed the same trajectory in a stock market generally at half mast.
Same, not the same
“If the United States suffers a crash, it will be one of the most predicted financial implosions in history,” said the British magazine last week. The Economist.
“Stock market veterans often say that it’s impossible to realize that you’re living in a bubble. The irrational nature of a behavior only becomes clear to us in retrospect, when common sense has returned,” noted for his part the Financial Times.
The gap between the value attributed on the stock market to companies specializing in AI and their real annual revenues is not yet as disproportionate as on the eve of the bursting of the technology bubble at the end of the 1990s. But to generate even a 10% return on the approximately 7,000 billion that they intend to invest in the sector by the end of the decade, they would have to draw improbable revenues of 700 billion per year.
It must be said that, although there has been a lot of talk about AI in recent years, its real adoption, particularly by businesses, is still awaited. While ChatGPT alone already has 800 million users around the world, only 3% of AI users pay, according to a recent study by Menlo Ventures.
Less than 10% of U.S. companies with 250 or more employees have integrated AI into their processes, and 95% of those that have tried it have found no benefits, reported The Economist. In Canada, only 2% of businesses saw a return on their investments.
Concentration and circular financing
What is more worrying is the degree of concentration of the current fever, the International Monetary Fund (IMF) warned last month. As during the last tech bubble, the information technology sector currently accounts for around 35% of the main American stock market index, the S&P 500. But this time, we are mainly talking about our “Magnificent Seven”, with 33%.
Until now, we were reassured by the fact that these giants were so rich that they did not even need to resort to debt to build their immense data centers.
But that is starting to change, reported the New York Timesamong other things because new players also want to be part of the party. Of the 3 trillion investments in data centers by 2038, a third will be on credit advanced by banks, investment funds and other kinds of players.
The interrelations between all these people become more complex and confused as the phenomenon extends its ramifications, noted the Wall Street Journal. We have seen, for example, the emergence of circular financing agreements where Nvidia advances to clients, such as OpenAI, Anthropic and CoreWeave, the money necessary to build data centers which will in return use its processors.
And then, it’s not just a question of money. The problem is also finding the electricity needed to run all these energy-intensive machines.
A crash? And after?
And if even part of this building collapsed, would it be so serious?
The AI investment boom and the wealth effect brought by stock markets to American consumers have so far been the two main counterbalances to tariffs, inflation and Donald Trump’s war on immigration, recalls The Economist.
A stock market shock equivalent to that which occurred during the bursting of the techno bubble would this time have a much greater impact, causing American households to lose 20,000 billion, and would subtract at least two percentage points from economic growth, calculated last month Gita Gopinath, Harvard professor and former chief economist of the IMF. The United States could hardly avoid a recession.
The rest of the world would not be spared. Just in stock market values, foreign investors would lose more than 15,000 billion, compared to the equivalent of 4,000 billion during the technology bubble. This would be bad news for other developed economies that are already indebted and weakened.
Although painful, a sudden return to earth of the stock markets, which for several months have been a little too jovial for the IMF’s liking, would obviously not necessarily call into question the AI revolution. Just like the bursting of the tech bubble didn’t kill the Internet. Maybe it will just give us a little more time to prepare for it.
