18th September 2026
Artificial intelligence has become one of the biggest stories in the stock market.
Companies involved in AI have reached extraordinary valuations, with investors betting that the technology will transform business and generate enormous profits for years to come.
But there is now a growing question hanging over markets: what happens if the expected growth does not materialise quickly enough?
That does not necessarily mean AI itself is a failure. The technology can continue to transform the economy while the shares of companies associated with it fall sharply.
This distinction is important.
The problem for investors is the price they are paying today for profits they expect tomorrow.
A very concentrated market
The American stock market has become unusually dependent on a relatively small number of enormous technology companies.
The Bank of England reported that AI-related companies accounted for 44% of US equity market capitalisation and 67% of the S&P 500's return in 2025 at the time of its analysis.
Goldman Sachs has also highlighted the concentration, estimating that the ten largest companies account for around 36.5% of the S&P 500.
That creates an obvious vulnerability.
If investors suddenly become less enthusiastic about AI, they are not selling a small corner of the stock market. They are selling some of the largest companies in the world.
And because those companies have such a large weighting in major investment funds and stock-market indices, the effects can spread much further.
What could trigger a fall?
It would not necessarily require an AI disaster.
A slowdown in the enormous spending on data centres and computer chips could be enough.
Companies such as Microsoft, Amazon, Alphabet and Meta are spending vast sums building AI infrastructure. Investors are expecting those investments eventually to produce much larger revenues and profits.
If businesses begin questioning whether they are getting enough benefit from AI to justify the cost, spending could slow.
That would hit the companies supplying the technology first. But then investors could start questioning the valuations of the entire sector.
This is where a correction could become self-reinforcing.
A falling share price makes investors nervous. Nervous investors sell. Falling prices make other investors question whether the boom has ended, producing more selling.
Would the whole stock market crash?[b]
Not necessarily.
There is a major difference between an AI share-price correction and a financial crisis like 2008.
Today's leading technology companies are generally highly profitable businesses with substantial cash flows. That is one of the reasons some analysts argue that comparisons with the dot-com crash are exaggerated. Reuters recently noted that today's leading technology companies are much more profitable than many of the businesses involved in the 1999-2000 bubble.
But the concentration means that a major fall in the biggest technology companies could still produce a substantial fall in the wider stock market.
Pension funds, investment funds and ordinary savers holding index funds would not escape simply because they did not buy individual AI shares.
[b]And then comes the economy
This is where an AI correction could become more serious.
If technology companies suddenly cut investment, construction of data centres could slow. Orders for chips and equipment could fall. Companies supplying electricity, cooling systems, networking equipment and construction services could also feel the effect.
Share prices could fall at the same time as business investment weakens.
That could affect economic growth.
The International Monetary Fund has already identified high valuations and concentration in AI-related companies as potential sources of downside risk to financial stability. It has also warned that market stress can be amplified by leverage and forced selling.
But there is another possibility
The AI boom does not have to end with a spectacular crash. Markets can also correct gradually. If company profits continue rising but share prices stop climbing, valuations can come down over several years without a financial catastrophe.
Indeed, there are already signs that investors are becoming more selective. Morningstar noted that a June 2026 sell-off had exposed the risks created by high valuations and increasing market concentration.
That may ultimately be healthier than a sudden collapse.
The real danger
The biggest danger is not that artificial intelligence turns out to be useless. It is that AI becomes enormously important to the economy but investors have already priced in too much of its future success. If that happens, the technology can continue advancing while its shares fall.
And because AI-related companies have become such a large part of the world's biggest stock markets, an AI correction would probably be felt far beyond Silicon Valley.
For British investors, including those whose pensions are invested globally, this matters.
The next stock-market shock may not come from companies going bust. It could come from investors simply deciding that the future profits they have been paying for are worth rather less than they thought.
That is how a boom can turn into a bubble, even when the underlying technology is real.