21st October 2026
Technology shares have powered stock markets higher, but growing doubts about valuations, spending and future profits are raising an uncomfortable question: are investors financing a revolution, or creating another financial bubble?.
For several years, artificial intelligence has been one of the biggest driving forces behind the stock market. Companies developing advanced chips, building data centres and supplying the computing power needed to run AI systems have attracted enormous investment.
American technology giants have committed hundreds of billions of dollars to expanding their AI capabilities. Investors have rewarded many of them with soaring share prices, encouraged by the belief that artificial intelligence will transform business, industry and everyday life.
But the mood is becoming less comfortable. Reports of selling pressure in US technology shares, renewed interest in alternatives among some European investors and growing questions about the returns from AI investment have brought an old financial question back into view.
Are we witnessing a temporary correction in a promising industry, or is the AI investment boom developing into a bubble that could burst?
The answer is not yet clear. However, there are reasons to take the question seriously.
A revolutionary technology does not guarantee successful investments
One of the difficulties in assessing the AI boom is that the technology itself is genuinely impressive. AI systems can assist with programming, research, administration, customer service and the analysis of enormous amounts of information.
Businesses are experimenting with ways to reduce costs, improve productivity and develop new services. There is little reason to believe that AI will simply disappear or become irrelevant.
But history offers an important lesson. A technology can transform the world while many of the companies associated with its early development disappoint investors.
The internet provides a familiar example. During the late 1990s, investors poured money into businesses expected to benefit from the growth of online commerce. Share prices rose dramatically, often on the promise of future profits rather than established earnings.
When confidence collapsed in 2000, many internet companies failed and investors suffered substantial losses. Yet the internet continued to expand, eventually becoming central to the global economy.
The lesson is not that AI must follow the same path. It is that being right about a technology does not necessarily mean being right about the price paid for its shares.
The spending is becoming enormous
The present AI boom is supported by a relatively small group of exceptionally large American technology companies, including Microsoft, Amazon, Alphabet and Meta.
They are spending heavily on data centres, specialist chips, networking equipment and the electricity needed to operate increasingly powerful computing systems.
This spending benefits companies supplying the infrastructure. Nvidia, for example, has become central to the market because of its role in producing advanced AI chips.
But there is a question investors cannot avoid: how much money will customers ultimately pay for AI services, and will those revenues justify the cost of building the infrastructure?
Data centres are expensive to construct and operate. Equipment can become obsolete, electricity demand is substantial, and the technology is evolving rapidly. Companies must keep investing simply to remain competitive.
If AI services generate enough additional revenue and productivity savings, the investment could prove worthwhile. If customers are unwilling to pay enough, or if competition drives prices down, returns may be disappointing.
The danger is that companies collectively invest on the assumption that demand will keep growing rapidly, only to discover that the financial rewards are smaller than expected.
When everyone is investing, who makes the profits?
Another concern is the possibility of excessive duplication.
Technology companies are competing to build their own AI systems and computing infrastructure. Cloud providers are expanding capacity, businesses are buying access to AI tools, and investors are financing new projects.
This creates opportunities, but it also raises the prospect of overcapacity. If too many companies build infrastructure in anticipation of future demand, the industry could eventually find itself with more computing capacity than customers need at profitable prices.
There is also a distinction between the companies selling AI infrastructure and those buying it.
A chip manufacturer can make substantial profits from selling equipment today, even if the customer who purchases it takes years to recover the cost. If customers later reduce their spending, the suppliers could find that an important source of growth has weakened.
The entire chain depends on AI ultimately creating enough economic value to support the investment.
Why are investors looking elsewhere?
Recent reports of pressure on American technology shares and renewed interest in European equities suggest that some investors are reconsidering how much of their money they want concentrated in the US technology sector.
That does not mean Europe is leading a coordinated campaign to abandon American AI companies. Nor does it mean every investor expects a crash.
Investors may be taking profits after substantial share-price increases, reducing exposure to expensive companies or spreading their money across different industries and countries.
European markets also offer businesses in banking, industrial manufacturing, energy and infrastructure, providing alternatives to the technology-heavy American market.
However, European technology companies are not immune to the same pressures. If global investors decide that AI spending has gone too far, the effects could spread well beyond the United States.
The movement of money between markets is therefore worth watching, but it is not proof by itself that a bubble is about to burst.
The warning signs of a bubble
Financial bubbles are difficult to identify with certainty while they are developing. Investors can correctly recognise a major economic change and still push share prices far beyond what future profits can justify.
Several warning signs deserve attention.
The first is valuation. If share prices rise much faster than the profits companies are actually generating, investors increasingly depend on expectations of exceptional future growth.
The second is the scale of spending. When companies must continually increase investment to satisfy market expectations, even strong businesses can disappoint if growth slows.
The third is the availability of finance. If AI projects increasingly depend on borrowing, outside investors or complex financial arrangements, rising interest rates or a loss of confidence can make expansion more difficult.
Finally, there is the question of customers. Exciting demonstrations and widespread experimentation are not the same as businesses paying enough for AI services to make them consistently profitable.
None of these factors proves that a crash is imminent. Together, however, they provide a framework for judging whether enthusiasm is becoming excessive.
What would cause the bubble to burst?
A sudden collapse would not necessarily require AI technology to fail.
It could begin with disappointing results from a major technology company, evidence that customers are spending less than expected, or an announcement that a large data-centre project has been delayed or cancelled.
Investors might then reconsider the expected profits of companies across the sector. Falling share prices could prompt further selling, while companies facing weaker valuations might find it harder to finance expansion.
A decline in the shares of major technology companies could also affect broad stock market indices, pension funds and investment funds around the world.
There is a further complication. Some companies have become so important to stock market performance that their fortunes influence the wider market. If a handful of large firms account for a substantial proportion of an index's value, their decline can pull down the overall index even when other businesses are performing reasonably well.
However, a sharp fall in share prices would not automatically mean that banks or the wider financial system were facing a crisis. The consequences would depend on the scale of the decline, the amount of debt involved and how financial institutions were exposed.
What would it mean for Britain?
The UK may not dominate the development of the largest AI models, but British savers and investors are connected to the sector through pension funds, investment funds and global stock market indices.
A sustained decline in US technology shares could reduce the value of some pension investments and household portfolios. It could also affect business confidence and the willingness of investors to finance new projects.
There could be consequences for the wider economy if companies cut investment, lenders become more cautious or falling share prices undermine confidence.
For Scotland and the Highlands, the effects would be indirect but still relevant. Businesses depend on a healthy economy, while public finances can be affected by changes in growth, tax receipts and investment.
There could also be opportunities. If the AI industry continues to expand but becomes more disciplined about costs, demand for reliable electricity, engineering, specialist skills and digital infrastructure may remain strong. The important question is whether projects can deliver lasting economic benefits rather than simply attracting money during a period of enthusiasm.
A correction may be more likely than the disappearance of AI
There is a danger in treating every fall in technology shares as evidence that a bubble has burst. Markets rise and fall, and investors regularly reassess the value of companies after periods of rapid growth.
AI could continue to expand while some of the companies associated with it experience substantial share-price declines. The technology's long-term success and the short-term fortunes of its investors are separate questions.
Equally, it would be unwise to assume that the largest technology companies are immune to financial disappointment simply because they have considerable resources and established businesses.
The crucial test is whether the revenue and productivity improvements generated by AI eventually justify the cost of developing, operating and financing it.
If they do, today's investment may prove to have laid the foundations for a new period of economic growth. If they do not, investors may discover that they have paid too much for the promise of future profits.
For now, the AI boom remains a mixture of genuine technological progress, ambitious business investment and considerable financial expectation.
The bubble may not burst tomorrow, and it may not burst at all in the dramatic way some fear. But the growing questions about spending, valuations and returns suggest that investors are beginning to demand more than promises.
The real test for the AI revolution is no longer just what the technology can do. It is whether it can make enough money to justify the extraordinary sums being spent on it.