Artificial intelligence (AI) is rapidly reshaping global markets, driving significant investment into technology stocks, AI infrastructure and data centres. As companies continue spending billions on AI development, many investors are asking the same question: are we witnessing the early stages of a long-term technological revolution, or is the AI sector entering bubble territory similar to the dot-com era?
The honest answer is that it may be both.
The scale of the bet
Start with the raw numbers. According to Morgan Stanley Research, major global technology companies have announced $740 billion in capital expenditure on data centres, chips and AI infrastructure in 2026 alone, an increase of 69% from 2025's record.
At the centre of this sits Nvidia, maker of the specialised chips that power AI systems. Nvidia alone now accounts for around 7% of the S&P 500, which also makes it worth more than the entire UK stock market. Technology now makes up c. 33% of the S&P 500 and c. 26% of MSCI World.¹ This means that investors holding US or global trackers are highly exposed to the AI theme, which would be problematic should the bubble burst.
Does the maths add up?
For all that spending to earn a decent return, research by Panmure Liberum argues that the companies building data centres need to find between $2 trillion and $5 trillion of additional annual revenue by 2030, essentially quadrupling their combined revenue without a proportional rise in costs. No mature industry has ever grown at the 29% compound annual rate currently baked into analyst forecasts for AI infrastructure.
Then there is the awkward matter of the companies supposed to monetise all this computing power. OpenAI, maker of ChatGPT, is forecast to burn roughly $17 billion in 2026, with projected cumulative losses approaching $85 billion by 2028. Its CFO reportedly told colleagues the firm might struggle to pay its committed compute bills. Anthropic, the rival behind Claude, burned around $3 billion in 2025. Both are rushing toward public listings at valuations of approaching $1 trillion, which is a way of transferring risk from the founders to equity investors - many of whom will be private individuals. There is widely reported scepticism regarding the routes to profitability for both firms. A further concern is how the 'hyperscaler' tech stocks are investing tens of billions of dollars into them in return for a commitment to spend enormous sums on cloud services and chips. This circular funding can make revenues look stronger than they actually are and poses systemic risk to the sector should these loops break.
The technology itself also has limits. Large language models have a persistent tendency to 'hallucinate', a problem that academic research has traced to the fundamental way these systems are trained, not something engineers can simply patch out. That matters because much of the hoped-for revenue depends on deploying AI in high-stakes settings such as legal work, accounting and medical decisions, where getting things wrong is expensive. A New York Times experiment found ChatGPT miscalculated tax liabilities by over $2,000 on average.
Déjà vu
We are seeing parallels to the dot-com bubble. The Philadelphia Semiconductor Index rose 48% in April alone, logging 18 consecutive up days, something last seen in February 2000, just before it fell around 80% over the following 18 months. Michael Burry, who famously foresaw the 2008 housing collapse, has said the market is "feeling like the last months of the dot-com bubble."
Another comparison is with the "peak oil" mania of 2006–2008, when investors became convinced that whoever controlled oil supply would enjoy permanent pricing power. Oil surged to $150 a barrel even as economies slid into recession, until demand destruction caused the bubble to burst. The logic driving semiconductor stocks today is very similar: there won't be enough chips for everyone, so prices and margins will stay high regardless of the wider economy. History says that belief tends to be right in the short term and often wrong in the medium term, because capital eventually responds to high margins.
Is there a bubble that is about to burst?
That said, there are arguments to suggest this is not a bubble story. AI is undoubtedly transforming lives and workplaces. In the medium term, it is possible that the companies most exposed to this do start to monetise effectively, as individuals and businesses become willing to pay substantial amounts for what they currently receive for free or very little.
And if we are in a bubble, they also have a habit of outlasting sceptics. Investors warned of "irrational exuberance" in December 1996 but tech stocks did not peak until March 2000.
Implications for investors
AI remains one of the most promising, yet volatile, themes for modern investors. The massive opportunity must be navigated with the risks that come with innovation - this is nothing new; it has been the case with technological revolutions time and time again.
The dominance of technology in global indices means an AI bubble that bursts would really hurt many passive investors, with a significant percentage of many portfolios exposed to stocks that may be in bubble territory. We believe that many sectors set to benefit from AI can still be found at attractive valuations, however. We are seeing efficiency gains across a wide range of industries, from banking to life sciences.
We believe that active management has an important role to play in this environment. Rather than seeking to concentrate exposure in a narrow group of AI leaders, our approach focusses on diversification across the broader AI value chain, including sectors that benefit from efficiency gains rather than just the infrastructure builders; second, disciplined long-term positioning to ensure clients participate in structural growth while managing concentration risk; and third, realistic expectations. We are not promising to call the top of any bubble, but we are working to ensure clients are not overexposed to the most speculative parts of the market.
If you are considering your investment options or would like a second opinion, you may wish to speak with one of our advisers based locally in Longridge, Preston.
¹ S&P 500 Information Technology sector weighting: 32.91%, S&P Dow Jones Indices, as at May 13th 2026 (spglobal.com). MSCI World Information Technology sector weighting: 26%, MSCI World Index factsheet, as at May 13th 2026 (msci.com). Weightings are subject to change.
² Morgan Stanley Research, AI Capex 2026: $740 Billion Signals Bank Tailwinds, 10 March 2026 (morganstanley.com). Figure based on announced capital expenditure guidance from major technology companies.
This article is intended for general information purposes only and does not constitute investment advice, a personal recommendation, or an offer or solicitation to buy or sell any investment. The views expressed are those of the author as at the date of publication and are subject to change without notice. Past performance is not a reliable indicator of future results. The value of investments and the income from them can fall as well as rise, and you may get back less than you originally invested. Any forecasts, projections, or forward-looking statements contained in this article are based on assumptions and estimates. There is no guarantee that such projections will be realised. References to specific securities, sectors, or indices are for illustrative purposes only and do not constitute a recommendation.
