Artificial intelligence has become one of the dominant investment themes of recent years, driving strong gains across technology markets and reshaping how capital is being deployed globally. While enthusiasm around AI is understandable, recent analysis suggests the sector may now be displaying many of the characteristics historically associated with late stage investment bubbles.

A useful way to assess whether a bubble is forming is to look across four broad indicators: valuation, investor behaviour, capital investment and leverage.

From a valuation perspective, technology shares linked to AI have risen sharply. In past market cycles, major bubbles such as gold in the 1970s or internet stocks in the late 1990s were marked by prices increasing roughly tenfold over a decade or more. US technology stocks have recently approached similar territory, and AI related shares have outperformed the broader market by close to levels that historically increased the probability of a sharp correction. While some argue that AI will deliver stronger productivity gains than previous technologies, current valuations are already close to the highest levels seen in modern market history.

Investor behaviour also suggests growing excess. US households now hold a record share of their wealth in equities, particularly technology stocks, exceeding levels seen during the dotcom boom. Trading activity has surged, with daily share volumes rising sharply and short dated options trading increasingly dominated by retail investors. The most widely held stocks on retail trading platforms are overwhelmingly concentrated in a small group of large technology names. Although there remains significant cash on the sidelines, liquidity has been strong enough to keep pushing markets higher and has, in some cases, forced even cautious institutional investors to remain fully invested.

Capital investment into technology and AI is also at elevated levels. Technology investment now accounts for more than six per cent of US GDP, surpassing the peak reached during the early 2000s. Major technology companies are committing vast sums to data centres, computing infrastructure and energy generation to support AI development. While projected spending continues to grow rapidly, the commercial payoff remains uncertain. Adoption across the broader economy is still limited, and there are emerging signs that implementation may be slower and more complex than initially expected. In addition, large scale labour disruption could prompt political and regulatory responses that affect long term returns.

Leverage, while not yet extreme at the household or corporate level, is increasing in other areas. Several of the largest technology companies now carry more debt than in recent years as spending accelerates. At the same time, leverage is rising within financial markets through the growth of leveraged exchange traded products, which amplify exposure and increase sensitivity to market movements. Government debt also represents a key risk, with ongoing deficits leaving markets more vulnerable to changes in interest rate expectations.

Taken together, these indicators suggest that AI as an investment theme shares many features seen in past bubbles. Importantly, history shows that bubbles do not burst simply because valuations are high or enthusiasm is widespread. They often continue until financial conditions tighten, liquidity dries up or interest rates rise in a sustained way.

For investors, this does not mean avoiding innovation or abandoning technology exposure altogether. Rather, it reinforces the importance of diversification, valuation discipline and risk management. Periods of strong thematic growth can coexist with heightened volatility, and prudent portfolio construction remains essential as markets navigate the balance between genuine long term opportunity and short term excess.