Artificial intelligence – bubble or balloon?
By, Aleeshia Naicker – Senior Investment Consultant
Since 2020, the surge of interest in artificial intelligence (AI) has marked the beginning of a new technological era, transforming industries like education, communications and healthcare services.
At the heart of this revolution are large language models (LLMs) and generative AI systems, which demand immense computing power, specialised graphics chips, and vast amounts of energy. This unprecedented growth has fuelled soaring stock prices for AI-focused companies, sparking concerns about speculative bubbles. How much further can AI stock prices go in the current environment? Are we in a bubble? Or is there still some elasticity to grow?
What is a bubble?
An investment bubble is when the prices of assets rise far above their real economic value because investors keep buying simply because they expect prices to keep going up, and not because the assets are truly worth more. An example of this is the dot-com bubble of the late 1990s, when excitement about the internet led investors to pour money into tech companies with little revenue or proven business models, driving share prices to unrealistic levels, until the bubble burst in 2000 and prices collapsed.
What is the root of the current fears?
The current market environment bears a stark resemblance to that of the dot-com bubble.
The dot-com bubble was built on the belief that the internet would transform everything instantly. Today, it’s the belief that AI will reshape the entire economy at unprecedented speed.
In the late 1990s, Cisco, Intel, Microsoft and a few others drove most of the market’s gains. Today, companies like Nvidia, Microsoft, Amazon and a small cluster of AI-heavy names dominate index returns. The top 10 companies today make up over 40% of the entire S&P 500 Index, meaning a small group of mega-cap companies are carrying most of the market’s weight.
Both periods show stretched price-to-earnings (P/E) ratios and markets priced far above long-term averages. The S&P 500’s forward P/E ratio (at the time of writing) is around 23x, which is more than two standard deviations above its 10-year average. When we remove the “AI 9” (Nvidia, Microsoft, Apple, Amazon, Meta, Alphabet, Broadcom, Oracle and Palantir) from the index, the forward P/E ratio drops to one standard deviation below the 10-year average, suggesting valuations of the AI 9 are stretched and driving the index.
Large tech firms are increasingly investing in one another or structuring compute-for-equity deals and cloud credits for equity, or creating complex cross-exposures that raise systemic risk. In the 1990s, venture capitals, investment banks and public markets reinforced each other’s valuations. Today, these circular megadeals can inflate valuations without underlying economic performance.
But it’s not all bad news. It could be argued that, just like the internet was not just a fad but a structural revolution, AI could be a similarly foundational shift. While many internet-linked businesses failed throughout the 2000s, survivors like Amazon, eBay and Google created outsized value.
The applications of AI are broad, giving it the potential to reshape multiple industries. So even if we are in bubble territory, the underlying technological transformation could add significant social and economic value. Major AI companies like Microsoft already have real revenue and business models. More so, we have already seen businesses adopting AI for productivity, cost savings and innovation. The presence of real, profitable AI businesses reduces the risk that the boom is pure speculation. Even if overhyped players fail, established companies may anchor sustainable value creation.
How do we select stocks or assets that can withstand the potential burst?
Speculative assets like tech equities or high-yield debt surge during bubbles but fall the hardest afterwards. Defensive assets like government bonds, cash and certain dividend stocks tend to preserve capital and recover faster.
Prioritise companies that benefit from AI adoption like those using the technology to reduce costs or expand margins. Build exposure across the ecosystem – from infrastructure providers, such as cloud computing and semiconductor leaders, to software firms with durable recurring revenues – while avoiding concentrated bets on businesses driven only by hype or narrative. In this environment, the most important investment pillars are diversification and selective stock picking.
25 February 2026
Sources: Bloomberg, GQG, Amplify Investment Partners, Laurium Capital, Bloomberg, October 2025, FactSet, Standard & Poor’s, J.P. Morgan Asset Management, September 2025, Refinitiv, Jeff Weniger, October 2025