There is growing talk in investment markets that the AI “bubble” may soon burst. However, evidence suggests those predictions are misplaced, although investors should remain cautious about risks.

In market terms, a bubble refers to irrational pricing. It happens when hype overshadows reason. Prices reflect investors’ urgency to buy because they fear missing out. This hype drives prices even higher, reinforcing positive sentiment in a self-perpetuating loop. Bubbles occur when asset prices far exceed sustainable value. Previous examples include tulip prices in the 1600s and tech stocks in the early 2000s.

Some analysts believe this is now happening with artificial intelligence (AI) shares. I am not convinced – as it is hard to foresee a world where AI becomes less relevant in our daily lives.

There have been many examples where analysts suggested stocks were irrationally overbought for years until they accepted higher prices as normal – think Naspers or Capitec a decade ago. Often in those periods analysts fail to appreciate that the market valued a stock or sector based on future potential earnings that had not yet materialised. Mr Market is, by and large, remarkably efficient at pricing assets and predicting trends.

The 2000 dot-com crash serves as a lesson from the past that informs the current market. In the early 2000s, there were naysayers who wrote the obituary for a tech industry that – at the time – looked like it had died before it had even matured. A few analysts were saying, “We told you so. This was all hype, all bubble, no substance.” But hindsight shows us that the market was not irrational in valuing highly the companies that would ultimately benefit from widespread Internet adoption.

Instead, the dot-com crash was simply a case of not all tech companies becoming winners. There’s a lesson there for today’s AI companies and today’s investors.

Markets tend to be remarkably resilient and efficient over time. The dot-com crash simply preceded an era of enormous stock market growth. Many of the companies that succeeded in the Internet age drove this. Were there failures? Of course. After the correction, many analysts pointed to the irrational behaviour of companies that were too eager to build the Internet’s infrastructure. This included laying the same fragile fibre-optic undersea cables that now enable our global connectivity. At the time, that infrastructure investment might have looked excessive. In hindsight, it proved essential. The rapid growth of AI may follow a similar path, but unlike the early internet, it will depend heavily on the infrastructure required to support it at scale. 

The long-term market correction that followed the 2000s dot-com bubble highlighted the importance of staying calm and avoiding panic selling. It also showed why a diversified and disciplined risk management approach will always beat jumping onto investment bandwagons or trying to “time the market”.

Investors should be cautious yet open-minded about the current AI bull run. Will there be pain from AI? Yes. Some companies will disappoint. Are valuations stretched? I would agree, but traditional accounting isn’t great at measuring technology company value.

Like AI, other assets – such as cryptocurrencies and commodities – also face “bubble” warnings. But labelling everything a bubble is not helpful. It simply creates fear among investors. They then see those industries or stocks as irrationally priced. This affects their behaviour, and so they stay on the sidelines.

Unsurprisingly, bubbles and subsequent busts get a bad rap. But arguably the optimal amount of bubbles and busts over time is not zero. Society needs them to occur. If you consider truly societal game-changing technologies – railways, air travel, the Internet – it’s clear that we need periods where people lose their short-term sensibilities, so that the long-term infrastructure can be built that moves societies forward.

So instead of worrying about bubbles, investors should take a pragmatic, long-term view of the market. AI stocks may look expensive today based on fundamentals. But how relevant are those current fundamentals over the long term? Many of these companies are building infrastructure for tomorrow’s AI-powered world – not just digital platforms, but the underlying systems needed to make them work in the real world.

What is often overlooked in this discussion is that AI is not just a digital story, it is also a physical one. Like the Internet era before it, which required massive investment in fibre-optic networks and data infrastructure, AI’s expansion depends on reliable and scalable energy supply. Data centres, cloud computing and advanced chip manufacturing are all highly energy-intensive.

Recent disruptions in global energy markets have shown how quickly constraints in supply can ripple across industries and economies. While much of the current debate focuses on valuations, what happens next for AI will increasingly depend on whether energy is available, reliable and affordable.

The AI rollout assumes that data centres will have a stable, continuous supply of energy to keep these facilities cooled and running efficiently. Any serious energy disruption could prove disastrous for AI – and for markets in general. Market shocks, like the current oil price shock, can have far-reaching knock-on effects. For investors, this reinforces the importance of diversification – especially when outcomes are harder to predict.

We believe that investment in companies developing tomorrow’s AI infrastructure is still a sound move, but as with previous technological shifts, the winners will not only be defined by innovation but by their ability to operate in practice, including the infrastructure that powers it.

Keen for more? Listen to the recent podcast on this topic here:

Disclaimer

Satrix consists of the following authorised Financial Services Providers: Satrix Managers (RF) (Pty) Ltd and Satrix Investments (Pty) Ltd. The information does not constitute financial advice. While every effort has been made to ensure the reasonableness and accuracy of the information contained in this document (“the information”), the FSPs, their shareholders, subsidiaries, clients, agents, officers and employees do not make any representations or warranties regarding the accuracy or suitability of the information and shall not be held responsible and disclaim all liability for any loss, liability and damage whatsoever suffered as a result of or which may be attributable, directly or indirectly, to any use of or reliance upon the information.