Since the launch of Chat GPT in early 2023, AI stocks have become the only game in town. The promise of a new era of innovation and efficiency gains have helped drive the S&P 500 to successive record highs throughout 2024. Leading the charge has been Nvidia (NASDAQ: NVDA). But with its stock now trading at 36 times sales and 64 times its earnings, I am beginning to wonder whether it can ever grow into such a lofty valuation.

Bubble characteristics

One of the hallmarks of any bubble is the idea of the greater fool theory. When investors start to believe that the only thing that matters is a ticker’s price, then alarm bells should be sounding.

The tulip mania in the 17th century is a classic lesson of what can happen when individuals ditch fundamental investing principles in favour of speculation and greed.

Another major characteristic of a bubble is that it sucks everyone in. Following the collapse of the South Sea bubble of 1720, Sir Isaac Newton famously said: “I can calculate the motion of heavenly bodies, but not the madness of people.”

Tech bubble

The dot.com bubble of the late 1990s is a modern day example of what can happen when stock prices diverge from underlying fundamentals.

The release of the Netscape browser in 1994 heralded the dawn of the Internet. Within five years, companies were going public with nothing more than a PowerPoint presentation and a URL.

One fact about the evolution of the Internet that is completely lost today, is that the companies who helped build it, did not end up being the eventual winners. The likes of Vodafone and Cisco succumbed to a new breed of business models led by the likes of Alphabet, Apple, and Meta.

Today is different

The question for investors today is simple: are the early days of the AI revolution any different from previous manias? I don’t believe it is.

Nvidia, and the hyperscalers reliant on their chips, such as Google and Microsoft, are highly profitable businesses, something very few companies were back in 2000. But then again even a great company can make a bad investment. Even 25 years later, neither Vodafone nor Cisco have surpassed their mania highs.

In the early days of a new technology, one would expect to see an explosion in startups. But what is happening instead is that AI is becoming more and more centralised. The vast majority of Nvidia’s revenues are coming from a few mega-cap companies within the Magnificent Seven.

Despite investing tens of billions of dollars each in Nvidia chips, Google, Microsoft, and Meta are yet to see a return on investment. Will they ever?

Let me hypothesise the unthinkable. What if the generative AI models of today are pretty much the peak of a hype cycle? One fact is indisputable: no one killer app has emerged. Indeed, I am beginning to question whether large language models are even a real form of AI.

I don’t know how the future of AI plays out. I do believe that AI will be transformative in the same way as the internet was. But I am not willing to make a bet on Nvidia being at the centre of it; not at its present valuation, anyway.