If AI looks like a bubble, acts like a bubble, then…

IN JULY 2007, just before the global financial crisis erupted, Citigroup chief executive Chuck Prince famously remarked: “As long as the music is playing, you’ve got to get up and dance.”

His words have since become one of the defining quotations of every speculative boom: where investors may recognise that stock valuations are becoming absurdly inflated, but few are willing to leave the dance floor while prices continue to rise.

That sentiment feels strikingly relevant today.

Artificial intelligence has become the dominant investment theme of the decade, with investors paying little attention to valuations.

For instance, financial tracker website Investing.com places SpaceX’s trailing price/earnings ratio at about -51, which means the company is trading at a premium even though it is making a net loss currently.

Because the company has no profits, many analysts evaluate SpaceX using its price/sales ratio, which sits at a roughly astronomical – pun intended – 138.

Meanwhile, hundreds of billions of dollars are being committed to AI chips, cloud computing, data centres and electricity infrastructure.

Investors are said to have embraced the view that AI will transform virtually every sector of the global economy.

Even though they are probably right, it is precisely this belief that makes the comparison with the dot-com boom and bust of 2000 both fascinating and uncomfortable.

Parallels to the dot-com bubble

For those who can recall it, the Internet revolution of the late 1990s genuinely changed the world by transforming communication, retailing, entertainment and finance.

Investors correctly identified one of the greatest technological shifts in history, as terms like “business-to-business” and “business-to-consumer” were coined to sell the idea of fabulous future riches.

However, what they got wrong was the actual worth of Internet companies at that time. Many had little more than a website, an ambitious business plan and a seemingly plausible growth story.

The problem was that these firms’ revenues were negligible, profits non-existent and valuations detached from reality. When expectations eventually collided with commercial reality, the dot-com bubble burst spectacularly.

To be fair, today’s AI leaders are fundamentally different.

SpaceX apart, Nvidia, Microsoft, Alphabet, Amazon and Meta are immensely profitable businesses with solid balance sheets and dominant market positions.

Furthermore, unlike companies in the dot-com era, today’s tech giants are financing their AI ambitions from real cash flows rather than investor optimism alone.

However, even though the difference is significant, it does not eliminate the possibility of a bubble. History shows that bubbles arise when investors extrapolate genuine success too far into the future.

Technological innovations such as railways, electricity and the Internet all transformed society, yet each experienced periods when their market valuations became impossible to justify.

The question is not whether AI will succeed in reshaping the way we live, but whether today’s shares are pricing in perfection for too many companies in the sector.

Investors are increasingly rewarding almost any company with a convincing AI narrative. Traditional valuation measures are frequently dismissed because of the oft-repeated phrase in markets: “This time is different”.

Fear of missing out has become every bit as powerful as disciplined analysis.

The lesson from 2000 remains clear – investors who had bought the Nasdaq-100 at its peak were right about the Internet, but were wrong about the price.

The truth is, no one knows when today’s enthusiasm will fade. No one knows what event may eventually trigger a correction. As Chuck Prince discovered, the music can continue longer than many expect. But it does eventually stop.

After all, if it looks like a bubble, acts like a bubble and is driven by the same psychology that inflated every previous speculative boom, then it probably is a bubble.