The potential for an AI bubble burst - and what we should do to prepare for it

A trader works at his desk on the floor of the New York Stock Exchange (Photo by ANGELA WEISS/AFP via Getty Images) | AFP via Getty Images

The IMF and Bank of England have joined those warning that the surge in AI-driven tech valuations could end in a painful correction. Many are pointing to another time of heady optimism, the late 1990s, when the internet began to enter everyday life and investors were convinced it would change everything.

When my parents finally got online in 2002, the boom had long since turned to bust, but the internet itself was only just beginning to change daily life. Despite UK equities falling more than 40 per cent after the dot-com bubble burst, the real economy barely flinched. Growth slowed but there was no recession. In the United States, the Nasdaq - heavily weighted towards technology - collapsed by almost 80 per cent, and the economy briefly tipped into recession.

Investors who bought at the peak, however, had to wait 15 years to recover their losses. The technology kept advancing even as markets returned to normality.

Spotting the warning signs

Economists tend to look for the same clues: high valuations, market concentration, circular investment, excessive leverage, and the ever-dangerous belief that this time it’s different.

The most common yardstick is the price-to-earnings ratio, or P/E. It measures a company’s share price relative to its annual profits. Historically, the S&P 500 (an index tracking 500 of the largest listed US firms) has averaged around 15–20, though since 1990 it has been closer to 25. Today it sits around 28, expensive, but not yet outrageous. Crucially, the pressure is concentrated in technology. The so-called ‘big six’ - Nvidia, Microsoft, Apple, Amazon, Alphabet, and Meta - now account for over 30 per cent of the total value of the S&P 500, the highest concentration in decades. Their valuations are lofty: Nvidia trades around 50x earnings, Microsoft and Apple about 40x, Amazon roughly 35x, Alphabet 27x, and Meta 26x.

Investors are betting that the AI revolution will deliver extraordinary gains in productivity and profit. That optimism likely won’t last; when it unwinds, a sharp correction could follow.

And dominance today is no guarantee of future supremacy: few of the biggest tech names in 2000, think Cisco, Intel and Nokia, remain household leaders today.

There are also signs of circular investment. OpenAI buys Nvidia’s chips; Nvidia invests in OpenAI. These loops can inflate valuations and feed on themselves. It’s the digital equivalent of selling houses to one another at ever-higher prices.

contributed

Another warning sign is leverage. High valuations encourage borrowing: firms use debt to fund buybacks or expansion, and investors use margin loans to amplify gains. Margin debt in the US has risen sharply.

When markets fall, leverage ratios spike and forced selling can quickly turn a correction into a broader sell-off.

Scotland’s challenge: build capacity, not hype

Closer to home, Scotland’s technology sector is thriving. It contributes around 4–5 per cent of national output and is expanding faster than the wider economy, healthy by European standards, though well below the scale of the global leaders. The challenge is one of scale and investment intensity.

Scotland has world-class universities and strengths in data science, fintech, health technology and space technology. Yet our venture-capital pool is shallow, and we lack large anchor firms that can sustain investment and innovation when conditions turn. When global valuations tighten, external capital retreats first - as it did after the dot-com crash.

The priority, therefore, should be to build durable capacity, not chase hype. A policy- driven ‘AI rush’, in which funding follows headlines rather than strategy, risks spreading resources too thin. Instead, Scotland should use the current momentum to invest in the foundations that make growth sustainable: compute power and data infrastructure so start-ups and researchers can afford access; connectivity and renewable energy to support data-intensive industries sustainably; and, critically, broad digital literacy and mid-career reskilling.

This isn’t about replicating Silicon Valley. It’s about ensuring that Scottish firms, workers and institutions are prepared for the hard work of adoption once the hype fades. Durable capacity outlasts market cycles; hype-driven expansion rarely does.

Are we in bubble territory?

Today’s environment echoes the late 1990s, though there may still be some way to run. Back then, price-to-earnings ratios for internet firms often exceeded 200, and the Nasdaq as a whole reached 86x earnings before falling by three-quarters. Most of those companies had no profits and business models built on ‘eyeballs’ — the idea that web traffic alone would one day translate into cash.

The AI boom is different. Nvidia, Microsoft, and the rest are immensely profitable and investing heavily in data centres, chips, and cloud infrastructure that will retain value even if share prices correct. If there is a bubble, it has substance beneath the froth. Nvidia estimates that global spending on AI infrastructure could reach $4 trillion by 2030 — more than the UK’s entire annual GDP.

Competition is fierce. Every major firm is racing to make the breakthrough that defines the next decade — to become the monopoly, the new superpower. There will inevitably be winners and losers, and future profits may not accrue to today’s giants.

A bursting AI bubble would not be contained. Those six mega-caps dominate global equity indices and fill pension funds and ETFs worldwide. A sharp correction would ripple through household wealth, credit markets, and currencies. Venture capital would retreat, IPOs would stall, and the funding pipeline for smaller ecosystems, Scotland’s included, would narrow.

The signs of a bubble are there: high valuations, circular investment, record leverage, and claims that ‘this time it’s different.’ Yet the technology is real, and its potential is transformative. Both can be true.

For Scotland, the wise course is to ride the wave of innovation while keeping both feet firmly on the ground. Use the moment to invest in the skills, infrastructure, and institutions that will still matter when the headlines move on. When the correction comes, what endures is not the price of a share, but the capacity left behind.

As for me, I’ll stay diversified. The upside may be smaller, but the crash will hurt less.

Dr Jonathan Swarbrick is a Lecturer in Economics at the University of St Andrews