By Eman Nazar
Almost two decades after the dot-com crash wiped trillions from global equity markets, a familiar anxiety is setting over the trading floors of NYSE and LSE. Investors are once again debating whether technological enthusiasm has outpaced economic reality.
The rapid ascent of artificial intelligence has propelled a market rally concentrated in a strikingly small number of stocks, echoing the late 1990s dot-com era when a handful of technology firms dominated market gains. At the time, companies like Cisco Systems (a networking hardware giant that became one of the most valuable firms globally during the internet boom) were seen as indispensable to the digital future, attracting outsized investor enthusiasm and driving index performance. Today’s AI leaders occupy a similar position, raising questions about whether this concentration reflects enduring structural change or speculative excess. The Shiller CAPE ratio – which smooths year-to-year profit swings by averaging earnings over a decade – tells a troubling story: US stock valuations are now more expensive than before the 1929 Wall Street Crash, and the only time they were more stretched in recent history was during the dot-com bubble of 1999.
Yet, the macroeconomic environment differs fundamentally from 2000. In that year, the bubble burst just as the Federal Reserve was aggressively raising rates to cool an overheating economy. Today, markets seem to be navigating a ‘higher-for-longer’ interest rate environment which is supported by a resilient labour market and massive private capital investment. The Bank of England has nonetheless warned that the risk of a ‘sharp market correction’ has increased, stating that global markets could tumble if investor sentiment sours on AI’s prospects.
To further understand the debate, one must look at the ‘pick and shovels’ providers of both eras – companies that supply the core infrastructure behind a technological boom rather than the end products themselves. In 2000, Cisco Systems played this role, acting as the “shovel” provider by supplying the routers that directed internet traffic and enabled data to flow across the web for dot-com companies (the “gold seekers”). At its peak, Cisco traded at a Price-to-Earnings (P/E) multiple – the price investors pay for every dollar of profit of over 150. Today, Nvidia holds a similar position as the “shovel” provider in the AI boom, producing the chips that power AI applications. Yet, despite the stock’s meteoric rise, Nvidia’s P/E ratio has typically hovered in the range of 30 to 40, creating a striking paradox: while its stock price has surged, Nvidia generates profits at a scale Cisco never approached, suggesting far more robust underlying earnings power.
This leads to the central tension of the current boom: the explosion in capital expenditure, or ‘CapEx’ as tech giants pour billions into data centres and AI infrastructure. This echoes the fibre-optic buildout of the late 1990s, when telecom companies laid vast amounts of cable in anticipation of future demand. The difference is that fibre was a passive utility, while AI data centres are active processing hubs. For sceptics like Jim Covellom, Goldman Sachs’ , Head of Global Equity Research, the key issue is return on investment: high upfront costs, rapid depreciation, and unclear revenue visibility make it difficult to know whether this spending will generate durable profits.
Jeff Bezos has made a similar point, saying that in periods of intense excitement investors often struggle to distinguish between good ideas and bad ones.
The scale of today’s investment frenzy dwarfs its predecessor. Venture Capitalists invested roughly $10.5bn into internet companies in 2000 – around $20bn in today’s money. In contrast, investors are on course to spend well over $200bn investing in AI stocks this year alone, with ten AI start-ups gaining close to $1 trillion in valuation in over just twelve months. More optimistic voices like Joseph Briggs of Goldman Sachs, suggest AI could eventually automate 25% of the work tasks and boost global GDP by 7% – a prize that would justify much of the current spending if delivered.
The divide has created a bifurcated market. Investors are rewarding companies with tangible technology moats: firms like Nvidia, TSMC, and ASML, which possess specialized hardware advantages and report record profits. The speculative fringe is a different matter. As Brooke Masters of the Financial Times has observed, “We tend to overestimate the short-term impact of brand-new technology and underestimate the long-term impact””- not every company investing in AI will be a winner, but the technology itself will most certainly prove transformative.
The strongest contrast to 2001 lies in the financial health of today’s speculative fringe. Pets.com – the poster child of the earlier bubble had no path to profit and collapsed just a month after its IPO. Today’s more volatile names, like C3.ao or CoreWeave, at least provide high-demand infrastructure services rather than selling consumer goods at a loss. Historical analysis suggests that technology booms which eventually prove transformative are routinely accompanied by speculative excess that shakes out weaker players while those with stronger foundations endure. After the dot-com burst, it took the NASDAQ 16 years to make a lasting break back above its previous peak, a sobering reminder of how long the ‘hangover’ can last even when the underlying technology ultimately succeeds.
Some things never change. The rise of ‘AI washing’ – where companies claim to be AI-powered simply to inflate their share price – is one of them. The SEC recently charged investment advisers Delphia and Global Predictions for making false and misleading claims about their AI capabilities. These cases illustrate the perennial peril of hype outrunning reality.
Some analysts also point to ‘circular financing’ as a structural risk: hardware giants invest in AI startups that then use that capital to buy the same giant’s chips, creating a ‘tangled spaghetti’ of demand in financial statement. Whether this represents genuine end-user adoption, or an elaborate hall of mirror remains a pressing and unanswered questions. Bubble indicators examined by Financial Times suggests that anxiety was also growing before the dot-com crash, a year prior, the San Francisco Fed had raised the spectre of the 1929 stock market crash and many economists echoed those fears in the same manner seen today.
Whether the current market is a bubble or a breakthrough remains fiercely contested. Valuation multiples are elevated but generally remain well below the triple-digit peaks of 2000. The companies at the centre of this rally generate real revenues and real profits – a luxury their dot-com predecessors rarely enjoyed. Expert opinion is itself bifurcated: fund managers tend to be more optimistic than journalists, while most sceptical voices sit somewhere in between. The world is witnessing a massive, capital-intensive bet on the future of human productivity. For the rally to sustain itself, the intelligence being bought today must finally deliver the revenues promised for tomorrow.
The views expressed in this article are the author’s own, and may not reflect the opinions of The St Andrews Economist
Image Credits: Wikipedia
