When investors start reaching for metaphors from the late 1990s, it is rarely a sign of calm thinking. The torrent of money pouring into artificial intelligence (AI) has invited comparisons with the dotcom boom, an era when “new economy” firms soared on the promise of the internet, only to crash spectacularly in 2000-01.
As valuations wobble after a year of giddy gains, investors could be forgiven for wondering whether history is about to rhyme once again.
Yet those who remember the dotcom years argue that today’s AI frenzy, while exuberant, is built on firmer ground. The Association of Investment Companies (AIC) recently canvassed fund managers who lived through the first internet bubble. Their consensus: AI is not immune to hype, but the foundations are stronger and the risks, though real, are better understood.
Annabel Brodie-Smith of the AIC notes that today’s mania differs in a crucial way. “The companies behind the boom are well established and highly profitable,” she says. OpenAI’s $38bn deal with Amazon is emblematic: vast sums, yes, but anchored by corporations with balance sheets that can bear them. Still, she cautions, no one knows how long the excitement can last. “Maintaining a diversified portfolio is key,” she adds
Not everyone is convinced that the word “bubble” applies at all. Craig Baker of Alliance Witan [LON:ALW] points out that while some smaller firms have been buoyed by investors’ enthusiasm for all things AI, many of the sector’s heavyweights, including Microsoft, Nvidia, and Alphabet, have genuine earnings to match their valuations. “But even if there’s enough evidence to worry about being in a bubble, there’s not enough to suggest it’s about to burst,” he says.
Others, such as Mike Seidenberg of Allianz Technology Trust [LON:ATT], argue that the scale of corporate investment from Meta and Google suggests the demand for AI services is still in its infancy. “Most of these companies are building these services for contracted revenue and not future capacity.” he explains.
Laying the foundations for the future
Still, markets can overreach even when the underlying technology is transformative. The late 1990s internet buildout produced overcapacity and ruinous losses, but it also laid the groundwork for the digital economy that followed.
Ben Rogoff of Polar Capital Technology Trust [LON:PCT] thinks the parallel is apt. “The dotcom infrastructure build is probably the best (and most recent) parallel for today’s cycle,” he says. “A profound new technology that’s likely to change the world cannot be supported with existing infrastructure. Then, circuit-switched networks; today, CPU-based computing power.”.
Then it was circuit-switched networks; today, CPU-based computing. The difference is speed. “The AI revolution is likely to unfold far faster than the internet because it builds upon earlier advances: the smartphone, the cloud, and the prior digitisation of knowledge.”
- Seeing Machines lands breakthrough North American deal
- MedPal AI wins NHS pharmacy approval
- Is it time for investors to go big on Japan?
Unlike the days when pets.com floated on little more than a sock puppet, most of today’s AI champions are cash-generating giants. Richard Clode of Bankers Investment Trust [LON:BNKR] points out that valuations, while lofty, are “dramatically lower than the dotcom era” relative to profits. Moreover, the market has not yet seen the excesses of 1999, when dozens of firms rose by 1,000% and hundreds rushed to IPO. Caution, it seems, has not been entirely discarded.
Lessons from that era still resonate. Paul Niven of F&C Investment Trust [LON:FCIT] lists the warning signs that preceded the 2000 crash: stretched valuations, tightening monetary policy, and speculative financing. Today, he says, some of those conditions are absent. Capex is rising, but “largely financed from cash.” There are pockets of speculation, but no full-blown mania, at least not yet. Markets, however, have a habit of staying irrational longer than investors stay solvent.
Even if AI’s promise is real, the path will be uneven. Rogoff warns that “volatility is endemic to new cycles.” Between 1995 and 1998, before the dotcom “melt-up”, the Nasdaq endured seven corrections over 15%, and still rose 350%. Seidenberg agrees: “Our job is to navigate the trust through the ups and downs.”
Don’t look for “straight line” returns from AI
The larger risk, some argue, lies not in hype but in the technology’s trajectory. A slowdown in AI model progress, or breakthroughs that cut computing demand, could upend growth assumptions. Others fret about leverage: if companies borrow heavily to fund AI infrastructure, a credit squeeze could puncture valuations faster than any collapse in enthusiasm.
The fund managers’ parting advice is characteristically pragmatic. Diversify. Keep scepticism handy. And remember that even transformative technologies rarely deliver returns in a straight line. The internet bubble ended in tears, but a trillion-dollar industry rose from the wreckage. AI may yet prove a similar story, less a bubble bursting than a long, noisy adolescence on the way to maturity.
Comments (0)