AI has quickly gone from tech buzzword to investment obsession, with “the AI trade” now shorthand for backing the companies most likely to benefit from its growth.
But before you assume today’s obvious winners will still look obvious in a decade, it is worth revisiting the last time a world-changing technology captivated markets. The dot-com era offers a surprisingly useful lens for thinking about what comes next, and what tends to go wrong when investors get ahead of themselves.
What is “the AI Trade”?
Artificial intelligence has been a buzzword for a few years now, and it has become a staple topic in investing. The “AI trade” is shorthand for investing in a way that aims to benefit from the growth of AI. That can include the obvious candidates – companies building AI models, software, and hardware, but it also extends to the enablers and downstream beneficiaries: semiconductors, data centres, power infrastructure, copper and other critical minerals, and the businesses that supply them.
The investment debate tends to split into two questions.
First, which companies are most likely to benefit as AI adoption accelerates, and what is the most sensible way to identify them?
Second, which companies are likely to be disrupted by AI – where AI cannibalises their product or service – and how do we spot those risks early?
Can we learn anything from the dot-com bubble?
Many people reading this blog will not remember the dot-com bubble. Fortunately (or perhaps unfortunately 😩), I am old enough to remember it first-hand.
The dot-com bubble started to build in the mid-1990s, as the internet moved from novelty to commercial reality. Netscape’s IPO in 1995 is often cited as an early catalyst (Netscape was a web browser product), and in hindsight it marked the point where excitement about “the internet” began to translate into extreme optimism.
The core investment thesis was not wrong: the internet would transform existing businesses and create entirely new ones, with the potential to scale far faster than traditional companies through a global reach. The problem was how the market chose to value that promise. Attention shifted away from business fundamentals and towards proxies such as website traffic, “eyeballs”, and vague notions of first-mover advantage.
By around 2000, the bubble had reached its peak. I saw the disconnect from fundamentals up close while working in corporate finance at Deloitte at the time, only a few years before I started ProSolution. Valuations became detached from economic reality, and some companies attracting venture capital or even listing on stock exchanges were pre-revenue! In other words, they were not really businesses yet – they were ideas being priced as if success was already guaranteed!
There are clear parallels with AI today. Investors are again trying to work out how a powerful new technology (AI) will ultimately translate into economic value.
Who were the dot-com bubble darlings and where are they now?
The table below highlights eight companies that were widely considered “market darlings” during the dot-com boom. It compares each company’s peak market capitalisation, most of which peaked in the first half of calendar 2000, with its global ranking at the start of 2000, to show that these are not cherry-picked examples.
The data in this table is compelling. In year 2000, around 6 technology companies made it into the ten most valuable companies in the world and all except one ultimately failed to deliver long-term value to shareholders. The only clear winner was Microsoft. Therefore, if you invested equally in the tech market darlings, your investment returns would have been very poor.
Australian share market investors had a similar experience. Dot-com darlings on the ASX included companies like One.Tel, Davnet, Solution 6, eCorp, and REA Group. Except for REA Group, all these companies ultimately failed to deliver long-term value to shareholders, so the win-loss ratio was similar to the global experience.
AI is probably the biggest tech advancement in my lifetime
I hope the above sub-headline doesn’t come across as hyperbole.
I suspect AI will be the most significant technological shift I experience in my lifetime. Don’t forget, when I started my career, everyday tools we now take for granted like the internet and email didn’t exist. I have seen a lot of change over the past two to three decades, but AI feels different in potential impact and how wide that impact will be.
What I am confident about is that many of the long-term winners probably do not even exist yet. The most valuable businesses are likely to be those that are built on capabilities we cannot fully anticipate today, because the technology is evolving so quickly. Some estimates suggest large language models like Chat GPT are improving at a rate that roughly “doubles” in capability every 7 months, although the precise pace is hard to measure and will not be linear.
AI is also likely to be broad-based rather than confined to a single sector. I think it will touch almost every industry and geography – healthcare, finance, logistics, manufacturing, and plenty more. Which means, to some extent, investors already get meaningful exposure to AI’s long-term impact simply by owning a broad index, because the beneficiaries (and the disrupted) will increasingly be represented across the market.
But you don’t need to work it out
The dot-com bubble is a useful reminder that most of today’s stock market darlings, such as the Magnificent Seven, are unlikely to deliver great outcomes for investors over the long run. There will be long term winners, and short-term losers.
That conclusion also fits neatly with our evidence-based approach to investing: index investing. The basic logic is simple. It is not just hard to consistently pick the sectors and companies that will deliver the best returns over the next few months or years, it is even harder to know when to sell them after they have peaked, and then repeat the process with the next “sure thing”.
The more realistic conclusion is to use a rules-based approach: spread your money across a broad basket of companies and let competitive markets sort the winners from the losers.
A good local example is Commonwealth Bank. When CBA has traded at what many considered an eye-watering valuation, some commentators blamed index funds, arguing that the flow of money into passive investing creates demand that pushes prices higher. But index funds buy proportionately across the whole market. They do not “choose” CBA; they simply reflect its weight in the index. This was supported by Vanguard – it estimates that index funds contribute less than 5% of daily trading volume in large-cap stocks. You may be aware that CBA’s share price has fallen from its peak of $190 to around $160 over the past 7 months despite index fund inflows remaining unchanged – this proves that index funds do not move prices.
Prices move because buyers and sellers are constantly negotiating in real time. That price discovery process is driven by institutions, hedge funds, high-frequency traders, super funds, and everyday investors. Some will be right, and some will be wrong.
But as evidence-based investors, we do not need to play that game.
In the end, the share market will work out which companies benefit most from AI. We shouldn’t change our investment approach to accommodate any new technology themes, no matter how compelling they may sound.
