The AI sell-off tells investors that valuations will always matter
Revolutionary tech doesn’t always mean a good investment; it’s not just what you buy, but also what you pay
ABOUT four months ago, I highlighted the very real possibility that a bubble had inflated in the artificial intelligence segment. Even though optimism and momentum were running high, a correction would inevitably occur sooner or later.
Over the past fortnight, AI stocks plunged. And on Nov 5, in the wake of the US midterms, the Nasdaq sank 5.5 per cent, erasing about US$2.5 trillion in market value.
In truth, the selling had already been gathering momentum since mid-October.
Sentiment looked like it was already weakening, possibly in anticipation of a Democrat victory in the midterms, and because of growing doubts over whether AI-led earnings could justify the vast expenditures involved.
Some might argue that these moves are not indicative of a burst bubble and should better be described as an AI-led market correction.
Semantics aside, if there is one takeaway from these events, it is that valuations matter; they always have and always will.
For much of the last three years, Wall Street seemed to have forgotten this basic principle, forgoing fundamentals in favour of momentum, a good story and faith in the maxim “this time is different”.
The prevailing argument was that AI represents a unique technological revolution comparable to the advent of the Internet.
Hence, conventional valuation measures are inadequate – as was the case back in 2000 during the dotcom boom.
Maybe so, but it is worth remembering that even the greatest tech revolution does not make every company associated with it a good investment.
Consider what you are paying for
Even after the sharp sell-offs of the last few weeks, the sector’s valuations are still steep.
AMD, for instance, might have fallen almost 40 per cent since early October, but still trades at a trailing price-to-earnings ratio of around 96; Palantir’s equivalent is 112 times earnings.
Interestingly, Nvidia now trades at a more reasonable 26 times earnings. And there could still be further downside, given the recent scandal involving its CEO Jensen Huang, which has yet to play out fully.
This illustrates the point that not all AI stocks are equal, and that investors need to be discerning.
Adding further uncertainty is the Democrats’ sweeping victory in the midterms on Nov 3. It has raised expectations of tougher scrutiny of AI, particularly the construction of energy-intensive data centres.
Concerns about electricity costs, environmental consequences and the enormous resources required to sustain AI infrastructure are no longer just commercial considerations, but political issues now.
For investors who assumed virtually unlimited expansion of AI infrastructure, this represents a potentially significant shift.
However, it would be premature to conclude that any possible new regulation would derail AI. Several US brokerages have stated that political risks may have been exaggerated.
Still, risks remain. The obvious parallels with the dotcom bust of 2000 cannot be denied. The Internet did transform the world, just as AI probably will.
But that did not prevent investors from suffering devastating losses after paying absurd prices for technology and dotcom shares. There is, after all, a crucial distinction between a revolutionary technology and a seemingly attractive investment.
The fundamental lesson is that no company, no matter how exciting its prospects might be, is worth buying without carefully considering its value. In investing, what matters is not just what you buy, but also what you pay.
