AI could therefore have a similar impact on company productivity to the introduction of Windows or the customer relationship management software created by SAP.
Currently, investors are backing most of the world’s largest stocks to be winners in this area, but we also are aware that many technologies have ended up dominated by only one or two players — there is only one Windows, Google search, and Amazon.
However, all these companies are spending billions to be in the game, and if you own an index fund — happy to have 20 per cent of your savings exposed to AI and even happier not to have to choose winners and losers — you will own all the winners, but also all the losers.
So the investment question is this: will the total long-term revenue gain from AI be large enough to produce a high return on all the capital being invested? And this is a tricky question.
Given the immense total capital investment seen from the so-called Magnificent Seven tech group in the past year, the odds are that the return on total capital will be mediocre, possibly poor. There will be winners, but losers might have to write off all they have risked.
As the competition progresses, it is likely that a few companies will continue to offer waterfront AI services, while others will specialise in the many niches that emerge: AI for healthcare, AI for logistics, AI for government services.
What might be next?
Any cooling down of the land grab could also see a moderation of chip spend, at least from the year-on-year growth we have seen recently. This explosion of demand for the best AI chips has, among other things, taken Nvidia’s operating margin to 54 per cent from 26 per cent five years ago.
I expect I will be much too early taking profits in Nvidia. Its stock performance has been extraordinary, having been the leader in processors for gaming, then bitcoin and now AI.
Nvidia is also likely to lead in chip-on-wafer-on-substrate (the next generation) and TSMC is likely to dominate fabricating them.
However, I have two investment rules: be wary of share prices that have risen on the back of margin expansion — margins go down as well as up. And second, when the chief executive of a stock you own appears on the cover of newspaper colour supplements, you might think that things have gone far enough — this was a symptom of the highs of the 2000 tech stock bubble.
Oh, and a third rule: it is much better to sell too early than too late.
