Presently, I’m looking at the Olympic medal tally and can’t help but notice a certain antipodean nation sitting in 4th position behind a bunch of other nations 3-30x the size. I don’t follow the Olympic Games that closely. However, it involves a nation of which I am a citizen and crucially a ranking. It is therefore competitive, and I must win. The most ludicrous part of this whole endeavour is that I recognise the lunacy of my position… and continue to hold it.
Thank God I wasn’t old enough to know what the Olympic games were in 1992. At the Barcelona Olympics of that year, Australia did what it had always done: sent a team of amateur sportspeople while the rest of the world sent professional athletes. The outcome was somewhat predictable: the Australian team was collectively smoked like a cheap cigar.
The message was clear: if you want to be competitive at the Olympic Games, you need to bring professionals. While the cost of competing was much higher, the alternative was a steady slide into athletic irrelevance. In other words: don’t bring a knife to a gunfight.
AI hyperscalers appear to have learned this lesson. Said hyperscalers (Microsoft, Amazon, Google and Meta) are investing vast sums in AI infrastructure and don’t yet know how to monetise the tools they’re creating. However, if they don’t invest now, their competition will win and they will slide into irrelevance. It’s a classic example of game theory:
The Google and Meta CEOs seem to agree:
Sundar Pichai on the Q2 earnings call: “the risk of under-investing is dramatically greater than the risk of over-investing. Over-investing scenarios still leave you with the infrastructure that has a long useful life… while not investing to be at the frontier has more significant downsides”
Mark Zuckerberg in a recent interview with Bloomberg: “If products are able to grow massively over the next few years, which I think there’s a very good chance of, I’d much rather over-invest and play for that outcome than save money by developing more slowly… [there’s a] meaningful chance a lot of companies are over-building now… but they’re all making rational decisions because the downside of being behind leaves you out of position for the most important technology over the next 10-15 years.”
It seems like that AI spending isn’t going anywhere.
What we’re dealing with is a kind of gold rush. Rather than spending a lot of time and effort betting on who strikes gold first, we could instead place a bet on the hyperscalers’ collective efforts to dig for it. In gold rush terms you would bet on the providers of picks and shovels. Or to return to our Olympic analogy: we’d sell the team apparel, not bet on the medal tally. In this case the purveyors of picks and shovels are the companies that provide the wires, electricity and cooling systems to the datacentres.
In recent weeks, we’ve seen a pullback in the hyperscalers’ stock prices because the market is pricing in uncertainty as to the size and timing of the payoffs from their AI investments. Anything AI-related has fallen with them. However, it is the hyperscalers that bear the risk of striking gold, not the pick and shovel companies – you need those regardless. It seems to me part of this sell-off might be overdone.
コメント