Will capital spending on AI continue to boom?
Spending a whole lot of money on chips with the hopes of making a whole lot of money via AI has been the dominant strategy for most of America’s leading companies. Two noteworthy exceptions to this trend are Nvidia and Broadcom, which are designing chips that power the AI boom.
The AI-linked outlays from the S&P 500’s “hyperscalers” — Microsoft, Amazon, Alphabet, Meta, and Oracle — are estimated to total in the hundreds of billions in 2024, prompting shortages of the cutting-edge semiconductors to train and refine generative-AI models and a frenzied build-out of data centers to harness their power. This is a big source of current profits for some tech giants that’s giving another group of tech giants something to dream on (and start to enjoy).
Narratives around the merits of all this capital spending have evolved and shifted over time. But with every hyperscaler besides Microsoft handily outperforming the S&P 500 in 2024, it’s hard to argue that investors are overly pessimistic on the prospective return on investment.
Right now, a shorthand summary of investors’ view is that this is a case of throwing good money after good. This raises the risk that a negative turn in how much companies are willing to spend building out these new capabilities coincides with a more pessimistic view on the returns that will be generated from these capital outlays.
Of course a universe of more benign scenarios exists, including relatively uncorrelated outcomes from a correlated investment boom — that is, clear winners and losers. Or this tree of capex seemingly growing to the sky. But to quote the famous statistician and trader Nassim Taleb: “I've seen gluts not followed by shortages, but I've never seen a shortage not followed by a glut.”