Most companies reporting 29% quarterly growth in their most important division would be celebrated on Wall Street like Olympic gold medal winners.
But when Microsoft said on Tuesday that its Azure cloud unit—a key component of its AI strategy—had grown at that pace, investors reacted as if Simone Biles had just flubbed the uneven bars. They sent the company’s shares tumbling as much as 7% in after-hours before they recovered somewhat; the stock is down just 1.6% at mid-day today.
Never mind that Microsoft’s overall quarterly results slightly exceeded analyst expectations.
There are two key takeaways from this episode:
One is that the current AI frenzy has gotten so out of hand that investors are pricing stocks to perfection and as if rapid growth will continue into perpetuity. When companies fall just a little short, at least in Wall Street’s mind, stocks take a beating.
Microsoft’s 29% cloud growth in its most recent quarter was slower than the 31% in the previous quarter and below the 30% that analysts had expected. Investors worry the deceleration may signal that the AI boom is maturing and slower growth ahead.
Of course, Microsoft executives spun the slowdown as partly a consequence of being unable to keep up with demand. Building cloud infrastructure—i.e. data centers—takes time, which is why Microsoft has partnered with other companies including Oracle to fill in the gap.
That brings us to a second takeaway from yesterday’s earnings: Building those AI-friendly data centers and buying AI chips is super expensive. Big Tech companies are spending tens of billions of dollars each annually on data center construction and expansion, with no end in sight.
Microsoft, for example, poured $13.9 billion into capital expenditures in the latest quarter (and $19 billion if you include leases), a 55% increase from the same period a year ago. Executives expect to continue spending at the same level in the future.
Last week, Google-parent Alphabet reported its own high infrastructure costs (and suffered a big drop in its share price as a result). Wall Street will be paying close attention to the equivalent numbers at Facebook-parent Meta when it reports its numbers later today and Amazon and Apple’s tomorrow.
Investors seem to think that AI profits can be harvested without much upfront investment. Sorry to break it to you, but no.
To bring it back to today’s Olympic theme, I have no idea which Big Tech company will ultimately win the AI race. But I do know this: Winning a gold medal takes years of preparation and patience.
Verne Kopytoff
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
