Microsoft Latest Results: Missed Cloud Expectations Amid Infinite AI Horizons
I had intended to write earlier and more often this week, but as often happens, life had other ideas. My good intentions were ambushed on this occasion by an ear problem, the details of which I will elide in the interests of decorum. After all, some of you might be eating.
I tried to write on Monday, but the pain and the throbbing made for poor prose and a foul mood. I visited the doctor yesterday, and I now have a cure in hand. In a few days I should be back in normal working order for a human my age. It’s not much, but it’s all I’ve got.
Even now, I think I’m capable of banging out a few coherent sentences, but you, as always, are a better judge of my output. The condition from which I am recovering is, like a lot of things in life – except for what happens at the very end of the road, when the pictures stop permanently – merely annoying and exasperating, not in any sense lethal. Unless I get hit by a car – always a possibility when walking in the city – I will survive to torment the keyboard another day.
As luck would have it, I'm fit enough to give the keyboard some punishment today, too.
Preliminaries aside, let’s get to the main event. Regular readers of this forum will recall a post I wrote last week on Alphabet’s (Google’s) latest set of quarterly financial results, which weren’t at all bad, or generally perceived as such, except for a conspicuous surge in capital spending relating to datacenters and infrastructure supporting genAI services. Wall Street analysts and institutional investors had concerns and questions, as is their wont. They wanted to know when this spending, lavish and persistent by any conventional standards, would deliver a return on investment. Google equivocated and prevaricated, offering generic homilies of passionate belief in genAI’s restorative powers while preaching to an audience characterized by agnostic pragmatism.
Google’s lack of specificity in answering investors’ questions rankled, though in fairness to Google, I must say that its answers, when compared with the obfuscations and subterfuge of those offered on the Tesla earnings call, were the model of transparency.
Google and Tesla discomfited the market, setting the stage for quarterly results this week from four other charter members of the Magnificent Seven: Microsoft, Meta Platforms (Facebook), Amazon, and Apple.
Undone by Unforced Error
Microsoft had the dubious privilege of launching proceedings this week, releasing its quarterly financial results yesterday and offering guidance on quarters to come. Microsoft somehow managed to make favorable results appear lackluster, and the guidance it offered added shades of ambiguity and murk.
In all seriousness, the Microsoft executive team should study old transcripts of Cisco’s results in the late 90s, when CEO John Chambers and CFO Larry Carter would dazzle their Wall Street interlocutors like the Harlem Globe Trotters running circles around the Washington Generals. (Of course, it went pear-shaped for Cisco’s dynamic duo when the dotcom boom went bust in 2001.)
Back then, before the bubble burst, it seemed that Cisco not only had a thriving business, but also had a constant measure of transactions occurring in the field. Cisco almost invariably beat consensus earnings expectations by a penny or two, delivering revenue that surpassed estimates, but not by too much. Each Cisco quarterly earnings cadence was masterfully plotted and smoothly executed, with Cisco first setting beatable expectations and then surpassing them just enough to discourage the Wall Street crowd from getting ahead of the game by raising the bar to a higher degree of difficulty. That’s how you manage the Street, and there are companies today that run the table just as proficiently as Cisco did in the days of yore. Microsoft is more than capable of accomplishing such a feat, though yesterday’s results qualify as a minor misstep.
All up, Microsoft actually managed to meet its overall targets, yielding overall revenue and earnings for its fourth quarter that slightly surpassed expectations. The problem, if one can be so bold as to characterize it as such, was in Microsoft’s Azure cloud business, whose revenues grew 29% against Wall Street expectations of 30%, and down from the 31% growth for Azure in the prior quarter.
If we allow ourselves to step back and study these results from the perspective of a disinterested observer, 29% y/y revenue growth in a business as sizable as Azure – remember, the law of large numbers eventually means revenue growth tapers as it reaches higher altitudes – remains an impressive achievement. Alas, the markets don’t have time for detachment and dispassionate appraisal. From the market’s standpoint, the fact remains that 29% growth is not 30% growth, and that was the expectation that Microsoft had encouraged among the Wall Street cognoscenti. Nobody likes to be shortchanged, even when the payout is respectable.
When a company falls short of meeting expectations, Wall Street’s disappointment can quickly turn to vexation. As with Alphabet (Google) last week, irritation found purchase in Microsoft’s burgeoning AI-related spending, which has raced far ahead of AI-related customer demand. In missing the market’s consensus estimate for the Azure cloud business, Microsoft put itself in the hot seat for a post-earnings call interrogation regarding when investors should expect the company’s massive spending on AI datacenters and infrastructure to provide material returns on investment.
Twitchy Trigger Fingers of the Hyperfocused
Cloud and AI are inextricably linked, since the former is where much of the latter is likely to run. The Wall Street pack was already restless after last week’s mixed messages from Google and Tesla, and Microsoft gave the hornet’s nest a sharp poke when it fell short of cloud estimates. This quote from a Wall Street Journal article clearly attests to the sensitivity of investor sensibilities:
“There’s a segment of the investing community that is hyperfocused on very small changes to the Azure business,” said Brad Reback, an analyst at Stifel Financial.
The word hyperfocused might represent inelegantly redundant phrasing, but it captures the twitchy specificity of the number watchers who anxiously anticipate cloud’s skyward thrust powered by genAI’s booster rockets. When the launch fails to trouble the firmament, thwarted expectations demand an explanation.
As recounted in a Business Insider article, Microsoft disclosed that its capital expenditures, including financial leases, amounted to $19 billion in the fourth quarter, with AI-related spending accounting for nearly half of that amount. About half of the total was for building and leasing datacenters, and the other half, on an approximate basis, was allocated to servers, CPU, GPUs (including new AI accelerators from AMD and Nvidia), and presumably networking, though the latter often gets short shrift (if any shrift at all) in the business press.
That’s a lot of money, and Microsoft flatly stated that it would continue spending heavily in pursuit of its AI aspirations, which the company appears to measure on the timescale of historic empires rather than on quarter-to-quarter increments. A Reuters article summarized the current facts and figures, reporting that Microsoft’s horizons for AI-related investments stretch far beyond conventional fields of vision. Quoting Reuters:
Microsoft said its capital spending rose 77.6% to $19 billion in its fiscal fourth quarter that ends June 30, with cloud and AI-related spending accounting for nearly all of the expenditures. For all of fiscal 2024, capital spending totaled $55.7 billion.
Group CFO Amy Hood said the spending was necessary to support demand for AI services and the company was investing in assets that "will be monetized over 15 years and beyond."
Serenity Now . . . and for the Next 15 Years
Can you imagine? Not just 15 years, but beyond. I’m already retired, but many of you will have happily ended your careers by the time Microsoft’s genAI’s assets are fully monetized. That’s asking Wall Street for a superabundance of patience it has does not typically possess. What Microsoft is requesting from Wall Street is unshakable belief. Can the Street remain indefinitely devout to the genAI religion? Will Wall Street analysts have to exchange business-casual attire for sackcloth and ashes? (They can, of course, keep their smart watches.)
To give Microsoft its due, the company offered quantitative metrics on current genAI adoption in the Azure cloud. CEO Satya Nadella, a full-throated proponent of AI as a business proposition over a limitless temporal horizon, reported that Azure AI was now used by more than 60,000 customers, up nearly 60% year-on-year; he also noted that average spend per customer on AI continues to increase, though hard numbers on that measure were not provided.
If that old-time genAI religion fails to enthrall the market makers, Nadella and his executive team will have to keep priming the promotional pump with evidence of commercial progress. They are advised from this humble quarter to provide as much detail as numbers provide and discretion allows. Nadella and his team are also strongly advised to not miss their cloud numbers next quarter. Woe betide those who fall short of the expectations they helped set, as evidenced by this quote in the Reuters article:
"The street doesn't have a lot of patience. They see you spending billions of dollars and they want to see a pickup in revenue of that amount," said Daniel Morgan, senior portfolio manager at Synovus Trust, which holds shares in Microsoft.
"If these companies do not hit it out of the ballpark and are far better than the estimates, then they are going to be knocked back," he added.
The days of repeating the word genAI as an incantation and watching your stock rise unabated appear to be over, unless you’re Nvidia, a company to which I will return my critical gaze tomorrow. One analyst yesterday said that Microsoft’s mildly disappointing cloud results, combined with its ever-expanding AI capital expenditures, represented a “transfer of wealth from Microsoft shareholders to Nvidia shareholders.”
Tomorrow I will challenge such a facile view, which is at best glancingly accurate and then only for a limited time, qualifiers that seriously diminish its utility as valid benchmark for a long-term thesis.