The CapEx Two-Step
It's that time of year again: quarterly earnings for Big Tech. And you know what that means... time for Wall Street to freak out over CapEx spend related to AI build-out once again. Sure enough, Google went first and upped their forecasted spend pushing the range past $200B for the first time. The market puked. Next up, Meta. They simply raised the low-end of their previous guidance (to $130B up from $125B). The market puked.1 Then there's Microsoft...
Wait, what's this? The stock is currently up over 8% in after-hours?! Why? Well I think it's mainly continued impressive Azure growth BUT it's at least in part because unlike with Google's and Meta's relatively strong earnings, that growth isn't negated by yet another resetting of CapEx expectations. In fact, Microsoft is now guiding that their number will come in at $175B for the 2026 calendar year. That's actually $15B below the previous estimate.
So is this Microsoft crying "uncle" and giving in to Wall Street demands to pull back spending? Well a lot of the reporting and commentary sure makes it sound like that's what's happening. Here's Deborah Mary Sophia, Stephen Nellis and Aditya Soni on this topic for Reuters:
The first-quarter forecast was below analyst estimates of $56.02 billion, according to Visible Alpha data, and the calendar 2026 estimate was below Microsoft's own previous estimate of $190 billion.
But wait a minute. Did you miss it? The paragraph immediately preceding that one:
Microsoft also said it would now spread long-term leases on data centers over 25 years rather than 15, which has the effect of lowering its annual reported capital expenditures. Microsoft said its spending plans remain unchanged and that it expects reported capital expenditures of $50 billion for the fiscal first quarter of 2027 and $175 billion for the 2026 calendar year.
Aha! The wording makes this sound like a sort of sleight of hand designed to make Microsoft’s CapEx look smaller. And that is one effect, but the actual mechanism is a little more nuanced.
Microsoft is extending the estimated useful lives of its data centers (and office buildings) from 15 years to 25 years. That change, in turn, means more of its future data-center leases will be classified as operating leases rather than finance leases. Microsoft includes the latter in its reported CapEx number but excludes operating leases. So the same underlying infrastructure commit now will mean a lower CapEx figure.
In other words, Microsoft isn’t spending less, it's changing how a portion of that investment appears. Optically, this makes it look like they're coming in under the guidance numbers, but per Microsoft's own comments on the matter, they're actually in line with the previous guidance. That's how you square that circle.
It's not like Microsoft is trying to hide this change, but they're also undoubtedly happy to benefit from the headlines that suggest they're being more prudent with their spend. At the same time, they're also clearly capacity constrained, so you could argue that it would be foolish to pull back on spend at the moment. Microsoft actually did that once before a couple years ago, pausing some spend and shifting others (as they tried to wiggle away from the massive OpenAI commitments), and it actually now looks like that was a mistake.
So instead, Microsoft has ended up with a way to make it optically look like they're being mindful while not actually changing anything! Rather brilliant.
To be clear and fair, I don't think this is all about obfuscation, but it's some masterful timing on the part of CFO Amy Hood. Microsoft has actually been a laggard in this regard, with Amazon, Google, and Meta all having longer such estimated useful lives – Amazon and Google actually extend up to 40 years in some cases (though there's a range – which Google also shifted in 2024 without any real talk of it) with Meta somewhere in the 25 to 30 year range.2 In other words, Microsoft is now benefitting from being the last-mover here in a time when Wall Street is most skittish about spend.3 Well played.
We'll see where Amazon lands with regard to CapEx later today. And, in turn, what Wall Street thinks about it. But it's entirely possible that only Microsoft comes out of this earnings season looking great from a stock-perspective.
Well, and probably Apple, which also reports later today. Look for their CapEx number to come in somewhere in the $9B to $10B range, just like pretty much every other quarter for years and years as their peers zoom spend to the moon.
Update July 31, 2026: Interestingly, after they both announced earnings, Amazon's stock surged (joining Microsoft), while Apple's fell. Amazon certainly isn't pulling back from CapEx spend, but AWS is growing so fast that it clearly eased any Wall Street fears on their spend.
Apple, meanwhile, issued some weak guidance going forward and showed some weakness in Services (the second most important business after the iPhone).
Apple's CapEx? $2.48B for the quarter – down 29% year-on-year. That puts them on pace for about $9B to $9.5B in CapEx for the year. Both Google and Amazon are now targeting past $200B.
M.G. Siegler
M.G. Siegler
M.G. Siegler
1 Though this seemed more directly tied this quarter to a profit miss, weak guidance, the decimation of free cash flow – though the latter is obviously directly tied to the CapEx spend! ↩
2 This is obviously different (though related) from actual server lifespans, which is far more contentious given the speed of AI development and chip deployments. There industry is even further from fully sorting that out, which could come back to bite the neoclouds (as well as Big Tech) eventually. ↩
3 Timing-wise, it also makes sense to do this at the beginning of a new fiscal year, which Microsoft just kicked off. ↩
Member discussion