While every week is CAPEX week here at Platformonomics, this particular week is our quarterly celebration of insane infrastructure instantiation, aka earnings! We’re breaking the quarterly update out to a standalone post instead of burying it in the weekly newsletter.
The four hyperCAPEX companies spent over $114 billion on CAPEX in Q3, up 18% from Q2 and up 76% from a year ago. We’ve cracked twelve figures!
CAPEX spending by Google, Meta and Microsoft is overwhelmingly data center spend while Amazon also has robots, warehouses and other paraphernalia associated with a half trillion dollar retail business. But Amazon now kindly breaks out AWS CAPEX spending on a quarterly basis (see below for detail).
Even with their copious CAPEX, the hyperCAPEX companies still have free cash flow ($209 billion over the last twelve months). Google and Microsoft’s free cash flow is even inching up despite booming CAPEX spending. Amazon’s is visibly shrinking while Meta’s is down about 20% in the last year (even as they resort to off-balance sheet funding). This data is trailing twelve months because Amazon can’t be bothered to report a quarterly number and I’m too lazy to derive it.
This quarter we add a new heuristic: how much does the Wall Street Journal earnings headline/subhead focus on CAPEX. More is obviously better.
Amazon
Amazon corporate CAPEX was $36.07 billion, up 7% from Q2 and up 83% from last year (including finance leases).
AWS CAPEX was $28.3 billion, up 76% from Q2 and up 97% from last year. So AWS CAPEX was 78.5% of Amazon’s total CAPEX, up from under half last quarter. A little over half of the finance leases were for AWS.
Wall Street Journal Headline Heuristic: No mention of CAPEX in headline or subhead (alas).
Not a very informative earnings call. The prolonged Trainium ad crowded out much other information (Amazon’s modest retail business didn’t show up until the tenth paragraph of prepared comments).
Guidance: $125 billion in 2025 (so $35 billion in Q4), and even more in 2026:
we expect our full-year cash CapEx to be approximately $125 billion in 2025, and we expect that amount will increase in 2026
Supply constraints:
Today, overall in the industry, maybe the bottleneck is power.
$23.95 billion in CAPEX, up 7% from Q2, 83% from a year ago.
Wall Street Journal Headline Heuristic: CAPEX (only) in the subhead.
Infrastructure is the foundation that “powers all”:
Our extensive and reliable infrastructure, which powers all of Google’s products, is the foundation of our stack and a key differentiator. We are scaling the most advanced chips in our data centers, including GPUs from our partner NVIDIA, as well as our own purpose-built TPUs, and we are the only company providing a wide range of both.
CAPEX (and “CapEx”): $24 Billion!
With respect to CapEx in the third quarter, our CapEx was $24 billion. The vast majority of our CapEx was invested in technical infrastructure, with approximately 60% of that investment in servers and 40% in data center and networking equipment.
Supply constraints: “Still!”
As I’ve mentioned on previous earnings calls, while we have been working hard to increase capacity and have improved the pace of server deployments and data center construction, we still expect to remain in a tight demand/supply environment in Q4 and 2026.
CAPEX Guidance: Bumping 2025 by $6-8 billion. Plus “a significant increase” for 2026!
We now expect CapEx to be in the range of $91 billion to $93 billion in 2025, up from our previous estimate of $85 billion, keeping in mind that the timing of cash payments can cause variability in the reported CapEx number.
Looking out to 2026, we expect a significant increase in CapEx and will provide more detail on our fourth quarter earnings call.
Depreciation: Yup and up!
In terms of expenses, first, as I’ve mentioned on previous earnings calls, the significant increase in our investments in technical infrastructure will continue to put pressure on the P&L in the form of higher depreciation expenses and related data center operations costs, such as energy.
In the third quarter, depreciation increased $1.6 billion year-on-year to $5.6 billion, reflecting a growth rate of 41%. Given the overall increase in CapEx investments, we expect the growth rate in depreciation to accelerate slightly in Q4.
Previous:
Meta
$19.37 billion, up 14% from Q2, and up 111% from a year ago.
Many words from Meta about their AI mulligan. Vague vagaries about capital allocation and potential revenue streams did not satisfy and Meta stock was down significantly after earnings.
I normally cheer any and all CAPEX spending but Meta is has yet to make the case there is a pony here somewhere. Without a cloud business, it isn’t clear where Meta is going to find vast new revenue streams to pay for their vast CAPEX spending.
Wall Street Journal Headline Heuristic: CAPEX in the headline AND the subhead — Meta is our Q3 winner.
Given Meta is the runt of the hyperCAPEX companies, this is a big claim:
We’re also building what we expect to be an industry-leading amount of compute.
CAPEX: $19.4 billion!
Capital expenditures, including principal payments on finance leases, were $19.4 billion, driven by investments in servers, data centers and network infrastructure.
Guidance for 2025: increase by $4 billion
We currently expect 2025 capital expenditures, including principal payments on finance leases, to be in the range of $70-72 billion, increased from our prior outlook of $66-72 billion.
Guidance for 2026: We’re still planning but up! (to paraphrase a lot of words)
A central requirement to realizing these opportunities is infrastructure capacity. As we have begun to plan for next year, it’s become clear that our compute needs have continued to expand meaningfully, including versus our own expectations last quarter. We are still working through our capacity plans for next year, but we expect to invest aggressively to meet these needs both by building our own infrastructure and contracting with third party cloud providers. We anticipate this will provide further upward pressure on our capex and expense plans next year.
As a result, our current expectation is that capex dollar growth will be notably larger in 2026 than 2025. We also anticipate total expenses will grow at a significantly faster percentage rate in 2026 than 2025, with growth driven primarily by infrastructure costs, including incremental cloud expenses and depreciation.
Personal Superintelligence is just engagement and ad optimization?
we feel pretty good that we’re going to be able to absorb a very large amount of that to just convert into more intelligence and better recommendations in our Family of Apps and ads in a profitable way.
Like, people ask us to be in the cloud business!
Like almost every week, people come to us from outside the company asking us to stand up an API service or asking if we have different compute that they could get from us.
Before scaling to billions, you still have to figure out what the (personal superintelligence) product is, and Meta doesn’t have a great zero to one track record:
We know — I mean, I would guess that Meta, I think has the best track record of any company out there of taking a new product that people love and getting it to billions of people in terms of usage.
My Never Take a Dependency on Elon Musk warning to businesses also, like, applies to Meta (they not as erratic but have a shabby record as a platform provider):
And then there are the business versions of all these too, like business AI.
Re: the off balance sheet financing of data centers:
[Analyst] …how do you see the on balance sheet versus off balance sheet financing of your AI initiatives? You’ve recently struck a deal with Blue Owl for the Louisiana data center. Is that part of the CapEx guide for ‘26? And if it’s not, how significant will that way of funding be for Meta going forward and basically will that slow down your CapEx growth past 2026?
[Meta CFO] So the JV that we announced with Blue Owl is sort of an example of finding a solution that enabled us to partner with external capital providers to co-develop data centers in a way that gives us long term optionality in supporting our future capacity needs, just given both the magnitude, but also uncertainty of what the capacity outlook in future years looks like.
In terms of how that is recognized as CapEx, our prior CapEx reflected a portion of the data center build cost prior to the joint venture being established. Going forward, the construction cost of the data center will not be recorded in CapEx. As the data center is constructed, we will contribute 20% of the remaining construction costs required, which is in line with our ownership stake, and those will be recorded as other investing cash flows.
Microsoft
$34.9 billion, up 44% from Q2, up 75% from a year ago (including finance leases).
Last quarter: CAPEX guidance for Q3 was “over $30 billion” with growth in spend for FY26 slowing relative to FY25. Such moderation.
This quarter: Just kidding! Microsoft is now projecting FY26 growth will be “higher than FY25 growth” which was 58% which suggests over $140 billion in CAPEX. Wall Street seemed to be bummed at having to update their Excel models, which sent the stock down.
We also saw a big increase in Microsoft’s use of finance leases (44% of total spend), and skewed to longer lived assets, i.e. data centers not servers, which is new.
Wall Street Journal Headline Heuristic: CAPEX in the headline
Infrastructure chest-beating:
We will increase our total AI capacity by over 80% this year, and roughly double our total datacenter footprint over the next two years, reflecting the demand signals we see.
Just this quarter, we announced the world’s most powerful AI datacenter, Fairwater in Wisconsin, which will go online next year and scale to two gigawatts alone.
And we have deployed the world’s first large scale cluster of NVIDIA GB300s.
CAPEX spending:
Capital expenditures were $34.9 billion driven by growing demand for our cloud and AI offerings. This quarter, roughly half of our spend was on short-lived assets, primarily GPUs and CPUs, to support increasing Azure platform demand, growing first-party apps and AI solutions, accelerating R&D by our product teams, as well as continued replacement for end-of-life server and networking equipment. The remaining spend was for long-lived assets that will support monetization for the next 15 years and beyond, including $11.1 billion of finance leases that are primarily for large datacenter sites. And cash paid for P, P, and E was $19.4 billion.
CAPEX guidance: “higher than FY25 growth”
With accelerating demand and a growing RPO balance, we’re increasing our spend on GPUs and CPUs. Therefore, total spend will increase sequentially, and we now expect the FY26 growth rate to be higher than FY25.
Supply constraints: yup!
demand remains significantly ahead of the capacity we have available
Therefore, we now expect to be capacity constrained through at least the end of our fiscal year.
Hello Larry!
you have to start with building out a very fungible fleet. It’s not like we’re building one data center in one region in the world that’s mega scale. We’re building it out across the globe for inference, for pre-training, for post-training, for RL, for datacenter, what have you.
(At some point we need to debunk the great “pivot to inference” story coming from the nano and neoclouds about how they will extend the useful life of their GPUs.)