explainx.ai0k
TrendingAI News TodayPathwaysSkills
Pricing
explainx.ai

Upskill in AI — 16 free pathways, live workshops & bootcamps, and 50+ courses from practitioners. Plus the skills, tools, and MCP servers to practice on.

follow us

follow on google

Add explainx.ai as a preferred source

corporate training

support@explainx.ai

get started

Find your pathTake Free Evaluation

community

Join the community

learn

mind: share how you thinkpathways — start freeworkshopsbootcampscoursescompare Explainxcertificationsmock testsexplainx universitycorporate traininglearn skills & mcp

discover

skillsmcp serversexplainx mcptoolsmdx readeragentsllmsdesignsdictionarypeopleagi trackerfelony benchranks

company

aboutvisionmissionteaminstructorsteach on explainxpartnershipscommunityhackathonscareers

content

daily AI newsstate of AI — live resultsblogreleasespromptsgeneratorsresource libraryfor LLMsexplainx.ai kids

solutions

all solutionsdeveloper upskillingmarketing upskillingproduct manager upskillingleadership upskilling

newsletter · weekly

Get AI news, tools, and insights in your inbox.

supportcontactprivacytermsdata rightshow we create contentsubmission guidelines

© 2026 AISOLO Technologies Pvt Ltd

explainx.ai

On this page

  • TL;DR: the numbers
  • What did Amy Hood actually say about capex?
  • How strong was Azure, and what changed in the reporting?
  • What is the money buying?
  • Cash flow and the "do not borrow yet" message
  • The people side: buyouts and cuts
  • Copilot: usage up, seats undisclosed
  • What this means for what you build or pay
  • How does this compare with the other hyperscalers?
  • What is not known
  • How to check it yourself
  • Bottom line
  • Related reading
← Back to blog

explainx / blog

Microsoft Guides to $170 Billion of AI Capex for 2027: What Hood Said and What It Means

Microsoft, Azure, AI Infrastructure, Data Centers, Earnings

Part of Microsoft, Apple and Amazon AI

Microsoft guided to about $170B of calendar 2027 capex as Azure grew 48% to $33.2B. What CFO Amy Hood said, what slipped, and what it means for AI builders.

Oct 28, 2026·8 min read·Yash Thakker
add explainx.ai
go deep
Microsoft Guides to $170 Billion of AI Capex for 2027: What Hood Said and What It Means

Microsoft told investors on October 28, 2026 that it expects to spend about $170 billion on capital expenditures and finance leases in calendar 2027, and that Azure, now disclosed in dollars for the first time, grew 48% to $33.2 billion. The reason for the spending, CFO Amy Hood said on the call, is simple: "We still have demand that exceeds supply." This post sticks to what CNBC's account of the call and the filing-level numbers say, flags what is guidance rather than fact, and spells out what it means for people who build on Azure AI.

TL;DR: the numbers

table · 2 cols
QuestionAnswer (source: CNBC's report of the earnings call)
What did Microsoft guide for 2027 capex?About $170 billion of capex and finance leases for calendar 2027
How big was fiscal Q1 spend?$45.2 billion, up about 30%, versus $50.67 billion expected
What is next quarter's capex?Above $50 billion for fiscal Q2
How fast did Azure grow?48% to $33.17 billion; guided 49% to 50% constant currency next quarter
Total revenue?$92.28 billion, up 19%
Free cash flow?$18.63 billion, down 27%
Why is spend lagging?Third-party capacity deliveries, mostly GPU-as-a-service providers, were delayed

What did Amy Hood actually say about capex?

According to CNBC, Hood said Microsoft is aiming at more than $50 billion of capex and finance leases in the fiscal second quarter, with about $170 billion for all of calendar 2027 because of capacity issues. The fiscal first quarter, which ended September 30, came in lower than Wall Street expected: $45.2 billion against a StreetAccount estimate of $50.67 billion.

The gap is not a pullback. Hood said "some third-party capacity deliveries we expected in Q1 were delayed", and that those delays involve GPU-as-a-service providers. In the second half of the fiscal year she expects spend to accelerate, "primarily from CPUs, inclusive of higher memory and storage costs, and GPUs, as well as the 3P capacity deliveries that shifted to H2."

One caution: the $170 billion figure is a forecast reported by a single outlet's account of the call. The primary document is Microsoft's own investor relations site and its fiscal 2027 Q1 release, which readers should check for exact wording. Capex figures that include finance leases are not the same as cash paid for property and equipment, so year-to-year comparisons need care about what counts as "capex."

How strong was Azure, and what changed in the reporting?

Azure revenue topped $33 billion in the quarter. CNBC reports that Microsoft disclosed Azure revenue in dollars for the first time, at $33.17 billion versus $32.32 billion expected. Growth was 48%, up from 42% in the June quarter, even with the delayed third-party capacity.

Microsoft also reorganized how it reports. After eleven years with three segments, it now reports two: Agents and Infra, where revenue rose almost 25% to $76.8 billion and which holds Azure, Microsoft 365 cloud products, server licensing and frontier services, and Devices and Consumer, where revenue slipped 3% to $15.48 billion. Sales of devices and Windows licenses fell 22%, which Microsoft tied to higher memory and storage costs and weak PC demand.

The segment name is itself a message. "Agents and Infra" puts AI agents next to the infrastructure that runs them, which matches how the company has been talking about Foundry Agent Service and the Copilot harness.

Data center power and server cabinet link illustrating Microsoft capex and AI data center capacityData center power and server cabinet link illustrating Microsoft capex and AI data center capacity

What is the money buying?

Hood said about 75% of capex is now tied to short-lived assets such as CPUs and Nvidia GPUs, up from about 66% in the past three quarters and about half a year earlier. Short-lived assets are the ones that turn into revenue fastest, and they also wear out fastest, which is why the mix matters for margins and for free cash flow.

Hood also flagged cost pressure: the guidance includes "higher memory and storage costs." That echoes what the device line shows. Memory prices are raising the cost of PCs and servers at the same time, a theme we covered in the context of AWS raising GPU Capacity Block prices.

Cash flow and the "do not borrow yet" message

Free cash flow fell 27% to $18.63 billion. CNBC notes Alphabet reported more than $9 billion of negative free cash flow in its quarter, and Meta said a week earlier that its cash flow turned negative for the first time. Microsoft is the one hyperscaler still reporting positive free cash flow.

Hood said Microsoft will keep generating free cash flow through fiscal 2027 but "will access additional sources of funding when it's attractive to do so," adding that it has not yet tapped debt markets despite its ratings. Net income rose to $36.63 billion. Adjusted earnings exclude a roughly $200 million loss on its OpenAI investment, a relationship we broke down in the filing showing about $10 billion of revenue from OpenAI in one quarter.

The people side: buyouts and cuts

The same quarter included 4,800 announced job cuts, followed on Monday by an offer of buyouts to up to 60,000 employees, as CNBC reports. That sits uneasily next to a $170 billion build plan, and it is the point of the "capital efficiency" framing: spend on chips and power, trim elsewhere. Our earlier post on the voluntary exit offer and the claimed-versus-verified look at the buyout details cover that side.

Copilot: usage up, seats undisclosed

Nadella said the number of people using Microsoft 365 Copilot is up 50% from last quarter. The company did not disclose paid seats. Hood said the new high-end Microsoft 365 E7 bundle has millions of seats. During the quarter Microsoft also set up an AI implementation unit and revealed a Copilot "super app" with coding abilities, per CNBC. Usage growth with no seat count is a classic "claimed, not verifiable" metric: treat it as direction, not adoption proof.

What this means for what you build or pay

This is a finance story, so here is the practical reading.

  1. Capacity, not price, is the signal. Nothing announced changes Azure AI pricing. But "demand exceeds supply" is what produces regional quota limits, waitlists for large GPU allocations and slow approvals. If your roadmap depends on a big Azure GPU commitment in the first half of 2027, ask for it early.
  2. Delays come from third parties. The slipped deliveries are from GPU-as-a-service providers that Microsoft rents from. That is the same layer behind projects like the Nscale Loughton datacentre that Microsoft stepped back from, a reminder that grid power and siting, not just chips, gate timelines.
  3. Inference economics matter more. Microsoft's own chip roadmap, including the Maia 300 inference chip promising 40% more tokens per megawatt, is aimed at turning scarce megawatts into more tokens.
  4. Design for model routing. When capacity is tight, cheaper models and caching are the lever you control. See the Microsoft token-wars hack on cutting cost per answer.
  5. Watch the denials too. An earlier report of 20% Azure cost cuts was denied by Microsoft; see what was claimed versus verified.

How does this compare with the other hyperscalers?

CNBC frames Microsoft as the outlier on cash: Alphabet posted more than $9 billion of negative free cash flow in the same period and Meta's cash flow turned negative for the first time a week earlier. All three are raising spend into the same constraint, which is chips, memory, power and sites arriving late. Microsoft's difference is that a large share of its spend is leased from outside providers, so its delays show up as third-party deliveries rather than its own construction schedules. For a reader choosing a cloud, the useful question is not who spends most but who can actually deliver GPUs to you on the date you need them.

What is not known

  • The exact split of the $170 billion between owned data centers, finance leases and chips was not in the reports we could open.
  • Microsoft did not give a paid Copilot seat count.
  • Whether the delayed third-party capacity arrives on schedule in the second half is a forecast, not a fact. Hood herself said spend will "accelerate" then.
  • The $170 billion number is guidance. Guidance can change in later quarters.

How to check it yourself

Open Microsoft's investor relations page for the fiscal 2027 first-quarter release and the prepared remarks, compare "capital expenditures including finance leases" to "cash paid for property and equipment," and read the CFO commentary for the calendar 2027 wording. CNBC notes cash paid for property and equipment grew faster than overall capex in the quarter. A figure like $170 billion can mean different things depending on whether leases are counted.

Bottom line

Microsoft is telling the market it will keep building at a pace that outruns even its own earlier plans, while cutting staff and watching cash. For builders, the takeaway is not the headline number but the constraint underneath it: AI capacity is still the scarce good, and the largest cloud still cannot get GPUs delivered as fast as customers want them.

Related reading

  • Microsoft booked about $10 billion from OpenAI in one quarter
  • Microsoft voluntary exit offer to 60,000 employees
  • Maia 300 inference chip and tokens per megawatt
  • Microsoft pulls out of the Nscale Loughton datacentre
  • Azure 20% cost cuts report: claimed vs verified
  • AWS Capacity Blocks GPU price hike

Figures reflect CNBC's October 28, 2026 report of Microsoft's fiscal Q1 2027 results and call; check Microsoft's investor relations materials for final wording.

Spotted something out of date? Let us know.

People in this article

  • Satya Nadella →Chairman and CEO of Microsoft
Explore people in AI →
Yash Thakker

Written by

Yash Thakker

Yash is an AI expert with over 300K learners. Join his workshops →

View Yash Thakker in People in AI →

Related posts

Oct 18, 2026

Microsoft Pulls Out of Nscale's Loughton AI Datacentre Over Power Supply

The Guardian reports that Microsoft told Nscale it is ending its agreement for the Loughton, Essex datacentre the UK government called its largest sovereign AI site. The trigger is a grid connection that may not arrive until the early to mid 2030s. Here is what is confirmed, what is not, and why power is now the binding constraint on AI buildouts.

Oct 14, 2026

Microsoft Azure 20% Cost Cuts: What The Information Reported, What Microsoft Denied

The Information reported on October 13, 2026 that leaders in Microsoft's Azure organization were asked to find at least 20% in cost savings. A Microsoft spokesperson says leadership has not directed teams to develop cost-cutting plans. Here is the claim, the denial, and what nobody has confirmed.

Oct 28, 2026

Microsoft Booked About $10 Billion From OpenAI in One Quarter: What the Filing Shows

Microsoft recorded roughly $10 billion in revenue from its commercial arrangements with OpenAI in the quarter ended September 30, 2026, according to TechCrunch reading a regulatory filing. Here is what is confirmed, what is not, and what it means for builders who pay for Azure and OpenAI models.