Azure Hits 48% and Finally Shows Its Dollars, but Microsoft Now Borrows to Build

Results vs. Consensus

Q1 FY27 Scorecard

Quarterly Comparison (YoY)

Quarterly Comparison (QoQ)

The sequential comparison carries a seasonal distortion that flatters the September quarter. Fiscal Q4 is Microsoft's heaviest quarter for operating expense, so opex fell 16.3% sequentially and operating margin rose 470bps from Q4. The useful signal is the year-over-year margin gain of 89bps, which happened while gross margin fell 126bps. A more telling sequential number is revenue: September grew 2.5% over a June quarter that is normally the seasonal high, because June carries the heaviest volume of multi-year contracts.

Quality of the Beat

• Revenue: The $1.7B beat against consensus was led by Agents and Infra, which cleared the high end of its recast guide by $1.05B, but it was not one-dimensional. Devices and Consumer also beat its range by $0.28B, on search and advertising ex-TAC growth of 12% and an XBOX content decline of only 2%, both better than guided. FX was negligible: total company growth was 19% on both a reported and constant-currency basis. Service and other revenue grew 25.7% while product revenue fell 8.1%, with Windows OEM driving the product decline.
• Azure decomposition: Management named three drivers: Foundry, where efficiency gains across the GPU fleet freed capacity that went to customers; higher revenue per token than expected; and CPU fleet engineering that found and monetized underutilized capacity. It also named the drag: third-party GPU capacity it expected in Q1 arrived late. Azure added $3,748M sequentially, 2.2 times the $1,672M it added in the same quarter a year ago.
• Margins: Gross margin fell 126bps to 67.8% on the mix shift to Azure, but operating margin rose 89bps to 49.8% because operating expense grew only 6.2%. Sales and marketing was flat in dollars year-over-year and total headcount declined 2%. In Agents and Infra, gross margin dollars grew 20.3% against 24.5% revenue growth, and segment operating margin slipped only 24bps to 52.8%. In Devices and Consumer, operating margin rose 183bps to 34.6% on lower Activision amortization.
• EPS: Non-GAAP EPS of $4.89 grew 18.4% against operating income up 21.0%. Three points of the gap are below the line: a higher tax rate (20.3% against 19.1%) and other income that turned slightly negative excluding OpenAI as finance-lease interest rose. The share count fell 0.3%. The GAAP growth rate of 32% is a comparison artifact, since the prior-year quarter absorbed a $3.1B after-tax loss on OpenAI.
• Cash flow: Operating cash flow grew 28.7% to $58.0B on cloud billings and collections. Cash paid for property and equipment more than doubled to $39.4B, leaving $18.6B of free cash flow. Below that line sits an investing outflow management did not discuss on the call: "other" investing cash of $12.9B, up from $6.2B, an increase the quarterly filing attributes primarily to facilitating the purchase of components.

Segment Performance

This is the first quarter under the two-segment structure announced in September. Agents and Infra combines the Microsoft Cloud, productivity and server licensing, and consulting and support. Devices and Consumer holds Windows, XBOX and advertising. The change also moved several businesses between metrics: GitHub cloud and Security Copilot left Azure for Microsoft 365 commercial cloud, and Healthcare and Life Sciences cloud left Azure for the new Industry solutions metric. Azure growth on the new definition was 42% in Q4 FY26, against the 43% reported in July. Every growth comparison below uses the restated basis.

Segment Results

Revenue by Product and Service Offering

Azure is now the largest single revenue line in the company at 35.9% of the total, and it contributed $10.8B of the $14.6B of year-over-year revenue growth, or 74%. Microsoft 365 cloud added another $4.1B. Every other line combined subtracted about $0.3B: the declines in Windows OEM, productivity and server licensing, and XBOX more than offset the gains in advertising, Industry solutions and support.

Agents and Infra ($76.8B, +25%)

Azure: $33.2B, +48%

Azure growth of 48% beat the recast guide of 44-45% cc despite a supply shortfall. Management attributed the upside to Foundry, where efficiency work on the GPU fleet created capacity it could allocate to customers, to higher revenue per token from the mix of models served, and to the CPU fleet, where engineering work located idle capacity that could be sold. Foundry revenue more than tripled year-over-year against more than doubling last quarter, and Foundry customers passed 120,000 from 100,000. Microsoft added 24 new datacenters across five continents, fewer than the 31 it added in Q4, and cut dock-to-live times for new GB300 systems in its largest regions by about 20% sequentially.

"Across our CPU fleet, engineering work enabled us to identify and monetize underutilized capacity this quarter, helping us serve more of the growing demand with our existing capacity." — Amy Hood, CFO

Management also gave investors an unusually explicit list of what makes the Azure growth rate move from quarter to quarter: the timing of engineering efficiencies, third-party GPU lease deliveries, Microsoft's own capacity builds, and revenue-per-unit changes from model mix. That list is a warning label as much as an explanation. Two of the four factors drove this quarter's upside, and one of them, token pricing, is set partly by model providers rather than by Microsoft.

Assessment: The acceleration is real and it is broad. Management described growth across enterprise and small and medium business and across core cloud services as well as AI infrastructure. The disclosure change is a material improvement in its own right: after years of quarterly growth rates without a quarterly base, investors can now see that Azure added $3.7B of revenue in a single quarter. The caution is about composition. Part of this beat came from finding capacity in the existing fleet and from favorable token pricing, both of which management described as hard to predict.

Microsoft 365 Cloud: $27.8B, +17%

Microsoft 365 commercial cloud grew 17% reported and 18% in constant currency, or 19% cc adjusted for a one-point in-period benefit in the prior year. On the restated basis it grew 15% cc in Q4, so this is a three-point constant-currency acceleration. The driver was ARPU: M365 Copilot, E5 and E7 on the seat side, with GitHub Copilot usage on the consumption side. E7, the suite combining Copilot, E5, Entra and Agent 365, has reached millions of paid seats across thousands of customers. The offset was execution. Management said sales of M365 E3 and stand-alone SKUs were weaker, and paid M365 commercial seats grew 5%, down from 7% in the restated Q4 figure even though that metric now includes paid GitHub seats.

"The number of Copilot customers actively consuming credits through usage-based billing grew over 75% over the quarter." — Satya Nadella, CEO

M365 consumer cloud grew 17% (16% cc), down sequentially as Microsoft begins to lap last year's price increase. The Q2 guide calls for M365 commercial cloud growth of about 19% cc and continued acceleration through the year, while consumer slows to the low teens.

Assessment: The revenue line did what the thesis needs it to do. Commercial cloud growth is accelerating in constant currency on ARPU, which is the mechanism we underwrote. The quality of the evidence, however, got worse. The 30-million-plus Copilot seat count disclosed in July was not updated, seat growth decelerated, and management attributed the miss in the core seat business to its own sales motion. We read the E3 weakness as a transition cost: a sales force told to sell E7 and consumption meters spends less time on the traditional renewal. That is a reasonable trade if ARPU keeps carrying growth, and the Q2 guide of about 19% says management expects it to.

Licensing, Industry Solutions and Support: $15.9B combined

Productivity and server licensing fell 3% (-2% cc) as the business lapped prior-year product launches, partly offset by contracts with more in-period recognition, and is guided to decline in the mid-to-high single digits in Q2. Industry solutions cloud, which combines Dynamics 365, LinkedIn Talent and Sales Solutions, and Healthcare and Life Sciences cloud, grew 10% against a high-single-digit guide, but management said bookings trends continued to moderate. Frontier and support services, formerly Enterprise and partner services, grew 4% (5% cc) on support.

Assessment: These lines are about a fifth of the segment and are not where the investment case lives. The one worth monitoring is Industry solutions, where Dynamics CRM has now shown moderating bookings for several quarters. Healthcare automation was the bright spot: 30 million patient encounters automated in the quarter, up 70%.

Devices and Consumer ($15.5B, -3%)

Windows OEM and Devices: $3.6B, -22%

Windows OEM and devices fell 22%, in line with the low-twenties decline management guided. Windows OEM alone fell 27% (28% cc), and management said that was below expectations. Three factors drove it: the prior-year Windows 10 end-of-support demand, OEM and channel partners drawing down the inventory they built ahead of component price increases, and weaker PC demand. Inventory remains elevated, and Q2 is guided to a decline in the mid-30s.

"Windows OEM revenue was below expectations and decreased 27% and 28% in constant currency, driven by the Windows 10 end of support demand in the prior year, inventory drawdowns by OEM and channel partners and weaker PC market demand." — Amy Hood, CFO

Assessment: In July we described two quarters of channel inventory builds as a borrowed benefit. This is the repayment, and the Q2 guide says it is not finished. Windows OEM and devices is now 3.8% of revenue, and its $1.0B decline cost the consolidated growth rate a little over a point. The strategic story, Windows as a "secure agent box" with Microsoft Execution Containers adopted by OpenAI Codex, Anthropic Claude Code and Meta Muse, is interesting and has no revenue yet.

Search and Advertising: $6.6B, +12% ex-TAC

Search and advertising, which now includes LinkedIn Marketing Solutions and Premium Subscriptions, grew 12% ex-TAC against a mid-to-high single-digit guide, on higher volume and rate across LinkedIn, Edge and Bing. Bing and Edge again took share. Q2 is guided to low double-digit growth.

"Search and advertising revenue ex-TAC increased 12% and was ahead of expectations, driven by higher volume and rate across LinkedIn, Edge and Bing." — Amy Hood, CFO

Assessment: The steadiest line in the segment. Combining LinkedIn's ad business with search creates a $25B annual advertising franchise on the FY26 restated figures, and it grew 14% ex-TAC in FY26 on the restated basis. It does not move the thesis, but it is the reason Devices and Consumer can hold its operating income while Windows contracts.

XBOX: $5.4B, -2%

XBOX revenue fell 2%, and content and services fell 2% (-1% cc), against a guided mid-single-digit decline. Third-party content and Game Pass declined, partly offset by first-party growth. Minecraft had a record quarter. Q2 content and services is guided up mid-to-high single digits on the holiday slate, with growth strongest in Q2 and slowing through Q4.

"Minecraft delivered a record quarter, reaching all-time highs in both engagement and revenue, while World of Warcraft had its largest beta in franchise history." — Satya Nadella, CEO

Assessment: Management committed in July to return XBOX to growth in FY27, and the Q2 guide is the first quarter that would deliver it. The segment's operating margin gain to 34.6% is mostly lower Activision amortization rather than an operating turn, and segment operating expense fell 2.8% on XBOX. The Game Pass decline is the line to monitor: it is the recurring part of the business.

Key KPIs

Investor Metrics on the Restated Basis (YoY Growth, Reported)

Operating and Platform KPIs

Two readings matter most. First, the Azure series shows the acceleration is concentrated in one quarter: 40%, 39%, 40% and 42% through FY26, then 48%. That is a step change, not a trend, and the Q2 guide of 49-50% asks investors to believe it holds. Second, the cash statement and the lease footnote now carry as much information about the capital program as the capex line does. Reported capex of $45.2B, $12.9B of other investing outflow, and a $35.5B quarterly increase in leases not yet commenced are three different measures of the same build, and only the first is in management's prepared remarks.

Key Topics & Management Commentary

Overall Management Tone: Confident, and unusually candid about volatility. Management led with delivery again, but it repeatedly named the factors that could make Azure growth uneven and told investors plainly that part of the Q2 guide depends on efficiency gains not yet achieved. The one area where the call was less forthcoming than the filings was the balance sheet: the call introduced external funding in one sentence, while the bond deal, the lease pipeline and the component outflows surfaced only in documents filed around it.

1. Azure at 48%, and the First Quarter Reported in Dollars

Azure revenue of $33.2B grew 48% against a recast guide of 44-45% cc, and the comparable figure for Q4 FY26 on the same definition was 42%. The acceleration happened while third-party capacity Microsoft had counted on for the quarter arrived late. Management's explanation was Foundry demand served from freed GPU capacity, better CPU monetization, and higher revenue per token.

"Azure revenue was ahead of expectations and increased 48%, even with the 3P capacity delays noted earlier, driven by Foundry and improved CPU monetization." — Amy Hood, CFO

The dollar disclosure is the structural change. The September recast restated two years of quarterly Azure revenue, and this is the first quarter reported on that basis. It shows Azure at $101.9B for FY26 and $33.2B for this quarter alone, an annualized run rate above $130B.

Assessment: This is the commitment we flagged as the first item to watch in July, and it was beaten. The 2.2x increase in sequential dollars added versus a year ago is the clearest evidence yet that demand, not just pricing or mix, is driving the line. The new dollar series also removes the most common objection to Microsoft's AI disclosure: a growth rate on an undisclosed base.

2. The 49-50% Guide Rests Partly on Efficiency Not Yet Found

December is guided to Azure growth of 49-50% in constant currency, above the Street's mid-to-high 40s. The guide assumes continued efficiency gains across the CPU and GPU fleet to offset third-party GPU-as-a-Service contracts that have moved into the second half. In Q&A, management said the team needs to find additional efficiencies to deliver it.

"In Azure, we expect revenue growth of 49% to 50% in constant currency with continued efficiency gains across our CPU and GPU fleet, helping to offset some 3P GPU-as-a-Service contracts that have been delayed to H2." — Amy Hood, CFO

The arithmetic deserves a closer look. Azure revenue was $24,129M in the December quarter of FY26. Growth of 49-50% implies roughly $35.95-36.19B this December, a sequential increase of about $2.8-3.0B. That is smaller than the $3.7B Microsoft added this quarter. The headline growth rate accelerates because last year's December quarter added only $1.7B. In dollars, the guide is a modest sequential slowdown.

Assessment: We do not read this as a warning. A $2.8-3.0B sequential add would still rank third in the nine quarters of restated history, behind only the last two, and the late third-party capacity should arrive in the second half. But it reframes the guide. Investors who extrapolate 48% to 50% as a demand curve are reading a comparison effect. Management also said that beyond Q2 growth will face a strong prior-year comparable that included accelerating growth in the second half. Expect the growth rate to become lumpier in calendar 2027 even if the dollars keep rising.

3. OpenAI Is 10.8% of Microsoft's Revenue

OpenAI is an equity-method investee in which Microsoft holds about 24% on an as-converted basis, down from about 25% in June. That makes OpenAI a related party, and Microsoft now discloses revenue from it. In the quarter that revenue, including revenue-sharing payments, was $10.0B, or 10.8% of total revenue. The FY26 annual report, the first filing with this disclosure, gave $24.1B for the full year, 7.3% of FY26 revenue. Accounts receivable from OpenAI rose to $7.2B from $6.0B in June. The quarterly filing for the March quarter carried no such figure.

"Secondly, and you mentioned it, token pricing will create a little bit of variability. We had higher token pricing than we expected in the quarter." — Amy Hood, CFO

Management also said Azure OpenAI benefited directly from big advances in OpenAI's models during the quarter. The two statements fit together: strong demand for new OpenAI models, served through Azure and through OpenAI's own consumption of Microsoft capacity, at higher revenue per token than planned.

Assessment: This is the most important new fact in the filing, and it received little attention in the first-day coverage we reviewed. At $10.0B a quarter, OpenAI revenue to Microsoft is running at roughly $40B a year against $24.1B in FY26, and it is now the counterparty behind more than a tenth of the company's revenue. In July we moved our OpenAI-concentration concern to contained because all sequential backlog growth came from other customers. That finding still holds for the backlog. But revenue concentration is the more direct measure, and it is rising. We do not think this is a crisis: OpenAI's demand is real and the contract runs to 2030 and beyond. We do think a customer at 11% of revenue, with a receivable that grew $1.2B in one quarter, belongs back on the risk list.

4. Microsoft Borrows: The $25 Billion Bond

In the prepared remarks, management added one sentence to its standard free-cash-flow line: Microsoft would access additional sources of funding when attractive. The next morning it launched a $25.0B senior unsecured offering in ten tranches maturing from 2028 to 2066. It priced that day, with the 40-year tranche at 105bps over Treasuries, and it settles November 4. Orders reportedly peaked near $115B. It is Microsoft's first US bond sale since 2017.

"To complement cash generated by the business and preserve financial flexibility, we will access additional sources of funding when we believe it is attractive to do so." — Amy Hood, CFO

The balance sheet context explains the timing. Cash and short-term investments fell $12.3B in the quarter to $64.5B. Total debt was $35.5B before the deal and becomes roughly $60.5B after it. Finance-lease liabilities, which are debt in all but name, reached $71.0B. Counting bonds and finance leases together, Microsoft carried $106.5B of obligations against $64.5B of cash and short-term investments before this deal priced.

Assessment: On the merits this is a sensible transaction. A AAA issuer that can borrow at 25-105bps over Treasuries should not fund a multi-year build entirely from cash, and the order book shows lenders agree. The signal is what matters for the equity. For three years the bull case has included the claim that Microsoft could fund the largest capital program in corporate history from operating cash flow. That claim ended this week. It does not change the thesis, but it changes the risk profile: the build is now financed with bonds and leases as well as cash flow.

5. Capex Came In Light Because Suppliers Were Late

Capital expenditures including finance leases were $45.2B against a guide of more than $50B. The reason was supply, not restraint: third-party capacity deliveries expected in Q1 were delayed. About three-quarters of the spend went to short-lived assets, mainly CPUs and GPUs, up from about two-thirds last quarter, and management noted higher memory and storage costs within that. Calendar-2026 capex is now expected at approximately $170B, down from approximately $175B in July.

"Capital expenditures were $45.2 billion as some third-party capacity deliveries we expected in Q1 were delayed." — Amy Hood, CFO

"Therefore, for the calendar year, spend should be approximately $170 billion due to the 3P capacity delays noted earlier." — Amy Hood, CFO

Q2 is guided to more than $50B, and management expects spending to accelerate in the second half of the fiscal year from CPUs, GPUs, memory and storage costs, and the delayed third-party deliveries.

Assessment: The calendar-year figure has now moved twice in two quarters for two different reasons: an accounting change in July and a supplier delay in October. Neither is a change in intent. Investors should treat roughly $175-190B as the economic run rate and the reported figure as a function of lease classification and delivery timing. The shift toward short-lived assets is double-edged. It makes the program easier to slow if demand weakens, which is the downside protection management described in July. It also means a larger share of the fleet depreciates over two to six years rather than over decades.

6. The Lease Pipeline Reaches $364.6 Billion

The capital program has a second, larger number that never appears in the capex line. At September 30, Microsoft had additional leases, mainly for datacenters, that had not yet commenced totaling $364.6B, with terms of one to 20 years starting between FY27 and FY34. The figure was $329.1B in June and $196.6B in March: $168.0B added in six months.

"As a reminder, we recognize the full principal value of finance leases and CapEx at the time of lease commencement." — Amy Hood, CFO

The quarter also shows the July useful-life change working through the lease accounting. Right-of-use assets obtained under new operating leases rose to $6.5B from $1.4B a year ago, while those obtained under new finance leases fell 43% to $5.2B. Operating-lease liabilities rose $5.6B in a single quarter to $27.5B. Operating leases sit outside reported capex, so every lease that moves from finance to operating treatment lowers the headline capital figure without changing the commitment.

Assessment: In July we asked what replaces the reclassified finance leases on the balance sheet. Part of the answer is now visible: the mix of new leases flipped from mostly finance to mostly operating in one quarter. The larger answer is the $364.6B pipeline, which equals roughly two years of reported capex at the current pace and was not discussed on the call. Not all of it will become capex, since some leases will be classified as operating. But it is the most complete measure of what Microsoft has already committed to build, and it grew by $35.5B this quarter.

7. Free Cash Flow Falls 27%, and a Component Line Sits Below It

Operating cash flow grew 29% to $58.0B on cloud billings and collections. Cash capex grew 103% to $39.4B. Free cash flow fell 27.4% to $18.6B. The FY27 guide remains only that Microsoft will stay free cash flow positive.

"Cash flow from operations was $58 billion, up 29%, driven by strong cloud billings and collections. And free cash flow was $18.6 billion, reflecting higher capital expenditures." — Amy Hood, CFO

The investing section holds a second outflow. "Other" investing cash was $12.9B, up from $6.2B a year ago, and the quarterly filing attributes the $6.7B increase primarily to facilitating the purchase of components. This sits outside additions to property and equipment, so it does not reduce free cash flow as Microsoft and most investors define it. Operating cash flow also absorbed a $6.4B increase in other long-term assets.

Assessment: The quarter tracks the FY27 decline our model already assumes: free cash flow of about $52.5B for the year, down from $67.0B. What is new is the component line. If Microsoft is now prepaying or financing suppliers to secure memory and other parts, that cash is part of the capital program whether or not it is labeled capex. Free cash flow after the full "other" investing outflow would have been $5.7B this quarter. That is not the right measure either, because the line contains other items, but the gap between $18.6B and $5.7B is the reason we now read the investing section in full every quarter.

8. A 50% Operating Margin, and a Voluntary Exit Offer

Operating margin reached 49.8%, up 89bps, in a quarter when gross margin fell 126bps. The lever was operating expense, which grew 6.2% against revenue growth of 18.8%. Sales and marketing was flat in dollars. R&D grew 11.1%, driven by compute capacity and talent. Total headcount declined 2%.

"This quarter, in particular, when you're delivering a lot of efficiency gains in a quarter and you can deliver revenue growth without spending additional costs, you, of course, see a benefit, which we did." — Amy Hood, CFO

Q2 guidance includes an estimate for a recently announced voluntary exit offer in both COGS and operating expense. Operating expense is guided to $18.5-18.6B, up 9%, and operating margin to be down slightly year-over-year. At the guidance midpoints, Q2 operating margin is about 46.8% against 47.1% a year ago.

Assessment: Microsoft is running a 50% operating margin while growing revenue 19% and absorbing the largest depreciation ramp in its history. That is the strongest available rebuttal to the claim that AI infrastructure must dilute margins. The voluntary exit offer and the 2% headcount decline say management intends to keep financing the build partly with operating expense discipline. The FY27 frame from July, operating margin down less than a point, was not repeated or changed. After a quarter 89bps above the prior year, it now looks conservative.

9. Copilot: Usage Up, Seat Count Withdrawn

The usage disclosures were strong. Copilot daily users rose nearly 50% sequentially and more than fivefold year-over-year. Automated tasks nearly doubled since July. Satisfaction rose nearly 20% to an all-time high. Customers with more than 50,000 seats rose more than eightfold, and Volkswagen plans to deploy to more than 330,000 employees. The number that anchored the franchise in July, more than 30 million paid M365 Copilot seats, was not updated, and management confirmed the omission when asked directly.

"And we did not disclose to answer your direct question, an update on the Copilot number this quarter. We talked about momentum in revenue and usage." — Amy Hood, CFO

The context is a business-model shift. Management described Copilot as a subscription funded by IT plus consumption meters funded by business units, and said the move to consumption will be faster in coding than in knowledge work.

Assessment: Withdrawing a metric in a strong quarter is rarely a good sign, even when there is a reasonable explanation. The likely explanation is that seats are becoming a less meaningful unit as revenue shifts toward E7 bundles and usage, and that is consistent with the ARPU-led acceleration in M365 commercial cloud. But three consecutive quarters of rising seat counts were the most concrete evidence the franchise had. Replacing them with sequential growth rates on undisclosed bases is a step backward in exactly the area where we asked for more.

10. Bookings Fall 12% While the Backlog Holds

Commercial bookings fell 12% (11% cc) against a prior-year quarter that included very large Azure commitments. Management said underlying commercial sales motions grew, helped by one large Azure commitment signed this quarter. Commercial RPO was $677B, up 73% year-over-year and roughly flat against $678B in June, a smaller sequential decline than seasonal patterns would suggest. About 30% will be recognized in the next twelve months, up 29%, and the longer-dated portion grew 98%.

"Commercial bookings decreased 12% and 11% in constant currency, driven by the impact from large Azure commitments in the prior year, partially offset by healthy growth in our commercial sales motions, including a large Azure commitment this quarter." — Amy Hood, CFO

Assessment: The bookings decline is a comparison effect and management flagged it in advance. The RPO growth rates are the cleaner trend, and they are slowing from very high levels: total RPO growth eased to 73% from 84% in July, near-term RPO growth to 29% from 37%, and long-dated growth to 98% from 112%. That is what normalization after a year of record contracts looks like. The unnamed large Azure commitment this quarter is the detail we would most like to know more about.

11. The MAI and Maia Cost Curve

Microsoft set a public target for its first-party models: match leading models within three to six months of their release at a fraction of the training cost. The supporting data were specific. MAI-Cyber now powers more than 40% of scans in the MDASH code scanner at 18% lower cost per scan. MAI Code Flash had 3 million users in GitHub Copilot last month. MAI-Base-2 completed pretraining with 4-12x greater training compute efficiency than other open-weight models tested. On silicon, Maia 200 delivers up to 1.7x more tokens per watt in Teams and Copilot, and Maia 300 is expected to deliver 40% more tokens per watt than leading GPU-based systems. Microsoft also docked its first NVIDIA Vera Rubin GPUs.

"Our goal is to advance the cost-to-outcome frontier, which means delivering capabilities comparable to leading models within 3 to 6 months of their release at a fraction of the training cost, thereby allowing us to transfer those cost benefits to customers." — Satya Nadella, CEO

Assessment: The three-to-six-month target is the most concrete statement Microsoft has made about its own model ambitions, and it is deliberately positioned behind the frontier rather than at it. That is the right strategy for a company whose margin depends on routing work to the cheapest model that is good enough: 90% of Copilot queries now use Auto mode, which routes across models. The remaining gap is unchanged from last quarter. We still do not know what share of inference runs on Maia or on MAI models, and that share determines whether this is a material margin lever.

Guidance & Outlook

The Q2 guide is above the Street on every headline measure. Revenue of $96.0-97.1B compares with roughly $95.6B, and Azure at 49-50% compares with consensus in the mid-to-high 40s. The cost guide is the more interesting part. COGS growth of 26-27% outruns revenue growth by eight points as depreciation and the Azure mix bite, and operating expense growth steps up to 9% from 6% partly because of the exit offer.

Implied sequential ramp: The revenue midpoint of $96.55B is 4.6% above Q1, and Agents and Infra is guided up about 4.7% sequentially. As noted above, the Azure growth guide implies a sequential dollar increase of about $2.8-3.0B, below Q1's $3.7B.

Street at: Consensus for the December quarter sat near $95.6B for revenue before the print. The guide midpoint is about $1B higher, and the Azure guide is a few points above where most models sat.

Guidance style: Specific on revenue and costs, explicit about volatility, and still vague on capital. Management gave segment ranges and product-level growth for every line, and for the first time told investors which factors drive Azure variability. It updated its calendar-year capex figure again but, for the fifth straight call, gave no fiscal-year figure, and it left the free cash flow guide at "positive." The pattern from July continues: precision where the numbers are favorable, ranges or words where they are contested.

Analyst Q&A Highlights

Where the Return on New Capacity Comes From

The opening question asked whether returns on new AI capacity will come mainly from better utilization and unit costs inside Azure, or from selling more layers of each workload, such as Fabric, Foundry and Copilot. The answer refused the choice. Management described one fungible pool of capacity allocated to the highest-value demand, and argued that a single customer workload now touches Copilot, WorkIQ, Foundry, a Postgres database, Agent 365 and the security layer at once.

Q: "Can I ask, as incremental AI capacity comes online, what do you see as likely to be the more important driver of returns over the next few years, improving utilization and unit costs within Azure or your ability to monetize more layers of each workload across Fabric, Foundry and Copilot?"
— Adam Wood, Morgan Stanley

A: "But your core question of where is the margin going to come from, where the leverage is going to come from. It's not going to come from any -- it's going to come from every one of them, and it compounds when we work them across."
— Satya Nadella, CEO

Assessment: The answer explains the segment change better than the September presentation did. If every workload runs across several product layers, a structure that reported Azure and Office separately was hiding the economics. The cost is that the new structure makes it harder to see which layer earns the return, since 83% of revenue now sits in one segment.

What Has to Go Right for 49-50% Azure Growth

A direct request to unpack the December Azure guide. Management walked through Q1 first: platform efficiency gains across apps, Foundry and the Azure platform; higher token pricing than expected; and capacity found in the CPU, storage and memory fleet. It then said the Q2 guide assumes those gains continue and that the team finds more, because supply expected in Q2 moved to the second half. It also flagged token pricing as an unknown.

Q: "Amy and Satya, I'm wondering if you could unpack what I thought was one of the key metrics from this call, and that is the guidance for Azure growth of 49% to 50% constant currency in December."
— Karl Keirstead, UBS

A: "So in Q2, we've got those efficiency gains, which is great, and those will continue into Q2. We need to find some more, which we're counting on, enable to get to that guidance that I gave because I have faith in the team's work in Q1 continuing into Q2."
— Amy Hood, CFO

Assessment: The most candid answer on the call. Management guided to a number that requires efficiency gains not yet achieved, and said so. That raises the execution risk on the guide somewhat, but this team has beaten its Azure guide in each of the last three quarters, and efficiency work compounds on an installed base that grows every quarter.

Enterprise Breadth Versus a Few Large Buyers

A question on how much of the AI strength comes from enterprises rather than AI-native companies. The answer cited named enterprise deployments in Europe, including order-taking systems, jet-engine maintenance planning and more than 20 end-to-end banking processes, several of them already moving from frontier models to cheaper models trained on the customer's own data. Management then drew a line between enterprise breadth and concentrated capacity buyers.

Q: "I'd like to get a sense on the drivers of the AI strength in Azure being from enterprise versus digitally native in general or AI native."
— Mark Moerdler, Bernstein

A: "But like everything else, the enterprise stuff is not like a quarter-over-quarter. It's a decade-long thing, and that's what I've been waiting for as opposed to like 4 customers who buy a lot of capacity from you and you can show a few quarters of growth. And that's not a business as far as I'm concerned."
— Satya Nadella, CEO

Assessment: A strong statement of strategy, and an awkward one to read next to the filing that shows a single customer at 10.8% of revenue. Management added that it will sell capacity to large buyers if the pricing is right. We take that at face value: Microsoft is not refusing concentrated demand, it is declining to chase it on poor terms. The disclosure gap is that investors cannot yet see how much of Azure growth comes from the breadth Microsoft says it prefers.

OpenAI's Share and How Token Value Will Shift

The only question to cite the new OpenAI revenue figure directly. It asked how token mix and value will shift between frontier models, Microsoft's own MAI models and enterprises building proprietary models. The answer described a toolbox: customers adopt a new frontier model when it leads, route across several frontier models by price for each task, and eventually train their own models from their traces. Management argued these choices change over time for the same workload.

Q: "At the same time, you have OpenAI at 11% of revenue in the quarter. So maybe just take a step back, can you give us your latest thinking on how you think token mix and token value is going to evolve between the frontier MAI and maybe what some of your more sophisticated enterprise customers are doing from a proprietary AI strategy standpoint?"
— Gabriela Borges, Goldman Sachs

A: "It's like I may start at the frontier, I may go to an open-weight, I may go to an MAI model. I may go back to the frontier for the next drop for a few months, come out."
— Satya Nadella, CEO

Assessment: Management did not address the 11% figure, and nobody pressed it. The answer is a coherent argument that Microsoft benefits whichever model wins, because it hosts all of them. It is less clear that it benefits equally: frontier model demand carried higher token pricing this quarter, and that demand is concentrated in a small number of providers. The question deserved a direct answer on concentration and did not get one.

Sustaining a 50% Operating Margin

With Azure growth and operating margin both near 50%, the question asked how both can hold. Management said the starting point is revenue growth rather than cost, then described efficiency as operational quality (reliability, security and the lowest cost per unit) rather than cost cutting. It credited product quality in the quarter and a deliberate choice to concentrate operating expense on new products, security and quality.

Q: "I think many ask how can you sustain this level of growth with 50% in the top and a 50% in the bottom?"
— Brent Thill, Jefferies

A: "Let me first say, best solution here is revenue growth. And so I would say I tend to start and focus there, even though I think many people seem to think I start at the bottom of the P&L and work up."
— Amy Hood, CFO

Assessment: A framework answer, not a commitment. Management did not say the 50% margin would hold, and the Q2 guide of a slight year-over-year decline says it will not in the next quarter. The useful detail was the admission that this quarter's margin benefited from efficiency gains that delivered revenue without additional cost. Those gains are lumpy by management's own description, so the margin will be too.

How Fast Consumption Spreads Beyond Coding

The question asked whether consumption revenue in core M365 will arrive as quickly as it did at GitHub, where the move from seats to usage unlocked rapid growth, or whether it will take significant sales effort. Both executives said it will be slower outside coding, where pay-as-you-go is not yet the norm. Management described a dual model of an improving subscription plus consumption meters, and cited investment in a forward-deployed engineering organization and new sales motions to teach customers to use both.

Q: "As you think about the tools in place now unlocked on the consumption side for the core M365 customer, do you think the unlocks is fast and without sales effort? Or is it going to take a fair amount of effort to get those meters spinning?"
— Brad Reback, Stifel

A: "And I think to Satya's point, it will be a slower move outside of coding. But I also still think we're still seeing encouraging signs, and I feel positive about that."
— Amy Hood, CFO

Assessment: An honest calibration, and it connects to the weak E3 execution disclosed in the prepared remarks. The sales force is being retooled to sell a two-part model, and the traditional seat business is paying for the transition. Investors modeling M365 consumption on the GitHub curve should slow the ramp.

Funding the Build: Leases, Debt and Other Options

The closing question asked how Microsoft would raise capital and in what mix of debt, equity and other sources. Management listed leasing as the vehicle already in use, noted access to the debt markets given its credit ratings, and mentioned other options without detail. It framed the decision around attractiveness and confidence in returns on capex. The same exchange confirmed that the Copilot seat count was not updated.

Q: "Did you disclose the Copilot numbers this quarter? But my question is more about the capital commitment or the capital strategy."
— Tal Liani, Bank of America

A: "I've talked about this before. There's a form of financing called leasing, and we've been using that one as a vehicle. There's also -- we're fortunate to have access in the debt markets given that where our ratings are and sort of we haven't been able to -- we haven't yet, I would say, tapped that."
— Amy Hood, CFO

Assessment: The answer was accurate on the evening of October 28, and the $25B offering launched the next morning. Management was not obliged to pre-announce a deal on the call, but the juxtaposition matters for how investors read the next "free cash flow positive" guide. The most useful word in the answer is "leasing": management named it first, and the lease footnote shows it is already the largest funding source for the build.

What They're NOT Saying

1. The paid Copilot seat count: More than 30 million in July and not updated this quarter, confirmed when asked. Usage growth rates on undisclosed bases replace the one hard number the franchise had.
2. Anything about OpenAI at 10.8% of revenue: The figure is in the quarterly filing, an analyst cited it, and management did not address concentration, the receivable, or how much of Azure growth it represents.
3. The size or timing of the bond deal: The call introduced external funding in one sentence. The $25B offering launched the next morning.
4. The $364.6B lease pipeline: Leases not yet commenced grew $35.5B in the quarter and $168.0B in six months. Neither figure was mentioned on the call.
5. What the component outflow is: "Other" investing cash of $12.9B, with the increase attributed mainly to facilitating component purchases, was not discussed. Investors do not know whether these are prepayments, supplier financing or something else, or whether they will recur.
6. The depreciation effect of the useful-life change: Still described only as minimal, and the quarterly filing gives no figure. Depreciation rose 56% to $11.1B, so even a small percentage effect is a meaningful dollar amount.
7. FY27 capex in dollars: The fifth consecutive call without a fiscal-year capex figure. The calendar-year figure that replaced it has now moved for two different reasons: lease reclassification and delivery timing.
8. The large Azure commitment this quarter: It held RPO flat and offset part of the bookings decline. Neither the counterparty nor the size was disclosed.
9. The cost of the voluntary exit offer: Included in Q2 COGS and opex guidance but not quantified.
10. The AI share of Azure: Azure now has a dollar figure but still no split between AI and core cloud services, in a quarter when management said growth was broad across both.

Market Reaction

• Pre-print setup: MSFT closed October 28 at $531.01, up 9.8% year to date against an S&P 500 up 13.1%, up 4.3% over the prior 30 days, and down 2.0% over twelve months. It entered the print 2.3% below its 52-week closing high of $543.45 and 50.5% above its 52-week closing low of $352.83. The stock had gained 17.7% since the July 30 close of $451.10 that followed the FY26 Q4 print.
• After-hours move (October 28): Shares rose roughly 3-4% after the release and through the call, briefly trading near $550, on the Azure acceleration and the Q2 guide.
• Reaction session (October 29): The stock opened up 1.7% at $540.22, traded between $519.56 and $541.29, and closed at $530.14, down 0.2% or $0.87, against an S&P 500 down 0.3%. Volume of 39.3M shares was 1.8 times the 30-day average of 21.4M.

Priced for the beat. A stock that had rallied nearly 18% in three months into the print, and sat within a few percent of its high, needed more than a beat to move. The Q2 Azure guide did move it after hours, and the reaction session gave the move back. Measured against the S&P 500, the stock finished essentially flat on volume well above normal, which is what a fully anticipated beat looks like.

The bond overhang. The fade began on the morning Microsoft launched its $25B bond deal, and coverage that day focused on capital spending growing faster than cash flow. We think the timing matters more than the debt itself. A company that guided free cash flow only to "positive" and then borrowed $25B within a day invites investors to ask how much of the build the business can fund internally, and the quarter's 27% decline in free cash flow made that question easy to ask. The rally into the print had priced the operating story, and the bond priced a little of the funding story back in.

The session's range is a useful summary of the debate. The open at $540.22 reflected the operating results. The low of $519.56, 2.2% below the pre-print close, reflected the funding questions. The close near the midpoint says neither side won on the day.

Street Perspective

Debate: Is the Bond Deal a Warning or an Optimization?

Bull view: A AAA issuer borrowing at 25-105bps over Treasuries, with demand several times the deal size, is optimizing its capital structure. Microsoft's credit spreads are a fraction of other hyperscalers', and long-dated debt at these terms is cheap funding for assets that management says earn high returns. Holding cash while competitors borrow would be the strange choice.

Bear view: Microsoft has not issued a US bond since 2017, and it chose to start the day after reporting a 27% decline in free cash flow and a $12.3B drop in cash and short-term investments. Combined with $71.0B of finance-lease liabilities and $364.6B of leases not yet commenced, the bond confirms that the build has outgrown internal funding.

Our take: Both are true, and they are not in conflict. The deal is good treasury management and a meaningful change in the investment case at the same time. We would not sell the stock on the bond. We do add it to the risk list, because the equity story no longer rests on self-funding, and the next downturn in AI demand would now meet a balance sheet carrying more fixed obligations than it did a year ago.

Debate: Does OpenAI at 11% of Revenue Change the Risk Profile?

Bull view: OpenAI revenue is a sign of strength. Its models drove higher token pricing, the relationship is contracted through the next decade, Microsoft owns about a quarter of OpenAI, and all sequential backlog growth in July came from other customers. A strong customer buying more is good news.

Bear view: One counterparty now supplies more than a tenth of revenue, its revenue to Microsoft is running at well above the FY26 pace, and its receivable is growing. OpenAI is a cash-consuming company that depends on continuous external financing, and the value of Microsoft's equity stake moves with the same risk.

Our take: The bear view has the better of the argument on disclosure and the bull view on near-term fundamentals. We see no sign of stress in the relationship, and the demand is real. But our July conclusion that OpenAI concentration was contained rested on backlog data. The income statement now shows concentration rising, so we are moving the concern back to emerging and will watch the receivable as closely as the revenue.

Debate: Is 49-50% Azure Growth a Demand Curve or a Comparison?

Bull view: Azure accelerated six points in one quarter on the restated basis, beat its guide despite late supply, and is guided higher again. Demand still exceeds supply, the delayed third-party capacity arrives in the second half, and efficiency gains add revenue without new capex.

Bear view: The guide implies a smaller sequential dollar increase than this quarter. Part of Q1's beat came from token pricing and found capacity, which management called unpredictable. Management itself warned of a strong prior-year comparable after Q2.

Our take: The bulls own the level and the bears own the trajectory. Q1 was the largest sequential dollar increase in the restated Azure series, and nothing in the guide contradicts continued dollar growth. But the 48-to-50% progression is partly a comparison effect, and we expect the growth rate to look uneven in calendar 2027 even as the dollars rise. Investors should track sequential dollars, which Microsoft now discloses, rather than the growth rate.

Model Update Needed

Valuation impact: Our FY27 estimates rise by roughly 1% on revenue and 1-2% on EPS, offset by a share price that has risen 17.5% since our July upgrade. At the October 29 close of $530.14 the stock trades at about 27x our updated FY27 EPS of roughly $19.60-19.90, against 23-24x when we upgraded, and about 23x the FY28 consensus we carried in September. That is a full but defensible multiple for revenue growing 18-19% with an operating margin in the high 40s and Azure accelerating. Two items push the other way in our view of intrinsic value. Interest expense now includes $25.0B of new bonds, which add roughly $1.3B a year on the fixed-rate tranches alone, and the lease pipeline implies a larger future depreciation and interest burden than the capex line shows. Net effect: modestly positive on earnings, neutral on intrinsic value after the move, and the upside from here depends on Azure sustaining dollar growth rather than on multiple expansion.

Thesis Scorecard Post-Earnings

Overall: The operating thesis strengthened and the financial risk profile weakened. All five bull pillars were confirmed, and the first, Azure compounding, had its best quarter of the cycle. On the bear side, the same quarter produced three escalations and one new point, and none of them is about demand: OpenAI concentration is visible in revenue, consumption visibility worsened when the seat count disappeared, the lease reclassification is now visible in the numbers, and the build is now financed partly with debt. In July the remaining complaint was disclosure quality. This quarter disclosure improved materially, with Azure dollars and OpenAI revenue, and what it revealed is a business with stronger operations and more financial leverage than we had appreciated.

Action: Maintaining Outperform. In July we said we would revisit toward Hold if the FY27 capex figure landed materially above the reclassified run rate, if Azure missed the committed first-half acceleration, or if Microsoft Cloud gross margin fell back below 64%. None of the three happened: calendar-2026 capex fell to roughly $170B, Azure accelerated to 48% with 49-50% guided, and Cloud gross margin printed 66%. The conditions we set in public were not met, so the rating stays. Our conviction is lower than in July for two reasons: the stock has absorbed much of the upside we identified (about 27x FY27 EPS against 23-24x), and the funding mix has moved toward bonds and leases. We would move toward Hold on: Azure sequential dollar adds falling below $2.5B in a quarter, OpenAI revenue concentration rising further while its receivable grows faster than its revenue, or FY27 free cash flow tracking below $45B. Monitor: (1) Q2 Azure against the 49-50% guide and the implied $2.8-3.0B sequential add; (2) OpenAI revenue and receivable each quarter; (3) leases not yet commenced and the operating-lease share of new leases; (4) the "other" investing line and any disclosure of component prepayments; (5) whether any Copilot seat or revenue figure returns; (6) XBOX content and services returning to growth in Q2 as guided.