tapebrief

MSFT · Q2 2026 Earnings

Cautious

Microsoft

Reported January 28, 2026

30-second summary

30-second take. Microsoft printed Q2 revenue of $81.3B (+17% YoY), Azure growth ticking down to +39%, and commercial RPO ballooning to $625B (+110% YoY). The eye-catcher: Hood disclosed in Q&A that Azure growth "would have been over 40" had all H1 GPUs been allocated to Azure — management is explicitly trading Azure-line growth for higher-LTV deployments in M365 Copilot, GitHub Copilot, and product R&D. Operating margin came in at 47% ahead of expectations, and Hood now expects FY26 operating margins up slightly. Customer-concentration disclosure also tightened materially, with the OpenAI share of commercial RPO now quantified at ~45%.

Headline numbers

EPS

Q2 FY2026

$4.14

Revenue

Q2 FY2026

$81.27B

+17.0% YoY

Gross margin

Q2 FY2026

68.0%

Free cash flow

Q2 FY2026

$5.88B

Operating margin

Q2 FY2026

47.1%

Key financials

Q2 FY2026
MetricQ2 FY2026YoYQ1 FY2026QoQ
Revenue$81.27B+17.0%$77.70B+4.6%
EPS$4.14$4.13+0.2%
Gross margin68.0%69.0%-100bps
Operating margin47.1%48.9%-180bps
Free cash flow$5.88B$25.70B-77.1%

Guidance

No quantitative guidance provided; Microsoft deferred forward guidance to earnings call.

No quantitative guidance provided; Microsoft deferred forward guidance to earnings call.

Segment performance

Q2 FY2026
SegmentQ2 FY2026YoY
Productivity and Business Processes$34.116B+16.0%
Intelligent Cloud$32.907B+29.0%
More Personal Computing$14.25B-3.0%
Microsoft Cloud$51.5B+26.0%

Platform metrics

Q2 FY2026
SegmentQ2 FY2026
Commercial remaining performance obligation$625 billion
Microsoft Cloud YoY Growth26%
Azure YoY Growth39%
Dynamics 365 YoY Growth19%

Profitability

Q2 FY2026
SegmentQ2 FY2026
Operating Margin47.1%
Gross Margin68.0%
Operating Cash Flow$35.758 billion

Other KPIs

Q2 FY2026
SegmentQ2 FY2026
Shareholder Returns$12.7 billion

Management tone

The framing of Azure growth has shifted from supply-constrained-but-accelerating to deliberately-allocated. Hood directly stated in response to Weiss that if she had taken the GPUs that came online in Q1 and Q2 and allocated them all to Azure, "the KPI would have been over 40." The deceleration relative to a hypothetical full-allocation scenario is a management choice, not a demand signal. Nadella's follow-on framed the choice as LTV portfolio optimization across Azure, M365 Copilot, GitHub Copilot, Dragon Copilot, and Security Copilot rather than headline Azure maximization — signaling management is preparing investors for an Azure line that may not be the right single metric to optimize.

Customer-concentration disclosure has tightened materially. Faced with a $625B RPO, Hood quantified the OpenAI share at ~45% (~$281B) and emphasized that the remaining 55% (~$344B) grew +28% YoY. The shift signals management has accepted that the OpenAI exposure is too large to wave off and is instead anchoring confidence in the diversified balance — a defensible breakout, but a notable concession that the question matters.

The duration-coverage argument has hardened. Hood's response to Moerdler that "the majority of the capital that we're spending today and a lot of the GPUs that we're buying are already contracted for most of their useful life" — with the longest-duration contracts going to the largest customers — is a stronger claim than prior framing, and directly addresses the 6-year hardware life vs. ~2-year RPO duration concern. The simultaneous extension of RPO duration from ~2.0 to 2.5 years reinforces the message — customers are committing for longer, not shorter. Nadella added that software optimization continuously extends competitive life of the aging fleet, and Hood noted that margins actually improve with time on already-sold capacity as efficiency gains compound.

The FY26 operating margin framing — Hood now expects margins up slightly — is the most material forward-looking posture, citing H1 prioritization work and a favorable mix shift toward Windows OEM and commercial on-prem.

Q&A highlights

Keith Weiss · Morgan Stanley

Earnings growing 24% YoY with margin expansion, yet stock trading down after hours. Investor concern centers on CapEx growing faster than expected and Azure growth potentially slower than expected. Request for clarity on capacity expansion yields, Azure growth trajectory, and ROI on CapEx investments.

Amy Hood explained CapEx allocation is not solely for Azure but distributed across M365 Copilot, GitHub Copilot, first-party apps, and R&D/product innovation. Satya added that if all Q1-Q2 GPUs had been allocated to Azure, growth KPI would exceed 40%. Emphasized building best lifetime value portfolio across multiple businesses (Azure, M365, GitHub, Dragon, Security Copilots) rather than maximizing one business. Noted compute is also R&D investment.

CapEx allocation: M365 Copilot, GitHub Copilot, Dragon Copilot, Security Copilot, R&D, then remainder to AzureAzure growth KPI would exceed 40 if all Q1-Q2 GPUs allocated to AzureOptimizing for long-term lifetime value portfolio across multiple businessesGPUs and capacity allocated to R&D and product innovation teams

Mark Mortler · Bernstein Research

Question on matching hardware CapEx investment (capitalized over 6 years) to revenue capture and margins. RPO duration increased from 2 to 2.5 years. Seeks comfort that AI-centric CapEx will generate sufficient revenue over 6-year hardware useful life to deliver solid revenue and gross profit growth comparable to CPU revenue trends.

Amy Hood stated majority of current GPU CapEx is already contracted for the entirety of useful life, mitigating duration/revenue capture risk. Mark noted Azure-only RPO is longer duration and GPU contracts for largest customers sold for entire useful life. Satya added that software optimization allows continuous running of latest models on aging fleet, and margins improve over useful life as efficiency gains compound.

Majority of capital spending and GPUs already contracted for useful lifeAzure RPO more extended than average due to longer-dated contractsGPU contracts for largest customers sold for entire useful lifeSoftware optimization enables efficiency improvements and margin expansion over asset life

Brent Thill · Jefferies

OpenAI represents 45% of commercial RPO balance ($625B total, implying ~$281B from OpenAI). Investor concern about durability and exposure. Request for commentary on perspective and health of OpenAI partnership.

Amy Hood reframed question to emphasize the remaining 55% (~$350B) RPO from diverse customer portfolio across solutions, industries, and geographies. Highlighted that non-OpenAI RPO grew 28%, demonstrating breadth and high confidence in durability. Characterized OpenAI as great partnership with continued confidence; noted Microsoft remains their scale provider and benefits from being cutting edge in app innovation.

45% of RPO ($281B estimated) from OpenAI55% of RPO ($350B) from diversified customer portfolioNon-OpenAI RPO grew 28% YoYGrowth consistent across customer segment, industry, and geography

Carl Kierstedt · UBS

Q2 added nearly 1 gigawatt of capacity, suggesting accelerating capacity additions. Requests detail on magnitude of upcoming capacity builds across Atlanta, Wisconsin, and other Fairwater sites, and how investors should think about capacity additions in coming quarters.

Amy Hood declined to focus on specific locations (Atlanta, Wisconsin) noting they represent multi-year delivery timelines. Emphasized Microsoft is working to add capacity globally, with substantial U.S. additions, to meet customer demand and usage increases. Stated focus is on securing power, land, facilities, then provisioning GPUs/CPUs as quickly as possible while optimizing operational efficiency. Did not quantify forward capacity additions.

Q2 added nearly 1 gigawatt capacityFairwater sites are multi-year delivery projectsCapacity being added globally with majority in U.S.Focus on power, land, facilities availability

Mark Murphy · JP Morgan

Maya 200 accelerator performance is remarkable relative to competitor solutions (TPUs, Tranium, Blackwell). Request for perspective on how much silicon development might become core competency for Microsoft and implications for gross margin profile, particularly for inference costs.

Satya emphasized multi-year silicon development history and excitement about Maya 200 progress, particularly FP4 precision performance running GPT-5.2. Explained innovation is end-to-end between model and silicon, optimizing systems including networking and memory. Stated preference for fleet flexibility across NVIDIA, AMD, and custom silicon to ensure best TCO at any time. Clarified custom silicon doesn't mean exclusive use; emphasized multi-generation competitiveness requirement.

Maya 200 delivers 10+ petaflops at FP4 precision with 30%+ improved TCO vs latest generationMulti-year history of proprietary silicon developmentInnovation spans model, silicon, networking, memory optimization at system levelFleet approach: diversified silicon sources (NVIDIA, AMD, Maya, Cobalt, DPU, NICs)

Answers to last quarter's watch list

Azure deceleration risk if capacity eases — Azure printed +39% (+38% CC), slightly ahead of the ~+37% CC Q2 guide. Hood disclosed the level was a deliberate allocation outcome — Azure would have grown over 40% had all H1 GPUs been allocated there. Supply constraints persist, but prioritization toward M365 Copilot, GitHub Copilot, and R&D is now visibly capping the Azure line.
Resolved negatively
Operating margin durability — Q2 operating margin came in at 47%, ahead of expectations, and Hood now expects FY26 operating margins up slightly.
Resolved positively
RPO duration disclosure as recurring metric — Hood disclosed commercial RPO weighted-average duration of 2.5 years, up from ~2.0 years last quarter. The metric has become a standing disclosure and is lengthening.
Resolved positively
More Personal Computing trajectory — MPC came in at -3%, with Xbox content and services -5% the main drag. Q3 guide of $12.3–12.8B implies continued pressure as Windows 10 end-of-support benefit normalizes.
Resolved negatively
CapEx trajectory — Q2 CapEx of $37.5B with ~1 gigawatt of capacity added; roughly two-thirds on short-lived GPU/CPU assets. Hood guided CapEx to decrease sequentially in Q3 due to normal variability and finance-lease timing, and emphasized majority of GPU spend is already contracted for useful life.
Continue monitoring

What to watch into next quarter

Azure growth allocation framing: Hood's "would have been over 40" disclosure is a new framework — watch whether management quantifies foregone Azure growth quarterly, or whether the headline KPI starts to be de-emphasized in favor of consolidated Microsoft Cloud growth.

OpenAI RPO share trajectory: at ~45% of $625B today; watch whether the non-OpenAI RPO sustains its +28% YoY growth pace, the cleanest durability signal stripped of the OpenAI commitment.

More Personal Computing trajectory: Q3 guide of $12.3–12.8B with Windows OEM expected to decline ~10% as Windows 10 end-of-support benefit normalizes. Watch memory pricing impact on the PC market.

FY26 operating margin durability: Hood now expects FY26 operating margins up slightly. Watch whether Q3's "down slightly YoY" operating margin guide and Q4 dynamics support that full-year frame.

Capacity additions cadence: Q2 added ~1GW; Hood declined to quantify forward adds. Watch whether quarterly GW disclosure becomes standing and whether the pace accelerates as Atlanta and Wisconsin Fairwater sites come online.

RPO duration: extended from 2.0 to 2.5 years; further extension would signal customers locking in longer, while a reversal would indicate hedging on AI commitments.

Memory pricing impact on CapEx and Cloud GM: Hood flagged rising memory prices as a CapEx headwind that will build gradually into Microsoft Cloud gross margins over the 6-year depreciation life.

Sources

  1. Microsoft Q2 FY2026 press release (SEC filing): https://www.sec.gov/Archives/edgar/data/789019/000119312526027198/msft-ex99_1.htm
  2. Microsoft Q2 FY2026 earnings call prepared remarks and Q&A (Weiss/Morgan Stanley and Moerdler/Bernstein exchanges referenced inline)

Get the next brief, free.

We publish analyst-grade earnings briefs the same day or morning after every call — headline numbers, segment KPIs, Q&A highlights, and tone analysis. Free during beta.

This is not investment advice.