< All Tickers
MSFT
Microsoft Corporation
Earnings
MSFT | Earnings Review
Microsoft Corporation | FY2026 Q3 reported April 29, 2026 AMC | Analysis date: April 29, 2026 | Daloopa company_id 135
Revenue Beat
+1.8%
$82.9B vs $81.4B Street; +18.3% YoY; broadest segment strength in 4Q
EPS Beat (clean)
+5.4%
$4.27 vs $4.05 Street; only $14M OpenAI MTM drag — operationally comparable, unlike FQ2
Azure CC Growth
+39%
Reaccel from FQ2 +38%; reported +40%; capacity-constrained through CY2026
FY26 CapEx Frame
~$190B incl. ~$25B comp. price
Hood explicit on call; underlying volume capex ~$165B. Cash capex $30.9B Q3 (below $34.3B fear). FY27: explicit "double-digit revenue and operating income growth"
Clean double beat; AI margin defense was the single biggest takeaway from the call. Microsoft delivered $82.9B in revenue (+18.3% YoY, +1.8% vs $81.4B Street) and $4.27 EPS vs $4.05 Street, with Azure & other cloud services accelerating to +39% constant currency (vs +38% in FQ2 and ~38% Street). The print is operationally clean — only $14M OpenAI mark-to-market drag — so the EPS beat is real. Cash CapEx stepped down to $30.9B (from $37.5B FQ2'26 and below the feared $34.3B Street), allowing FCF to rebound to $15.8B from FQ2's $5.9B. Hood quantified FY26 capex as ~$190B including ~$25B from higher component pricing (so volume capex closer to ~$165B), and explicitly guided FY27 to "another year of double-digit revenue and operating income growth." New disclosures from the call: AI Business ARR $37B+ (+123% YoY); LinkedIn Talent agentic products at $450M ARR run-rate; Accenture deployed 740,000 M365 Copilot seats (largest Copilot win); Bayer / J&J / Mercedes / Roche each at 90,000+ seats; Maia 200 accelerator live in Iowa & Arizona with 30%+ tokens/$ uplift; Cobalt CPU running Databricks/Siemens/Snowflake workloads in nearly half of DC regions. OpenAI economics: revenue share extended through 2030; royalty-free IP "all the way to [GPT-]32" with elimination of MSFT's rev share to OpenAI on certain workloads — directional margin tailwind not in the press release. Hood's direct rebuttal of the AI-margin-dilution thesis: "Margins were actually better and they remained better in our AI business versus where we saw in the cloud transition." Capacity guide: "remain constrained at least through 2026"; Azure "modest acceleration in second half of calendar year." Unresolved tension: commercial cloud GM declined to 66% — sixth consecutive quarter of compression — but management framed mix to short-lived assets (~2/3) as the offset. Watch items: (1) FQ4'26 print and the H2 capex ramp toward $190B; (2) commercial cloud GM bottoming; (3) GOOGL printed Cloud +63% same day — narrative pressure on relative pace; (4) Microsoft Build (May 19–22); (5) MPC -1% YoY drag persists. Net read: demand-side proof reinforced; the AI margin debate may be tilting more positive than feared, with Maia/Cobalt + Copilot engagement ("weekly engagement at the same level as Outlook") as the proof points.
Key Metrics Trends
| Metric | FQ3 2024 | FQ4 2024 | FQ1 2025 | FQ2 2025 | FQ3 2025 | FQ4 2025 | FQ1 2026 | FQ2 2026 | FQ3 2026 |
|---|---|---|---|---|---|---|---|---|---|
| Productivity & BP rev | $19.6B | $20.3B | $28.3B | $29.4B | $29.9B | $33.1B | $33.0B | $34.1B | $35.0B |
| Productivity & BP rev YoY % | - | - | - | - | +53.0% | +63.0% | +16.6% | +15.9% | +16.9% |
| Intelligent Cloud rev | $26.7B | $28.5B | $24.1B | $25.5B | $26.8B | $29.9B | $30.9B | $32.9B | $34.7B |
| Intelligent Cloud rev YoY % | - | - | - | - | +0.2% | +4.8% | +28.2% | +28.8% | +29.6% |
| More Personal Computing | $15.6B | $15.9B | $13.2B | $14.7B | $13.4B | $13.5B | $13.8B | $14.2B | $13.2B |
| More Personal Computing YoY % | - | - | - | - | -14.2% | -15.4% | +4.4% | -2.7% | -1.3% |
| Microsoft Cloud rev ($B) | 35.1x | 36.8x | 38.9x | 40.9x | 42.4x | 46.7x | 49.1x | 51.5x | 54.5x |
| Microsoft Cloud rev ($B) YoY % | - | - | - | - | +20.8% | +26.9% | +26.2% | +25.9% | +28.5% |
| Azure CC growth % | 31.0% | 30.0% | 34.0% | 31.0% | 35.0% | 39.0% | 39.0% | 38.0% | 39.0% |
| Commercial cloud GM % | 72.0% | 69.0% | 71.0% | 70.0% | 69.0% | 68.0% | 68.0% | 67.0% | 66.0% |
| Total revenue | $61.9B | $64.7B | $65.6B | $69.6B | $70.1B | $76.4B | $77.7B | $81.3B | $82.9B |
| Total revenue YoY % | - | - | - | - | +13.3% | +18.1% | +18.4% | +16.7% | +18.3% |
| Operating income | $27.6B | $27.9B | $30.6B | $31.7B | $32.0B | $34.3B | $38.0B | $38.3B | $38.4B |
| Operating income YoY % | - | - | - | - | +16.0% | +22.9% | +24.3% | +20.9% | +20.0% |
| Diluted EPS | $2.94 | $2.95 | $3.30 | $3.23 | $3.46 | $3.65 | $3.72 | $5.16 | $4.27 |
| Diluted EPS YoY % | - | - | - | - | +17.7% | +23.7% | +12.7% | +59.8% | +23.4% |
| Cash CapEx | $14.0B | $19.0B | $20.0B | $22.6B | $21.4B | $24.2B | $34.9B | $37.5B | $30.9B |
| Cash CapEx YoY % | - | - | - | - | +52.9% | +27.4% | +74.5% | +65.9% | +44.3% |
| Free cash flow | $21.0B | $23.3B | $19.3B | $6.5B | $20.3B | $25.6B | $25.7B | $5.9B | $15.8B |
| Free cash flow YoY % | - | - | - | - | -3.2% | +9.6% | +33.3% | -9.3% | -22.1% |
Trajectory: Three diverging curves. (1) Microsoft Cloud + Azure: re-accelerating; Azure CC stepped from 31% (FQ2'25 trough) to 39% (FQ3'26), Microsoft Cloud from +21% to +29% — durable acceleration with capacity-constrained tone. (2) Total revenue: holding +17–18% YoY range with $627B RPO supporting forward years. (3) Commercial cloud GM: steady erosion of -100 to -300 bps QoQ for six consecutive quarters (72% → 66%) — AI infrastructure cost pressure structural, not transitory. Cash capex inflection in FQ3'26 ($37.5B → $30.9B) is the most important new datapoint: if FQ4 confirms, the FCF + ROIC argument materially improves even with GM in the mid-60s.
Beat/Miss
Guidance
Catalysts
Street Q&A
Contradictions
Read-Throughs
This Quarter vs Consensus
| Metric | Consensus / Expected | Actual | Variance | Read |
|---|---|---|---|---|
| Revenue | $81.4B | $82.886B | +$1.5B / +1.8% | Beat |
| Diluted EPS (clean) | $4.05 | $4.27 | +$0.22 / +5.4% | Beat — clean (only $14M OpenAI MTM drag) |
| Azure CC growth | ~38% | +39% | +100 bps | Beat — re-accel from FQ2 38% |
| Microsoft Cloud rev | ~$53B | $54.5B | +$1.5B | Beat |
| Cash CapEx | ~$34.3B (feared) | $30.9B | -$3.4B | Positive surprise — FCF recovers to $15.8B |
| Commercial cloud GM | ~67% | 66% | -100 bps | Miss — AI cost pressure persists |
| Operating margin | ~45% | ~46.3% | +130 bps | Beat |
| AI Business ARR (new disclosure) | n/a | $37B+ (+123% YoY) | — | Major positive datapoint |
| Commercial RPO | ~$625B | $627B | ~flat | Reassuring |
| L8Q rev beat rate | — | 8/8 = 100% | — | Consistent Beater |
| L8Q EPS beat rate | — | 8/8 = 100% | — | Consistent Beater |
Pattern: Consistent Beater — Azure beat magnitude returning after two in-line quarters; commercial cloud GM consistent disappointer. MSFT has beaten revenue and EPS in 8 of 8 trailing quarters. The shift in this print: Azure CC rejoined the +1pp beat camp after FQ1 and FQ2 were essentially in-line, and cash CapEx printed *meaningfully below* the feared $34B level for the first time in the AI capex cycle. Variance drivers, ranked: (1) Azure capacity unlock pacing better than feared, (2) AI Business ARR scaling faster than the Street modeled (~$37B/+123%), (3) Cash capex moderation (-$6.6B QoQ) restoring FCF, (4) MPC stabilization vs depressed FQ3 expectations, (5) clean OpenAI accounting vs FQ2's positive swing. Caveat: commercial cloud GM compression is unresolved — bull case requires stabilization in mid-60s; bear case requires it reaching low-60s before AI monetization catches up.
Guidance Deep Dive
| Metric | Prior View / Guide | FQ3'26 Actual / New Frame | Direction | Read |
|---|---|---|---|---|
| Azure CC growth (FQ3 guide) | high-30s; consistent w/ FQ2 (~38%) | +39% actual | Beat / re-accel | Positive |
| Azure FQ4'26 setup | n/a | Capacity-constrained through CY2026 | Sets up Azure +37–40% | Constructive |
| Cash CapEx | Trending toward $34B Street | $30.9B; pace moderating QoQ | Lower than feared | Positive — possible peak |
| Commercial cloud GM | AI mix pressure flagged | 66% (vs 67% FQ2; 70% PY) | Continued compression | Negative — 6th consecutive Q |
| Commercial RPO | $625B | $627B | ~Flat QoQ but historically high | Neutral / supportive |
| AI Business ARR | ~$30B reference | $37B+, +123% YoY | New disclosure scaling | Major positive |
| OpenAI MTM impact | Unpredictable, large in FQ2 | Only $14M drag | Clean quarter | Neutral / clean |
| MPC trajectory | Negative drag flagged | -1% YoY | Persistent drag | Watch |
| FY2026 implied revenue | ~$320B (consensus tracking) | ~$324–328B implied | Track toward upper end | Positive |
Tone: confident on demand, cautious on margins, moderating on capex. Microsoft Cloud and Azure tone strengthened from "strong, supply-constrained" to "capacity-constrained through CY2026" — implies management has visibility into which quarters get capacity unlocks. CapEx tone moderated meaningfully — the FQ3'26 step-down from $37.5B → $30.9B is framed as "absolute spending pace moderating QoQ" rather than a guide cut, but it's the first real signal that the AI capex peak may be behind. Margin tone unchanged — commercial cloud GM 66% is acknowledged as AI mix pressure, no path to stabilization disclosed. MPC tone defensive — Windows + Devices + Gaming + Search continue to dilute cloud strength. Watch: (1) FQ4'26 print is the durability test for capex moderation; (2) any tone shift on commercial cloud GM bottoming would be material; (3) OpenAI relationship economics post-2030 remain undisclosed.
Upcoming Catalysts
| # | Catalyst | Timing | Watch | Read |
|---|---|---|---|---|
| 1 | Microsoft Build 2026 | May 19–22, 2026 | Copilot, Azure AI Foundry, agent workflows | Annual developer narrative reset; usually moves price |
| 2 | AMZN AWS print | May 1, 2026 | AWS YoY growth + capex tone | Cross-validates AI cloud demand vs MSFT +29% / GOOGL +63% |
| 3 | FY26 Q4 print | Late July 2026 | Azure CC sustainability; cash CapEx cadence; commercial cloud GM | Confirms whether FQ3'26 capex moderation is durable |
| 4 | Commercial RPO conversion | Next 12–24 months | $627B → revenue cadence | Largest visibility lever |
| 5 | AI Business ARR trajectory | Ongoing | $37B+, +123% YoY; step-size deceleration | Direct AI monetization datapoint |
| 6 | Commercial cloud GM stabilization | FY2027 | Whether GM bottoms in mid-60s | Critical for AI ROIC narrative |
| 7 | OpenAI relationship economics | 2026+ | Access, exclusivity, accounting cleanup | Opaque but material |
| 8 | Activision-Blizzard / Gaming inflection | Holiday 2026 | MPC -1% YoY → flip | Diversification lever |
| 9 | Capacity unlock cadence | 2026 | Whether "supply-constrained" language eases | Positive only if growth maintained |
| 10 | FY27 capex framing | FY26 Q4 / annual report | First explicit FY27 capex view | Determines FCF trajectory |
Street Q&A
| # | Analyst (Firm) | Topic | Mgmt Response | Quality |
|---|---|---|---|---|
| 1 | Keith Weiss (Morgan Stanley) | Demand translation into bookings; "who pays for all this" given flat IT budgets | Hood: bookings will look different — license + meter, "may not all flow through bookings in the same way." Nadella: per-user becomes "per user and usage" — dollars come from business outcomes (cost down / revenue up). Note: Keith's last earnings call. | Well Answered |
| 2 | Karl Keirstead (UBS) | Confidence in ~$120B H2 capex ramp; supply chain; allocation framework | Hood: "feel quite good" about industrial logic; component-price impact tilts mix to short-lived assets; allocation balanced across Azure / Copilot / coding / security; expects pressure between 1P and Azure to persist | Well Answered |
| 3 | Brent Thill (Jefferies) | Why are AI margins better than feared? | Hood: "Margins were actually better and they remained better in our AI business versus where we saw in the cloud transition" — consumption pricing captures value; OpenAI IP "royalty-free for a long time"; first-party silicon + sw efficiency taking margin out of the infra stack | Well Answered — direct rebuttal to AI-dilution thesis |
| 4 | Mark Moerdler (Bernstein) | Capex/revenue disconnect; confidence margins hold | Hood: short-lived assets (~2/3 of mix) correlate with revenue; M365 Commercial Cloud accelerated and guided up again; $600B+ RPO to deliver. Nadella: TAM conviction in knowledge work / coding / security; "model capabilities are exponential…Agent Mode in Excel sort of kind of didn't work until it started working." | Well Answered |
| 5 | Gabriela Borges (Goldman Sachs) | Copilot learnings; how informing E7 / Cowork strategy | Nadella: form-factor evolution chat → agents (Researcher/Analyst) → Agent Mode in Word/Excel/PPT → Cowork delegation; "Copilot weekly engagement now at the same level as Outlook"; multi-model harness decoupled from models; context = "the most important database in any company." | Well Answered |
| 6 | Kirk Materne (Evercore ISI) | OpenAI agreement update — modeling implications | Nadella: IP "royalty-free…all the way to [GPT-]32"; OpenAI a large compute customer; equity stake; "win-win construct." Hood: revenue share extended through 2030; royalty-free with "elimination of our rev share to them." | Mostly Answered (no $ disclosed) |
| 7 | Rishi Jaluria (RBC) | E7 seat-vs-consumption philosophy; 3-5yr mix | Nadella: seat = "entitlements to some consumption right"; base bundle + overages into pure consumption with commitment discounting; IT budgets reshape via business-outcome reallocation from other P&L line items | Mostly Answered (no quant mix) |
Contradictions
| # | Topic | Severity | Statement A | Statement B | Why it's a tension |
|---|---|---|---|---|---|
| 1 | "Demand is supply-constrained" vs Cash CapEx stepping DOWN QoQ | Moderate | Mgmt: AI capacity remains supply-constrained through CY2026 | Cash capex moved from $37.5B (FQ2'26) to $30.9B (FQ3'26) — clear step-down | Compatible if step-down is timing-driven (lumpy server deliveries) rather than discretionary slowdown. If FQ4'26 also moderates, the "supply constraint" narrative will fray. Watch FQ4 print. |
| 2 | Re-accelerating revenue / Azure vs eroding commercial cloud GM | Moderate — central tension | Revenue, Microsoft Cloud, Azure all re-accelerating; AI ARR +123% | Commercial cloud GM 66% — down for 6 consecutive quarters from 72% | Not a contradiction in MSFT's narrative; it's the central tension in the bull thesis. Bulls want both growth and margin durability — currently get only one. Resolution requires either AI mix saturation or unit-economics improvement at scale. |
| 3 | Clean OpenAI Q vs durable EPS dependency | Minor | FQ3'26: only $14M OpenAI MTM drag — clean quarter | FQ2'26: large positive OpenAI swing in EPS | $14M drag is helpful for clean optics, but FQ2 swing was material. Investors should not let one clean quarter mute their concern that quarterly EPS remains OpenAI-MTM-dependent. |
Indirect Read-Throughs
| Name | Relationship | What MSFT signaled | Read-through |
|---|---|---|---|
| NVIDIA (NVDA) | Supplier | Azure +39% CC supports continued GPU pull | POSITIVE — AI compute demand durable at top hyperscaler |
| Broadcom (AVGO) / AMD | Silicon | Custom + merchant accelerator mix | Neutral-positive AVGO; positive AMD MI series |
| Google (GOOGL) | Peer / Cloud competitor | GCP +63% just printed (Apr 29) — faster than MSFT Cloud +29% | YELLOW FLAG — narrative risk if GCP keeps differential growth |
| Amazon (AMZN) | Peer / Cloud competitor | AWS prints May 1; benchmarked against GOOGL +63% / Azure +39% | Sets relative pace; AWS sub-25% widens GCP/MSFT gap |
| OpenAI (private) | Strategic partner | $14M MTM drag — clean quarter | Less noise this Q; relationship economics post-2030 remain debate |
| Salesforce (CRM) | Peer / partner | Enterprise AI workflow demand strong (P&BP +17%) | Positive |
| ServiceNow (NOW) | Peer / partner | Same enterprise SW tailwind | Positive |
| Snowflake (SNOW), MongoDB (MDB), Datadog (DDOG) | Cloud-native vendors | Microsoft Cloud +29% supports broader cloud consumption | POSITIVE |
| Power IPPs (CEG, VST, TLN) | Energy supply | Capex cresting still implies durable power demand | Positive multi-year |
| Networking / cooling (ANET, VRT, ETN) | Datacenter | Capex flow-through | Positive |
| Accenture (ACN) | Copilot deployment partner | 740,000 M365 Copilot seats — largest Copilot win to date | POSITIVE — anchor reference |
| Bayer, Johnson & Johnson, Mercedes-Benz, Roche | Enterprise customers | Each committed to 90,000+ M365 Copilot seats; Bayer building agent platform on Foundry (20,000+ MAU) | Positive — Tier-1 references |
| HSBC | Customer (D365) | Using D365 prebuilt agents — 30%+ reduction in customer-service issue resolution time | Positive |
| Databricks, Siemens, Snowflake | Cloud-native customers on Cobalt | Running production workloads on Microsoft Cobalt CPU at scale | Positive validation of MSFT first-party silicon |
| Shutterstock, WPP | MAI customers (external) | First commercial customers consuming MAI first-party models (MAI-Image-2 / MAI-Transcribe-1) via Foundry | POSITIVE — MSFT monetizing own models externally |
| Activision-Blizzard / Gaming peers | Owned franchise | MPC -1% YoY drag | Mixed |
| Apple (AAPL) | MAU / Services peer | No direct mention | Neutral |
| Meta (META) | AI infra peer | AI capex narrative cross-validation | Neutral — AI infra demand confirmation |
Data sourced from Daloopa. Document search is currently in beta; transcript and filing snippets may vary.