Static snapshot, generated 13 November 2026. Not updated, no live prices.
Microsoft (MSFT) research report
Microsoft (MSFT, $523.95 delayed, 2026-11-13) pairs a contractual-looking AI backlog (commercial RPO +84% to $678B, Microsoft Cloud +27% to $214.4B) [E0145] with a thinning cash bridge: capex is 34.9% of revenue and FCF of $67.0B is half of $133.7B net income [E0570][E0580][E0591]. The price implies about 28.8% annual FCF growth for five years versus a 16.8% DCF stage-one rate [E1545][E1539], and the DCF ($279.97) and comps ($595.23) disagree by 113% [E0490]. The OpenAI rights, chip supply and power are the links that must hold into the estimated 2027-01-27 print [E1692][E1802][EARN-1].
Investment thesis
Microsoft sells productivity software, cloud and AI infrastructure, and personal computing products to enterprises and consumers. FY2026 revenue was $331.8B, up from $281.7B, with Microsoft Cloud at $214.4B and commercial RPO up 84% to $678B [E0202][E0203][E0145]. Operating margin is 46.8% and net income $133.7B [E0594][E0591].
What is happening now: the company is spending heavily to turn that backlog into capacity. Capex is 34.9% of revenue and FCF of $67.0B is about half of net income [E0570][E0580]. A multi-day Outlook, Teams and Copilot outage is hitting headlines, and a Canadian class action is reported but unverified [E1402][E0638].
What the price assumes: at $523.95 the market implies about 28.8% annual FCF growth for five years, against 16.1% three-year revenue CAGR and a 16.8% DCF stage-one rate [E1545][E0607][E1539]. DCF is $279.97 and comps $595.23, a 113% gap, so the blended $479.35 is low-confidence [E0487][E0486][E0490][E0489].
Our read: demand looks well supported, the balance sheet shows no funding wall (net debt $19.4B, 51x interest cover) [E0589][E0587], and the valuation question is whether capex converts to cash on schedule. Our inference is that the stock is priced for execution, not for disappointment, and that the weakest evidence link is how much of the RPO is tied to OpenAI, which is not disclosed [E0021].
Biggest risk to this read: a slip in the four must-hold links: OpenAI access [E1692], chip supply, power and capital [E1802][E1780], and tariffs or regulatory matters [E1824][E1758]. Operating income is also sensitive to rates and USD [E1325][E1333]. Insiders are net sellers (26 sells, 0 buys, mostly pre-planned), which is weak evidence either way [E0251][E0250][GOV-6].
Key debates and what is priced in
Debate 1: Is the $678B backlog a hard demand floor? Market take: yes, RPO +84% means capex is simply a supply problem [E0145]. Our read: probably largely yes, but the OpenAI share and customer concentration are not disclosed [E0021], so we cannot size how contractual it is. What settles it: RPO commentary and capacity additions on the 2027-01-27 call (estimated) [EARN-1].
Debate 2: Does capex convert to cash? Market take: spend is temporary and returns follow. Our read: FCF of $67.0B is below its 3-year average of $70.9B while capex is 34.9% of revenue [E0580][E0581][E0570]; the price needs 28.8% FCF growth [E1545]. What settles it: FCF and capex guidance in the next one or two quarters.
Debate 3: Is the outage a reliability scare or a one-off? Market take: sentiment hit but contained. Our read: the event is live, tone is -0.33 and the class action is unverified [E1402][E1460][E0638]; the Copilot trust angle is our inference. What settles it: outage duration, SLA credits disclosed and any confirmed filings.
Debate 4: Does the macro reach the earnings line? Market take: not priced. Our read: operating income sensitivity to the 10Y and USD is statistically reliable but price sensitivity is not [E1325][E1333][E1368]. What settles it: the Fed path (hike odds split 95.6% Dec vs 59.5% Jan) [E1022][E1027].
Reverse DCF: the market is pricing about 28.8% annual FCF growth for five years [E1545][E1540], against a DCF stage-one assumption of 16.8% [E1539], a 16.1% three-year revenue CAGR [E0607] and 17.8% latest revenue growth [E0608]. The implied rate is well above anything in the record. Consensus: 54 analysts have a $636.12 mean (range $440 to $870), +21% above price [E0568][E0523][E0522][E0520][E0626]; forward P/E is 21.7x vs 27.3x TTM [E0597][E0598]. Our base-case FCF growth is not above the history shown, so our inference is the gap sits in whether capex-heavy growth turns into cash.
Valuation
Model output, not a target. Blended fair value is $479.35 (-8.5% vs $523.95), weighting DCF 40%, comps 35%, analysts 25% [E0489][E0539][E0496][E0502][E0508]. DCF base is $279.97 (bear $185.05, bull $414.29) at a 9.76% WACC and 2.5% terminal growth [E0495][E0494][E0491][E1586][E1584]. Comps give $595.23 ($495.07 to $702.80) [E0501][E0500][E0497]. The 113% DCF/comps gap is flagged [E0490]. Multiples: P/E 27.3x TTM and 21.7x forward, EV/EBITDA 19.2x, PEG 1.54, P/B 8.3x [E0598][E0597][E0578][E0599][E0596]. DCF SIMULATED Monte Carlo puts 1.8% probability on fair value above price [E1228]. The DCF sensitivity grid runs from $197 to $479 across WACC and terminal growth [E1567][E1551]. No sum-of-parts or historical-analog return figures are in the pack; the analogs engine ran as context only [E1247].
Scenario Lab output is a model range, not a target. The SIMULATED Monte Carlo 12M P10 to P90 is $293.6 to $784.2 with a $495.4 median [E0516][E0517][E0514][E0518]; probability that DCF fair value exceeds price is 1.8% [E1228]. Our likelihood buckets: bull about 30%, base about 45%, bear about 25%.
| Case | What must be true | Early signs | Model range | Likelihood |
|---|---|---|---|---|
| Bull | Capacity converts the backlog, OpenAI access intact | RPO conversion, capex guidance holds | Upper end of MC range (P90 $784) | Medium (~30%) |
| Base | Demand strong, FCF stays below net income | FCF flat vs $70.9B average | Near MC median $495 | Highest (~45%) |
| Bear | Power/chip or OpenAI slip with higher rates | Capacity delays, hike path firming | Lower end (P10 $294) | Lower-medium (~25%) |
Catalysts and timeline
Next earnings release is 2027-01-27, estimated, not confirmed by the company [EARN-1][E0205]. Ex-dividend date is 2026-11-19 [E0210]. The annual meeting is 2026-12-08 [E0238]. NVIDIA reports 2026-11-17 per graph (tier 3) [G-40].
| Date | Event | Why it matters | What to watch |
|---|---|---|---|
| 2026-11-17 | NVIDIA earnings (graph, tier 3) | GPU supply read-through | Hyperscaler allocation |
| 2026-11-19 | MSFT ex-dividend | Income signal | None material |
| 2026-12-08 | MSFT annual meeting (DEF 14A) | Governance and pay votes | Proxy items |
| 2027-01-27 | MSFT earnings (estimated, unconfirmed) | Capex, RPO and AI capacity commentary | Capex guidance, FCF, Q&A tone |
Risks
By folder; details in table.
Company overview and business model
Microsoft develops AI, cloud, productivity and personal-computing technology for individuals and businesses [E0069][E0071]. Revenue by segment in FY2026: Productivity and Business Processes $140.0B (42.2%), Intelligent Cloud $137.8B (41.5%), More Personal Computing $54.1B (16.3%) [E0170][E0167][E0177][E0174][E0184][E0181]. Intelligent Cloud grew from $106.3B and Productivity from $120.8B [E0178][E0171]. By geography the US is $170.8B (51.5%) and non-US $161.0B (48.5%) [E0033][E0030][E0040][E0037].
The durable earner is the cloud and subscription stack: Microsoft Cloud revenue rose 27% to $214.4B, Microsoft 365 Commercial cloud +17% and Consumer cloud +28% [E0145]. LinkedIn grew 11% [E0145]; Search advertising was $15.2B [E0131]. Headcount is about 223,000 (121,000 in the US) [E0044][E0045]. Customer concentration is not disclosed after checking the 10-K and 10-Q layers [E0021]; key suppliers are not named in Item 1 [E0022].
Industry and competitive position
Microsoft names broad competition in Office (software and application vendors, AI-first application companies), Windows (alternative platforms and devices) and endpoint security [E0016][E0018][E0019]. It positions on integrated, secure tools that work with customers' existing technologies [E0017]. The filing-derived peer screen returns small names that cite Microsoft as a competitor (AvePoint, C3.ai, Skkynet, Turtle Beach, PC Connection) [E0774][E0782][E0790][E0798][E0806]; these are not size-comparable. On the AI infrastructure side the graph links NVIDIA and AMD as suppliers [G-8][G-27] and Amazon as the hyperscale comparison (tier 5, unconfirmed price read-through) [WEB-4]. Our inference: the moat is breadth of distribution plus the OpenAI access [E1692], on a sliding scale rather than a grade. Market size and share-trend data are not in the pack.
Management, governance, government and insiders
Satya Nadella and CFO Amy Hood are graph-tracked insiders [G-14][G-16]; other tracked insiders include Brian DeFoe and Alice Jolla [G-44][G-45]. Hood's 8 reported sales from 2026-09-14 to 2026-11-09 were about $15M and pre-planned under 10b5-1 [G-57][GOV-24][GOV-25]; the CMO also sold 5,000 shares on 2026-11-10 at $531 [E0254][E0257][E0256]. Insider ownership is 0.08% and institutions hold 76.5% [E0585][E0586]. A reported voluntary buyout of about 60,000 staff is tier 3 graph-only and unverified [G-7]. The annual meeting is 2026-12-08 [E0238]. Pay versus performance, board independence and related-party deals are not in the pack.
Form 4: 26 sells, 0 buys, about $92.5M sold in 180 days [E0250][E0251][E0253]; CFO Amy Hood's sales across Sep to Nov 2026 are flagged pre-planned (10b5-1) [GOV-24][GOV-25][GOV-6]. Congress: 7 trades by 5 members in 3 years, 3 purchases and 3 sales, none large in a cluster [GOV-1][GOV-10][GOV-11]. Lobbying: 225 filings worth $38.8M in 3 years, with $2.4M in Q3 from Microsoft itself across tax, trade, defense and tech issues [GOV-2][GOV-18]. Federal contracts: 37 awards worth $816.3M, largest source Department of Defense [GOV-3][GOV-23]. FDA, clinical trials and federal rules were checked and found nothing [GOV-4][GOV-5][GOV-7]. Physical-world signals: no sites registered [GOV-8]. 13D/G and 13F changes were not in the pack.
Financial analysis and model summary
Growth: revenue was $331.8B in FY2026 vs $281.7B and $245.1B in the two prior years, up 17.8% YoY with a 16.1% three-year CAGR [E0202][E0203][E0204][E0608][E0607]. EPS growth is 31.6% and net income growth 31.3% [E0577][E0592]. Margins: gross 67.9%, operating 46.8%, net 40.3% [E0584][E0594][E0593].
Cash conversion: operating cash flow was $182.9B, but capex intensity is 34.9% and FCF is $67.0B (20.2% margin) versus net income $133.7B; FCF is below its 3-year average of $70.9B [E0564][E0570][E0580][E0582][E0591][E0581]. Returns: ROIC 27.1% against a cost of capital (WACC) of 9.8% [E0614][E1586]; ROE 34.0% [E0612].
Balance sheet: net debt $19.4B, debt/equity 0.09, current ratio 1.23, interest cover 51x [E0589][E0572][E0571][E0587]. No debt-due ladder is in the pack. Capital allocation: payout ratio 19%, dividend yield 0.75%, share count down 0.16% YoY [E0595][E0573][E0615]. Piotroski is 6/9 (liquidity and gross-margin tests failed) and Altman Z is 9.17 (safe) [E1125][E1136][E1140][E1122][E1123].
Story versus numbers: earnings quality is high, but reported profit overstates near-term free cash generation because capex is front-loaded for AI capacity. Our inference is that FCF should be tracked ahead of EPS.
Key reported lines only; a full standardised five-year statement set is not in the pack.
| Metric | Value |
|---|---|
| Revenue FY26 | $331.8B |
| Net income | $133.7B |
| Operating margin | 46.8% |
| Operating cash flow | $182.9B |
| Free cash flow | $67.0B |
| Capex % revenue | 34.9% |
| Net debt | $19.4B |
| ROIC | 27.1% |
| Line | FY2026 | FY2025 | FY2024 |
|---|---|---|---|
| Revenue | $331.8B | $281.7B | $245.1B |
| Net income | $133.7B | n/a | n/a |
| Operating cash flow | $182.9B | n/a | n/a |
| Free cash flow | $67.0B | n/a | n/a |
| Total assets | $758.4B | n/a | n/a |
ESG and red-flag checklist
Thresholds are stated per item; items not checkable are unknown.
Count: 0 red, 4 amber, rest green or unknown (see checklist). Top 3 ambers: FCF at half of net income [E0580][E0591]; DCF/comps gap of 113% [E0490]; live outage with an unverified class action [E1402][E0638]. See the Red-flag checklist and Gaps.
| Item | Status | Threshold | Source |
|---|---|---|---|
| Cash conversion (FCF / net income) | Amber | FCF below 0.7x net income | FCF $67.0B vs NI $133.7B [E0580][E0591] |
| Valuation model agreement | Amber | DCF/comps gap above 50% | 113% gap [E0490] |
| Live outage / unverified suit | Amber | Any reported suit or outage | [E1402][E0638] |
| Customer concentration | Amber (unknown) | Disclosure of any customer above 10% | Not disclosed [E0021] |
| Insider selling | Green/amber | Net selling vs pre-plan | $92.5M, mostly 10b5-1 [E0253][GOV-6] |
| Leverage / bankruptcy risk | Green | Altman Z above 3 | Z 9.17 [E1122] |
| Liquidity trend | Amber | Piotroski liquidity test | Failed [E1136] |
| Tax / regulatory | Amber | Open audit or probe | IRS audit [E1890]; EU DMA/DSA [E1846] |
| Auditor / accounting issues | Unknown | Any restatement or auditor flag | Not checked in pack |
What we could not find
Unknown stays unknown. These gaps were listed in the report itself.
- Customer concentration and OpenAI share of the $678B RPO are not disclosed (business checks found none) [E0021]
- Key suppliers are not named in the 10-K Item 1 [E0022]
- OpenAI contract terms and economics are not in the pack; affects the OpenAI chain
- Short interest history was not exposed by the FINRA endpoint (only 0.82% of float from Yahoo) [E1377][E0616]
- GDELT tone data failed (partial failure) [E0378]
- CourtListener search was blocked, so all three legal matters are news-only and unverified [E0660]
- Social sentiment sources were skipped [E1522]
- No market size, share-trend or hyperscale peer financials in the pack; affects the Industry section
- Full standardised statements, debt-due ladder and proxy pay data are not in the pack; affects Financials, Management and Appendix
- Outage root cause, duration and customer SLA exposure are unknown; no Q2 capex guidance or power-contract detail
- NVIDIA CEO transition and the reported voluntary buyout of about 60,000 staff are graph tier 3 and not verified against primary sources [G-42][G-7]
- Web map truncated to 40 of 223 edges and prices are from an unofficial source; MiroFish adapter and the forecast model were disabled [E0399][E1587][E1263][E0541]
- Industry and competitive position: No market size, growth or share data and no hyperscale peer (AWS/Google Cloud) metrics are in the evidence pack.
- Quant and ML signals: ML pattern-match outcomes and live hit rate are not available: analogs are context only and the forecast model is disabled.
- Management and governance: Proxy detail on pay against performance, board independence and related-party deals was not available in the evidence pack.
- Appendix: financial statements: Full five-year-plus-TTM income statement, balance sheet and cash flow line items are not in the evidence pack, only the summary lines shown.
- Industry and competitive position: Part of the underlying data is missing: No market size, growth or share data and no hyperscale peer (AWS/Google Cloud) metrics are in the evidence pack.
- Management, governance, government and insiders: Part of the underlying data is missing: Proxy detail on pay against performance, board independence and related-party deals was not available in the evidence pack.
- Financial analysis and model summary: Part of the underlying data is missing: Full five-year-plus-TTM income statement, balance sheet and cash flow line items are not in the evidence pack, only the summary lines shown.
Disclosures
- Pegasus Alpha is a research tool. This report is not investment advice and is not an offer or a solicitation to buy or sell any security.
- Pegasus Alpha gives no rating and no price target. Where a range is shown it is a model range, not a target.
- Every claim is cited to the evidence pack or marked as inference. Evidence tiers 1 to 5 apply; anything below tier 3 is unconfirmed.
- Anything marked SIMULATED comes from a simulation. Simulations are never evidence.
- Figures come from the public sources listed in the Sources section, as they were available on the date of the run. Gaps are listed in the Gaps section.