TL;DR
MSFT (Microsoft Corporation) — Adversarial Due-Diligence Report
Run of 2026-08-14 (4th cycle; re-anchored after the 2026-08-14 price/history/policy refresh) · data as-of per the embedded manifest · price anchor $496.88 — a settled 2026-08-13 close (market.json currentPrice = the last history.json row 2026-08-13 to the cent; the refresh ran post-close, 03:09 UTC = 23:09 ET on 2026-08-13), with mcap $3,689,607M. The central 2026-07-10 trap is gone: Microsoft reported Q4/FY2026 on 2026-07-29 (10-K msft-20260630.htm + Q4 8-K), a prior refresh pulled the full FY2026 10-K durations into companyfacts, and metrics.json now keys on FY2026 (ended 2026-06-30) and is current (basis = "latest fiscal year ending 2026-06-30," stale_inputs = []) — no FY2025-staleness caveat, no TTM-from-9M roll-forward. All current-company figures below are the reported FY2026 full year. EV is the reconciled ~net-cash basis (mcap ± the 2026-06-30 net position ≈ $3.65–3.71T), not the provider EV $2,879,039M, which this cycle imputes ~$0.81T of phantom net cash and understates the multiple. FY2025 figures (revenue $281,724M, FCF $71,611M, the old ev_to_revenue 13.41) appear only as prior-year comparisons. The 4th cycle re-anchors the price/policy context one trading day forward, from the retired 2026-08-12 close $492.43 to this settled 2026-08-13 close $496.88 (filings.json/companyfacts.json byte-unchanged, so the FY2026 spine and the coverage ledger stand).
TL;DR
At $496.88 (settled 2026-08-13) Microsoft has recovered most of the −34.9% peak-to-trough de-rating — back +3.2% above the pre-Q2-crash $481.63 (2026-01-28), +40.8% off its $352.83 low (2026-06-25), but still −8.3% off its $542.07 52-week high (2025-10-28) and −4.6% YoY (vs $520.58 on 2025-08-13); a full round-trip to the high needs +53.6% off the low, not the +40.8% achieved, so this is a recovery of ~76% of the drawdown, not the whole move ("52-week," not "all-time" — history.json is a 262-close, 1-year window). It trades at 27.6× FY2026 earnings, ~11.0–11.2× reconciled EV/FY2026-revenue (the cohort's highest on EV/sales), and 55.1× FY2026 free cash flow (a 1.82% FCF yield). This is no longer the "cheapest de-rated hyperscaler" of the first cycle; the whole debate reduces to one question — does the record AI capex convert to free cash flow? — and both cases name the same event as their falsifier #1: the confirmed Q1 FY2027 print, Tue 2026-10-27 (after close).
- Bull. The current company answered the deceleration fear that drove the −10.0% Q2 crash: FY2026 revenue accelerated to +17.8% (above FY2025's +14.9%), operating margin expanded +116 bps to 46.78%, net income +31.3% (+22% ex-OpenAI), Azure re-accelerated to +43% in Q4 / +41% for the year, crossed $100B for the first time, and commercial RPO grew +84% to $678B (~2.0× revenue) — the market paid +15.5% for it on the print. Microsoft still throws off more absolute FCF (+$67.0B) than Alphabet (+$53.3B) on ~26% less revenue and is FCF-positive when Amazon (−$11.6B) and Oracle (−$23.7B) are not. You pay a full multiple for the highest-quality, re-accelerating franchise in the group.
- Bear. Concede all of that — none is contested. But over the very year the stock rallied +29.3% (from the 2026-07-10 read $384.36 to this settled $496.88), the buildout moved further from cash: FY2026 capex hit $115,948M (+79.6%, 34.9% of revenue), more than absorbing the +34.4% OCF growth, so free cash flow fell −6.5% to $66,987M — the incremental +$51,397M of capex bought −$4,624M of FCF. That spend is converting to depreciation (up +55.9%; gross PP&E doubled to $431,767M; capex is 3.38× the depreciation recognized), whose leading edge is the −88 bps gross-margin drift to a five-year low. The rally handed the bear the re-rating lever it lacked in July: MSFT is now the cohort's priciest name on EV/sales, at ~55× P/FCF on falling cash and a 1.82% yield, and the market has already erased ~$369.0B in a day on a one-point Azure miss.
- Priced-in break-even. At flat FY2026 operating cash flow ($182,935M) today's price is a mature 30× free-cash-flow multiple iff capex/revenue falls to ~18.1% — below anything in Microsoft's recent history — and even a full reversion to FY2025's 22.9% (from the current 34.9%) only re-rates it to ~34.5× FCF, still a premium. The conclusion flips on whether AI capex/revenue normalizes from ~34.9% toward ~23% while OCF grows (FCF converts, the premium is a quality multiple) or stays capex-suppressed (the 55.1× cash multiple compresses −18% to −46%). No buy/sell/hold — research tooling, not advice.
Watch next — a cold reader can act on these:
| Window | Event | If it lands bullish | If it lands bearish |
|---|---|---|---|
| Tue 2026-10-27 (after close, confirmed) | Q1 FY2027 earnings print — the fulcrum, both cases' falsifier #1; first quarter the market grades "does AI capex convert to FCF" after FY2026 FCF fell −6.5% | Quarterly FCF re-expands and capex/revenue starts down from 34.9%; Azure holds near the +43% Q4 pace / ~+45% cc guide; gross margin stabilizes ~67.2%; revenue in/above the guided $89.85B–$90.95B | Azure decelerates below ~38%; capex steps up again with FCF still falling; gross margin drifts below 67.2% — 27.6× P/E / 55.1× P/FCF is a value trap |
| each print; next 2026-10-27 | Depreciation vs gross-margin trajectory | Depreciation growth (+55.9% in FY2026) falls below revenue growth; the 69.05% → 67.20% gross-margin drift halts near ~67% | Depreciation keeps outrunning gross-margin recovery as the $431,767M PP&E base (~$81.6B/yr not yet expensed) matures into the cost line |
| each 10-Q; next 2026-10-27 | OpenAI/Azure relationship — swinging marks + backlog concentration + removed exclusivity | RPO ($678B) and LongTermInvestments ($36,348M) hold with OpenAI on Azure at scale; the equity-method mark (FY2026 net +$4,963M) stops swinging GAAP NI | OpenAI — now free to buy AWS/Google Cloud/Oracle after the April 2026 non-exclusive restructuring — shifts workloads off Azure; the mark turns to sustained losses (FY2025 was −$3,620M) |
| ~mid-September 2026 | Annual dividend raise + buyback status (cadence; not a standalone 8-K in the index) | Another ~high-single/low-double-digit raise from $0.91 quarterly; the $60B program ($40.6B remaining) continues | Raise/buyback pace slows as the capex draw claims falling FCF (FY2026 returns were 72.7% of FCF) |
| milestones through ~Nov 2026 → Feb 2027 | Cloud-antitrust overhang — FTC probe, EU DMA (Azure, final by Nov 2026), UK CMA business-software SMS (decision by Feb 2027) | Probes close or resolve benign; no conduct remedy on Azure/Copilot licensing | A DMA gatekeeper designation or CMA conduct remedy constrains Azure/Copilot licensing terms |