Target basis: Ten-year unlevered FCF/DCF at a 9.50% proprietary WACC with an explicitly modelled vintage depreciation schedule. The DCF is the preferred basis because the entire investment question here is what $135bn a year of capital earns — a question EBITDA multiples are structurally unable to answer. The multiples read is used only as a cross-check, and it disagrees.
Best growth year in a decade, worst return on the capital that bought it: incremental ROIC has fallen to 14.5% against a 9.5% cost of capital, free cash flow has declined two years running, and the shares have never been more expensive on cash.
Report date: 2026-08-09 · Data as of August 7, 2026 · all figures USD unless noted
Important notice & disclaimer. This report is general, impersonal research, prepared and distributed exclusively to the public. It is not investment advice and not a personal recommendation: it has been prepared without regard to any reader's investment objectives, financial situation, tax position or particular needs, and no advisory, fiduciary or client relationship arises from reading it. The publisher is not authorised or registered by the CNMV and provides no service requiring such authorisation. Readers should form their own view and, where appropriate, consult an authorised adviser before making any investment decision. The data and analysis are drawn from public filings, earnings-call transcripts and private data sources believed to be reliable but not independently audited; they may contain errors or omissions and may differ from company-reported figures. Forecasts, estimates, scenarios, valuation ranges and fair values are opinions about an uncertain future - not guarantees, and not reliable indicators of future results. Investing involves risk, including the loss of the whole amount invested. The publisher accepts no liability for loss arising from investment decisions taken on the basis of this report, save for liability that cannot lawfully be excluded. Who produces this research, how it is valued, what the ratings mean, and our conflicts of interest →
Verdict & quality scorecard
Scored on the fetched FY2026 filings, the four most recent earnings calls and our own cost-of-capital build. The business quality is not in question; the price and the return on incremental capital are.
What the market is pricing in at today's price
At $499.99 the market is not paying for fiscal 2026; it is paying for the next decade of it. Reverse-engineering our own model, the price requires 5.05 percentage points added to every single year of our growth path — FY2027 revenue +23.0% against the +16-17% management has guided for Q1, FY2029 +17.0%, FY2031 +13.5%, FY2036 +9.0%. That is a 14.0% compound revenue growth rate for ten years, taking revenue to $1.23 trillion by FY2036 from $331.8bn today, with EBITDA margins held near 60% and capital intensity falling exactly on our schedule. Microsoft's actual revenue CAGR was 15.1% over FY2020-26, so the market is asking it to repeat roughly its recent growth rate for another decade off a base three times larger, and to do it while normalising the largest capital programme in corporate history. Nothing in that is impossible. What is absent is any margin of safety. The divergence is stark: the Street's average price target is $547.17 with a low of $400 and 66 Buy ratings against zero Sells; our $380 target sits below the lowest published target on the Street.
Fiscal 2026 was Microsoft's best growth year in a decade — revenue $331.8bn (+17.8%), operating income $155.2bn (+20.8%), Azure through $100bn growing 41%, and the stock has risen 28.0% in seven sessions since the print. The same fiscal year consumed $115.9bn of cash capital expenditure against $38.5bn of depreciation and amortisation (3.01x), deployed $135.4bn of total capital equal to 40.8% of revenue, and produced free cash flow of $66.99bn — below FY2024's $74.07bn on 35.4% more revenue. Incremental return on newly deployed capital fell to 14.45% from 23.31%, leaving a spread of just 495 basis points over our 9.50% cost of capital, down from 1,381bp a year earlier. Management simultaneously extended data-centre and office-building useful lives from 15 to 25 years effective FY2027 and reclassified future data-centre leases from finance to operating, which removes them from reported capital expenditure and cut the calendar-2026 capex figure to about $175bn for accounting reasons rather than economic ones. Not one of the six analysts on the call asked about either change, and the one question on return on invested capital produced no number, no framework and no comparison. On accounting earnings the shares sit at the 45th percentile of their own eleven-year range; on cash they sit at the 100th, at 55.4x free cash flow against a prior eleven-year maximum of 51.6x and a 1.80% free-cash-flow yield against a prior minimum of 1.94%. Our DCF says $312 base case; we set the target at $380 to credit the Azure acceleration and the chance that capital intensity peaks sooner than we model. That is 24% below the price.
Earnings quality
Operating cash conversion is superb and improving (CFO/net income 1.37, best in five years) and the opex discipline is real — headcount down 2% against revenue up 18%. But 6.3% of FY2026 diluted EPS is non-operating: other income swung $15.6bn to +$10,697m, including a $3.2bn Anthropic gain, and $0.27 of Q4 EPS came from discrete items versus April guidance.
Balance-sheet strength
Net debt/EBITDA 0.52x, interest coverage 50.9x, debt 1.09% of market capitalisation. Nothing breaks here. The composition is the caveat: 68.7% of $128.8bn of total debt is finance-lease obligations, up from $32.2bn in FY2023, and bonded debt actually fell.
Returns on capital
Full-company ROIC 29.3% versus a 9.50% WACC is still exceptional. Incremental ROIC on FY2026's $133.2bn of new invested capital is 14.45%, a 495bp spread, down from 1,381bp in FY2025 — two consecutive years of decline on the largest capital deployment in the company's history.
Capital allocation
$135.4bn deployed into the build, $22.3bn of buybacks against $12.4bn of stock compensation leaving only 0.16% annual share-count shrinkage, and a $26.4bn dividend. FY2027 free-cash-flow guidance is 'we expect to remain free cash flow positive' from a company that earned $74.1bn of it in FY2024.
Management credibility
Guidance has been met or beaten and the Azure sequence 40% to 43% to a guided ~45% is the strongest in the dataset. Against that: a $37bn AI revenue run-rate disclosed in Q3 and silently dropped in Q4, a useful-life extension announced in the quarter it starts to matter with no quantification, and an unanswered ROI question.
Valuation
27.8x trailing earnings looks like an average version of Microsoft. 55.4x free cash flow is the most expensive it has ever been in eleven years of data, above the prior maximum, and the 1.80% free-cash-flow yield is below the prior minimum. The bridge between the two is $115.9bn of capex against $38.5bn of D&A.
Technical set-up
RSI(14) at 78.13 after a 28.0% seven-session advance, 22.8% above the 50-day and 15.5% above the 200-day, with the 50-day still below the 200-day and volume already fading. Momentum exhaustion, not breakout confirmation.
Previous recommendation. On 2026-08-08 we rated this SELL · fair value $335. This report supersedes it. read it
What has happened since
Nothing published since the last report met the threshold for inclusion. This is the result of a search, not the absence of one.
The rest of this report is one email away
Eight more sections: the earnings-call analysis, returns on capital, the balance-sheet
risks, both valuations, positioning, the technical read and the falsifiable thesis.
No card, no account — we send you a link that opens the whole document for three days.
This is a free sample — the full library is 206 names, auto-updated on earnings, for €39/mo plus VAT. See the plans → · JLPascual Associates