MICROSOFT CORP (MSFT) stock intrinsic value fair value estimate & owner earnings analysis

Microsoft sells business software and subscriptions (Microsoft 365, Dynamics 365, Windows), cloud infrastructure and AI services (Azure), and gaming and search advertising. In fiscal 2026 (year ended June 30, 2026) revenue rose 18% to $331.8 billion, Microsoft Cloud revenue rose 27% to $214.4 billion, Azure grew 41%, Microsoft 365 Commercial cloud grew 17% and operating income rose 21%. What decides the value is how much of that growth reaches owners as cash, because the company is spending at an unprecedented rate to build AI datacenter capacity. What owners actually keep. Cash from operations was $182.9 billion, but cash capital spending was $115.9 billion (it was $64.6 billion a year earlier and $44.5 billion two years earlier), and the company added a further $24.6 billion of datacenter capacity through finance leases, which are repaid through financing cash flows and so never appear in the capital spending line. We count both as capital spending. We also remove swings in working capital, including growth in unearned revenue, and deduct $12.4 billion of stock-based compensation as a real cost. Of the year's $140.6 billion of total spending, about $9.9 billion is treated as spending for growth (in proportion to the $50 billion revenue increase) and the rest as the cost of keeping the business running. The result is owner earnings of $44.7 billion, or $6.00 per diluted share, versus $6.92 in fiscal 2025, $7.55 in fiscal 2024 and $6.91 in fiscal 2021, when revenue was about half today's level. Adding back $1.5 billion of cash interest on debt and $2.5 billion of finance-lease interest gives the cash the business produces before its lenders: $6.55 a share in fiscal 2026, down 10.5% from $7.32 in fiscal 2025, with a five-year median of $7.32. Treating nearly all of the spending as a cost of staying in business is conservative: fiscal 2026 depreciation was $34.3 billion plus $5.4 billion of lease amortization, far below the spending. We have not credited a return to lower spending, because cash capital spending has risen every year since fiscal 2017 and signed commitments point to more. Earlier versions of this analysis used a different basis, and their figures are not comparable. What is locked in. Customers have contracted $684 billion of future revenue, with about 30% to be recognized within 12 months. Against that, Microsoft has signed $329.1 billion of additional leases, primarily for datacenters, that have not yet started (commencing between fiscal 2027 and fiscal 2033, some subject to conditions), $194.1 billion of purchase commitments (mostly datacenter-related; $169.0 billion falls due within a year) and $34.6 billion of construction commitments. These total about $558 billion, the same order of size as the customer commitments, and we credit neither side as a net gain. OpenAI is a large, concentrated counterparty: it contributed $24.1 billion of fiscal 2026 revenue including revenue-sharing and owed Microsoft $6.0 billion at year-end, and Microsoft holds roughly 25% on an as-converted basis, having funded $11.9 billion of a $13.0 billion commitment. That stake is not valued in any case. The IRS has proposed an additional $28.9 billion of tax plus penalties and interest for 2004 to 2013; Microsoft disputes it and says its reserves are adequate. One item we could not reconcile from the filings: cash taxes paid fell to $21.2 billion from $28.7 billion, while the deferred-tax line swung from a $7.1 billion deduction to a $14.2 billion add-back, and we found no explanation in the filing text. We treat it as a timing benefit that reverses in the bear case. What Microsoft owes. In every case we value the business before its lenders and then subtract what it owes: debt at its disclosed fair value of $36.5 billion (face value $46.1 billion, with $9.25 billion of principal due within a year) plus $66.6 billion of finance-lease liabilities, less $76.8 billion of cash and short-term investments. Net borrowings are $26.3 billion, or $3.52 a share. Base case: $94.15 a share. The business before lenders is worth about $97.67 a share; after $3.52 of net borrowings that leaves $94.15. We assume cash before lenders falls another 10.5% in the first year, matching fiscal 2026's decline, to $5.86 a share, and then stays there with no growth. Left out: Azure and AI demand beyond signed contracts, Microsoft 365 price and seat gains, and any fall in capital spending, because the record shows revenue and spending rising together (revenue nearly doubled since fiscal 2021 while owner earnings per share did not rise) and we cannot point to signed evidence that this reverses. What keeps the earnings going beyond ten years is the renewing base of business software and cloud contracts; revenue has risen every year from fiscal 2017 through fiscal 2026. Bear case: $34.89 a share. We use the 2001 to 2002 telecom and internet infrastructure bust as the episode: Cisco's annual revenue fell 15%, from $22.3 billion to $18.9 billion, between fiscal 2001 and fiscal 2002. Here the AI capacity proves overbuilt: Microsoft revenue falls 7.5% in fiscal 2027 and is 15% below fiscal 2026 from fiscal 2028, with no recovery credited. We assume three-quarters of lost revenue falls straight through to cash flow, and the $14.2 billion tax timing benefit reverses. Cash flow before spending (after stock compensation, before lender interest) falls from $179.5 billion to about $161 billion in fiscal 2027 and about $128 billion afterwards. Signed commitments are honored: spending is $150 billion in fiscal 2027, $140 billion, $125 billion and $112 billion in the next three years, then $105 billion a year as purchase commitments run off while the leases keep starting through fiscal 2033. Cash before lenders is then $1.45 a share in fiscal 2027, a loss of $1.64 in fiscal 2028 (covered by the cash on hand, against roughly $4 billion of interest), $0.38, $2.12, and $3.06 a share thereafter. The business is worth $42.29 a share; net borrowings are $3.52 and we deduct the full $28.9 billion IRS proposal ($3.88 a share), leaving $34.89. Bull case: $214.70 a share. The company's own levers execute: revenue grows 10%, 8%, 6%, 5%, 4%, 3% and 3% over fiscal 2027 to 2033 (well below its roughly 15% a year over the past nine years, and backed by the $684 billion of contracted revenue) and then stops growing; cash flow before spending is 52% of revenue (it was 54% in fiscal 2026 and 47% in fiscal 2025); and spending falls from 42% of revenue to 30% by fiscal 2032, the level reached in fiscal 2025, as contracted customers fill the new capacity. Cash before lenders rises from $6.75 a share in fiscal 2027 to $14.66 in fiscal 2033 and settles near $14.29. The business is worth $218.22 a share; after $3.52 of net borrowings that leaves $214.70. Margin-of-safety price: $50. We are sure of the cash Microsoft reports, because cash from operations is in the audited statements; we are not sure how much of the spending is a permanent cost versus a one-time build, or whether the new capacity earns its cost. The worst plausible outcome is AI demand falling short of capacity already committed, with a major customer such as OpenAI unable to honor its contracts and the IRS claim paid; the bear case models a revenue decline, the committed spending and the IRS claim, but not penalties and interest on that claim or a specific OpenAI failure beyond the lost revenue. The company would survive that: net borrowings are $26.3 billion against $182.9 billion of annual operating cash flow, and the $26.4 billion of cash dividends and $22.3 billion of buybacks paid in fiscal 2026 could be cut. If we are wrong about the two things we are most likely wrong about, whether spending stays near $140 billion a year and whether revenue keeps holding up, the shares are worth about $215 if spending falls as contracted customers fill the capacity, or about $35 if demand disappoints. The price assumes a holding period of about five years, and the company paid about $3.55 a share in cash dividends in fiscal 2026. At $50, fiscal 2026's $6.55 a share of cash before lenders would be about 13% a year on the price with no growth at all, and the dividend alone about 7%, so most of a 20% return would come from cash the business already generates. At a much higher price the estimate would depend on a fall in spending or on growth we cannot yet support with signed evidence, and the bear case shows what happens if that does not arrive.