Loading today’s stories
Invest1 distinct publisher3 min readPublished
Profit grew nearly a third in fiscal 2026 and the shares are worth less than a year ago. What the market marked down was the capital spending, not the results.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
Work backwards from the 31%. The prior year's net income was about $102 billion [1][1], and at about 37 times earnings that put the market capitalisation near $3.76 trillion [2][3][2]. The distance between that and today's figure is roughly $169 billion of value gone while the earnings base grew by a third [2][3]. The multiple did all of it and a little more: down about 27%, slightly steeper than the profit growth it was applied to [3][4].
The specific thing being repriced shows up in cash conversion. Reported earnings turned into about 50 cents of free cash flow per dollar in fiscal 2026, against roughly 70 cents the year before [1][7][1][5]. Chief financial officer Amy Hood's calendar 2026 plan of about $175 billion in capital expenditure and finance leases runs some $20 billion above the $155.2 billion of operating income the entire company produced last fiscal year, and she has signalled further growth in fiscal 2027 [5][6][6]. Almost all of that spending returns as depreciation against future earnings, and whether it earns its keep depends on AI computing demand being large enough in a few years to fill the capacity being built now [8].
There is also a quality question inside the 31%. Gains on the OpenAI stake added about $5 billion to fiscal 2026 net income, and the fourth quarter carried a $3.2 billion gain on the Anthropic holding [9]. Take both out and net income is nearer $125.5 billion, which puts the shares closer to 29 times earnings than 27 [7]. The same counterparty appears on the demand side: the commercial backlog reached $678 billion, up 84% year over year [10], and Microsoft said in January that about 45% of the then-current figure came from OpenAI alone [11]. The Motley Fool reports no updated share against the June number, so the largest visible block of contracted demand is being read off a stale disclosure.
The Apple comparison is sharper than the two multiples suggest. At about 35 times earnings and $4.51 trillion, Apple's implied trailing profit is roughly $129 billion [12][8], marginally below what Microsoft earned, and yet it carries about $920 billion more market value [8]. Investors are not paying that premium for profit. They are paying it for profit that arrives without a depreciation schedule attached.
Which leaves the two explanations on offer sitting awkwardly beside each other. The Motley Fool attributes part of the discount to a broad derating of software on generative AI disruption fears [13], and part to capital intensity [8]. Those point in opposite directions: if the threat is disruption, heavy compute spending is the defence; if the problem is the spending, the defence is the complaint.
Ranked by verification strength, evidence, and original report placement.
In fiscal 2026, the year ended June 30, Microsoft revenue rose 18% to $331.8 billion and net income grew 31% to $133.7 billion.
Fiscal 2026 operating income rose 21% to $155.2 billion, faster than revenue.
Almost every dollar Microsoft spends on data centres returns over time as depreciation that reduces future earnings, and the payoff depends on AI computing demand staying strong enough in a few years to fill the capacity being built today.
Microsoft said in January that about 45% of the then-current remaining performance obligation figure came from OpenAI alone.
Microsoft's market cap was about $3.59 trillion at the time of writing, roughly 4.5% below where it was a year earlier.
A year ago investors paid about 37 times earnings for Microsoft; they now pay about 27 times.
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Detailed but single-source figures with no primary filing cited
The numbers are specific, internally consistent and presented as company disclosures (revenue, operating income, net income, capex, cash flow, backlog, guidance), which is why this scores above the midpoint. It does not score higher because the entire cluster is one retail-investing column: no 10-K, press release, transcript link or second publisher corroborates any figure, and the analytical claims about depreciation drag and a sector-wide software de-rating are asserted without quantification.
Heavy contracted and capital commitment, concentrated in one counterparty
Adoption of AI-era cloud capacity is unusually well evidenced for a valuation story: Azure past $100 billion growing 41%, $678 billion of contracted commercial backlog up 84%, $115.9 billion spent and ~$175 billion guided. It is held below the high end because the disclosed concentration — about 45% of the January backlog from OpenAI alone — means a large share of the demand signal rests on a single counterparty, and no current concentration figure is given for the $678 billion total.
Framing slightly dramatises a modest decline; analysis is restrained
Mildly positive rather than aligned. The headline framing of '$169 billion of market value lost' dramatises what is a 4.5% move, and adjectives such as 'possibly one of the best years in its history' overreach the data given that roughly $8 billion of the profit came from OpenAI and Anthropic investment gains. The gap is small because the substantive argument is conservative and self-limiting: the author quantifies the cash-flow deterioration, discloses the investment-gain distortion and the OpenAI backlog concentration, and lands on 'more of a hold than a buy' rather than a promotional call.
Retail-investing publisher issuing an explicit stock rating
Measured from what the source itself shows: the publisher is a subscription retail-investing outlet, opens with a live price quote, includes a 'Key Data Points' module and closes with a personal recommendation ('more of a hold than a buy'). That business model rewards timely mega-cap stock commentary and rating calls. It is mid-range rather than high because the piece takes a cautious, non-promotional stance, volunteers the investment-gain adjustment and the OpenAI backlog concentration, and no undisclosed position, sponsorship or vendor relationship is visible in the supplied material.
Coherent arithmetic, one publisher, no corroboration
Confidence sits just above the midpoint. The internal arithmetic chain — growth rate, multiple compression, cash conversion, adjusted earnings — is consistent and reproducible from the disclosed figures, and the material disclosures are the kind a company reports rather than a writer invents. But a single publisher, no primary-document citation, one dated concentration figure and one wholly unevidenced causal claim keep this well short of high confidence.
invest
A $2B bet that voice replaces the text box, not just the keyboard3 distinct publishers
invest
The AI moat is now a balance sheet, so price the financing and not the model1 distinct publisher
product
The classroom channel nobody bid out: teacher training as Big Tech distribution1 distinct publisher
product
Rillet's $100M reads as proof mid-market ERP is rip-and-replace, mostly at the cheap end1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 26, 2026