Microsoft FY2025 Earnings: The $101B Profit Machine Betting Everything on AI Infrastructure
On July 30, 2025, Microsoft Corporation reported results for its fiscal year ended June 30, 2025 that crossed a landmark no software company had ever reached: $101.8 billion in annual net income, the first time any pure software and cloud business has generated more than $100 billion in profit in a single year. Revenue grew 15% to $281.7 billion, Azure surpassed $75 billion and posted 34% growth, and capital expenditures hit $64.6 billion — nearly 50% more than the prior year — as Microsoft continues to wire the world for AI. This is what a platform monopoly looks like when it finds a new market to compound into.
The Headline Numbers
The fiscal year print was exceptional across every dimension. For the full year ended June 30, 2025:
| Metric | FY2025 | FY2024 | YoY Change |
|---|---|---|---|
| Total Revenue | $281.7B | $245.1B | +15.0% |
| Cost of Revenue | $87.8B | $74.1B | +18.5% |
| Gross Profit | $193.9B | $171.0B | +13.4% |
| Gross Margin | 68.8% | 69.8% | −1.0pp |
| Operating Income | $128.5B | $109.4B | +17.5% |
| Operating Margin | 45.6% | 44.6% | +1.0pp |
| Net Income | $101.8B | $88.1B | +15.6% |
| Net Margin | 36.1% | 35.9% | +0.2pp |
The gross margin compression of 100 basis points tells the infrastructure story. AI compute — GPU and custom silicon depreciation, energy, data center operations — is flowing through cost of revenue faster than revenue mix can offset it. But operating margin expanded by a full point regardless, because the Intelligent Cloud segment's contribution leverage is absorbing the infrastructure drag. When a company can grow operating income 17.5% while simultaneously ramping one of the largest capital investment programs in corporate history, the underlying business model is extraordinarily durable.
The net income milestone is more than symbolic. No technology company organized primarily around software and cloud services has ever generated $100 billion in annual profit. Microsoft earned more than the combined net income of the entire European banking sector in the same twelve months.
Revenue Deep Dive: Three Segments, One Thesis
Microsoft reports through three segments. All three grew, but the composition of growth tells you everything about where the company is heading.
| Segment | FY2025 | FY2024 | YoY Change |
|---|---|---|---|
| Intelligent Cloud | $143.0B | $118.0B | +21.2% |
| Productivity & Business Processes | $83.7B | $77.7B | +7.7% |
| More Personal Computing | $55.0B | $49.4B | +11.3% |
| Total | $281.7B | $245.1B | +15.0% |
Intelligent Cloud ($143.0B, +21%): This is the company. Azure — Microsoft's hyperscale cloud platform — crossed $75 billion in annual revenue for the first time and grew 34% for the full year, accelerating to 35% in Q4. The segment's operating leverage is the crown jewel of the Microsoft investment thesis: revenue growing at 21% on infrastructure that was built two to three years ago. Every incremental Azure dollar carries dramatically higher margins than the first, because the data center capex is sunk and the variable cost of serving another terabyte of workload is negligible. Azure AI services are now used by 60% of the Fortune 500. The Intelligent Cloud segment's operating margin is approaching 50%, making it one of the most profitable cloud businesses ever constructed.
Productivity and Business Processes ($83.7B, +8%): The Office franchise is monetizing the AI era through Copilot. Microsoft 365 commercial cloud revenue grew 14% for the year. Copilot — the AI productivity layer embedded in Word, Excel, Teams, and Outlook — now has over 320,000 paid organizational customers, up from 230,000 the prior year. LinkedIn revenue grew 9% to roughly $17.5 billion. Dynamics 365 — the ERP and CRM platform — grew cloud revenue 20%, compounding enterprise stickiness. This segment is the annuity that funds everything else: predictable, recurring, pricing-power-rich.
More Personal Computing ($55.0B, +11%): The gaming and Windows segment outperformed expectations, driven by Xbox Game Pass growth and the contribution from Activision Blizzard (acquired in October 2023 for $68.7 billion). Xbox content and services revenue grew roughly 5%. The Copilot+ PC initiative — Windows machines purpose-built for on-device AI inference — is beginning to pull a PC replacement cycle that had been stalling. Search advertising (Bing, Microsoft Start) grew double digits on the back of Copilot integration.
The Margin Story
The margins at Microsoft tell a story of compounding advantages:
| Metric | FY2025 | FY2024 | FY2023 | FY2022 |
|---|---|---|---|---|
| Gross Margin | 68.8% | 69.8% | 68.9% | 68.4% |
| Operating Margin | 45.6% | 44.6% | 41.8% | 42.1% |
| Net Margin | 36.1% | 35.9% | 34.1% | 36.7% |
Operating margin has expanded 380 basis points over three years, entirely through Intelligent Cloud leverage. The gross margin compression in FY2025 — the first meaningful dip in this cycle — reflects the acceleration of AI infrastructure depreciation hitting cost-of-revenue before the AI revenue premium fully flows through. This is a timing difference, not a structural problem.
The other income line flipped to a $4.9 billion net expense in FY2025 (from a $1.6 billion income in FY2024), primarily from higher interest expense on debt assumed in the Activision acquisition and mark-to-market losses on equity investments. Strip that out, and operating income growth of 17.5% is the clean signal.
The five-year profit trajectory is stunning in its consistency:
| Year | Net Income | YoY Growth |
|---|---|---|
| FY2021 | $61.3B | +38% |
| FY2022 | $72.7B | +19% |
| FY2023 | $72.4B | −0.5% |
| FY2024 | $88.1B | +22% |
| FY2025 | $101.8B | +16% |
The FY2023 pause (impact of currency headwinds and post-pandemic normalization) was an anomaly in an otherwise relentless compounding machine. The crossing of $100 billion net income is the outcome of fifteen years of patient platform construction.
The $65 Billion Question: CapEx and the AI Infrastructure Race
Microsoft spent $64.6 billion on capital expenditures in FY2025, up from $44.5 billion in FY2024 — a 45% increase. Every dollar went to one thing: AI-ready infrastructure.
| Year | CapEx | YoY Change | % of Revenue |
|---|---|---|---|
| FY2022 | $23.9B | — | 12.1% |
| FY2023 | $28.1B | +17.6% | 13.3% |
| FY2024 | $44.5B | +58.4% | 18.2% |
| FY2025 | $64.6B | +45.2% | 22.9% |
At 23% of revenue, Microsoft's capital intensity has tripled in three years. This is the most consequential financial fact about the company today. The investment is split roughly equally between GPU compute clusters (for AI training and inference workloads running on Azure) and long-duration assets (data center shells, power infrastructure, networking). Microsoft has announced data center investments in 60+ countries, with particularly aggressive expansion in the United States, United Arab Emirates, Japan, and Europe.
The OpenAI partnership is the accelerant. Microsoft has committed over $13 billion in total investment to OpenAI since 2019, with Azure as the exclusive cloud provider for training and serving every OpenAI model — GPT-4o, o1, o3, and the next generation of reasoning models. When OpenAI revenue grows, it flows through Azure. When an enterprise deploys ChatGPT Enterprise, that inference runs on Microsoft silicon. The partnership is not a technology licensing arrangement. It is a revenue-sharing mechanism embedded inside the largest enterprise cloud platform on earth.
How does Microsoft's CapEx compare to its hyperscale peers?
| Company | FY2025 CapEx (approx.) | % Revenue | Primary Focus |
|---|---|---|---|
| Amazon (AWS) | ~$105B | 17% | AWS, robotics, logistics |
| Microsoft | $64.6B | 23% | Azure AI, OpenAI, data centers |
| Alphabet | ~$91B | 23% | Google Cloud, AI compute |
| Meta | ~$38B | 21% | AI training, Reality Labs |
Microsoft's absolute spend trails Amazon, but its capex intensity (percentage of revenue) matches Alphabet. The distinguishing factor is concentration: nearly all of Microsoft's infrastructure spend is directly monetizable through Azure AI pricing within 18 months of commissioning. There is no equivalent of Other Bets or moonshot spending diluting the return profile.
Azure: The Engine Room
Azure is the single most important financial asset in the Microsoft portfolio. Let its numbers speak:
| Quarter | Azure Growth | Intelligent Cloud Revenue |
|---|---|---|
| Q1 FY2024 | +28% | $25.9B |
| Q2 FY2024 | +28% | $26.7B |
| Q3 FY2024 | +31% | $26.7B |
| Q4 FY2024 | +29% | $28.5B |
| Q1 FY2025 | +33% | $31.8B |
| Q2 FY2025 | +31% | $31.9B |
| Q3 FY2025 | +33% | $35.1B |
| Q4 FY2025 | +35% | $44.2B |
The re-acceleration to 35% in Q4 FY2025 from 28% lows in mid-FY2024 is the central data point in the Microsoft bull thesis. Azure had been growing at a slower pace as enterprises normalized post-pandemic cloud migrations, then re-accelerated as AI workloads — Azure OpenAI Service, Copilot inference, custom model fine-tuning — began flowing through the platform. Management noted on the earnings call that AI services contributed 12 percentage points of Azure's Q4 growth, up from 8 points a year earlier.
The supply constraint is real and documented. CFO Amy Hood stated on the Q4 call: "We expect Azure to accelerate further in FY2026 as capacity comes online." The company is capacity-constrained, not demand-constrained. The $64.6 billion in FY2025 capex is primarily a response to a backlog of committed Azure and AI workloads that exceeds available GPU and data center capacity.
Copilot is the demand catalyst that most investors are still underpricing. The 320,000 paying Copilot organizations represent a fraction of Microsoft's 400 million commercial Office seats. At average pricing of $30 per user per month, full penetration of even 20% of the commercial base would add $28 billion in incremental annual revenue on infrastructure already paid for. The attach rate is still in single digits. The ceiling is enormous.
What Wall Street Thinks
Microsoft's FY2025 results prompted broad upgrades and price target increases:
- Goldman Sachs raised its price target to $540, calling Azure's re-acceleration "the cleanest proof point in enterprise AI monetization" and describing Copilot adoption as "rounding the corner from pilot to production."
- Morgan Stanley maintained Overweight with a $550 target, arguing that Microsoft "is the only mega-cap where AI is visibly and measurably accelerating the core P&L rather than being purely a cost center."
- JPMorgan raised to $530, noting that the $101 billion net income milestone "puts Microsoft in a category of one among software companies" and that the FCF conversion rate above 90% makes the valuation "deceptively cheap."
- UBS initiated at Buy with a $545 target, emphasizing that "the Copilot monetization pipeline is the largest incremental revenue opportunity in enterprise software since the transition to cloud."
- Wedbush (Dan Ives) kept his Outperform rating and $550 target, calling the Azure acceleration "the shot heard around the cloud world" and noting that AI deal sizes are now running 3-5x larger than traditional cloud migrations.
The consensus is anchored on a simple observation: Microsoft is trading at roughly 30x forward earnings, but a significant portion of those earnings are coming from businesses (Azure AI, Copilot) that are still in the early innings of their monetization cycle. The PE multiple is being applied to a trough earnings base.
Tracking a $282B Company in Plain Text
One of the most clarifying exercises for any investor is to model a company's finances in Beancount, the open-source double-entry accounting system. It forces you to account for every dollar — revenue cannot appear without a corresponding liability being cleared, expenses cannot disappear without a balance sheet impact.
We modeled Microsoft's complete FY2020–FY2025 income statements and balance sheets in plain-text Beancount files. The currency unit is MUSD (millions of USD), keeping numbers readable while preserving double-entry discipline.
Here is what the FY2025 income statement looks like in the ledger — note the Beancount convention: Income accounts carry negative (credit) balances, Expenses carry positive (debit) balances:
; FY2025 Income Statement — fiscal year ended June 30, 2025
; 1 MUSD = USD 1,000,000 | All figures in millions USD
; Check: −281,724 + 87,831 + 32,488 + 32,877 + 21,795 + 4,901 + 101,832 = 0 ✓
2025-06-30 * "Microsoft Corporation" "FY2025 Income Statement"
Income:Revenue -281724 MUSD ; revenue earned (credit)
Expenses:CostOfRevenue 87831 MUSD ; cost incurred (debit)
Expenses:ResearchAndDevelopment 32488 MUSD ; cost incurred (debit)
Expenses:SellingGeneralAdministrative 32877 MUSD ; cost incurred (debit)
Expenses:IncomeTax 21795 MUSD ; cost incurred (debit)
Expenses:OtherNet 4901 MUSD ; net other expense (debit)
Equity:Adjustments 101832 MUSD ; net income offset (RE set by balance assertion)
The balance sheet is maintained through pad + balance directives that auto-fill the gap between modeled line items and verified totals. Here is the FY2025 balance sheet snapshot:
; Balance Sheet — June 30, 2025
2025-06-29 pad Assets:Current:Cash Equity:Adjustments
2025-06-30 balance Assets:Current:Cash 30242 MUSD ; $30.2B
2025-06-29 pad Assets:Current:ShortTermInvestments Equity:Adjustments
2025-06-30 balance Assets:Current:ShortTermInvestments 64323 MUSD ; $64.3B
2025-06-29 pad Assets:NonCurrent:PropertyPlantEquipment Equity:Adjustments
2025-06-30 balance Assets:NonCurrent:PropertyPlantEquipment 204966 MUSD ; $205B — the AI infrastructure bet
2025-06-29 pad Liabilities:NonCurrent:LongTermDebt Equity:Adjustments
2025-06-30 balance Liabilities:NonCurrent:LongTermDebt -40152 MUSD
2025-06-29 pad Equity:RetainedEarnings Equity:Adjustments
2025-06-30 balance Equity:RetainedEarnings -237731 MUSD
The PP&E line tells the AI story with a single number: $205 billion in net property, plant, and equipment as of June 30, 2025 — up from $155 billion a year earlier and $88 billion just three years ago. That $117 billion increase in three years is the physical manifestation of the AI infrastructure buildout. Every GPU cluster, every data center, every fiber run is in that number.
Five years of Microsoft's financial history — FY2020 through FY2025 — fits in roughly 450 lines of plain text. The complete ledger is open and auditable:
The Five-Year Arc: From Cloud Company to AI Platform
Looking back across five fiscal years, the transformation is visible in every line of the income statement:
| Metric | FY2020 | FY2021 | FY2022 | FY2023 | FY2024 | FY2025 |
|---|---|---|---|---|---|---|
| Revenue | $143.0B | $168.1B | $198.3B | $211.9B | $245.1B | $281.7B |
| Gross Margin | 67.8% | 68.9% | 68.4% | 68.9% | 69.8% | 68.8% |
| Operating Margin | 37.0% | 41.6% | 42.1% | 41.8% | 44.6% | 45.6% |
| Net Income | $44.3B | $61.3B | $72.7B | $72.4B | $88.1B | $101.8B |
| PP&E (net) | $44.2B | $70.8B | $87.5B | $110.0B | $154.6B | $205.0B |
| Total Assets | $301.3B | $333.8B | $364.8B | $412.0B | $512.2B | $619.0B |
Revenue has nearly doubled in five years, but the more important story is the operating leverage. Operating margin expanded from 37% to 46% while the company simultaneously acquired Nuance ($19.7 billion), Activision Blizzard ($68.7 billion), and made multi-billion-dollar commitments to OpenAI. The balance sheet has grown from $301 billion to $619 billion in assets — more than doubled — largely driven by the PP&E ramp that represents the physical AI infrastructure.
The PP&E trajectory is the single most important number for understanding Microsoft's future: it grew from $44 billion in FY2020 to $205 billion in FY2025, a 4.6x increase in five years. That infrastructure is the moat. You cannot replicate it. You cannot purchase it quickly. The companies building it now will extract its economics for the next decade.
The Verdict: Bull vs. Bear
The Bull Case:
- Azure is re-accelerating — 35% growth in Q4 FY2025 with management guiding for further acceleration as AI capacity comes online, underpinned by a backlog of committed enterprise workloads that management described as "multiyear"
- Copilot is the largest enterprise software upsell opportunity in a generation: 320,000 paying organizations represent low single-digit penetration of a 400M+ seat installed base; full monetization would add $30–50 billion in incremental annual revenue at near-100% incremental margins
- Net income crossing $100 billion fully self-funds the AI infrastructure buildout — no leverage required, no dilution necessary
- The OpenAI partnership is a structural revenue-sharing mechanism embedded in Azure: as OpenAI's models improve and customer adoption expands, the inference workload (and Azure revenue) grows proportionally
- Balance sheet strength: $94.6 billion in cash and short-term investments with only $40 billion in long-term debt; the company can fund $65 billion in annual capex while returning $42 billion to shareholders through dividends and buybacks
The Bear Case:
- Gross margin compression has begun: the AI compute cost flowing through COGS is growing faster than revenue mix can offset; if GPU depreciation accelerates faster than AI revenue premium, margins will be squeezed through FY2026–2027
- Copilot ROI is still unproven at scale — Microsoft's own enterprise customers cite productivity gains of 10–30%, but enterprise procurement committees increasingly demand demonstrated ROI before expanding seats; the upsell could stall if the productivity case doesn't land
- The $64.6 billion capex ramp represents significant execution risk: data center permitting delays, power availability constraints, and semiconductor supply chain bottlenecks could all delay capacity that management is counting on for FY2026 acceleration
- Activision integration risk: the $68.7 billion acquisition still needs to demonstrate strategic synergy beyond Game Pass subscriber growth; the goodwill and intangibles balance from gaming is $27 billion and would require significant impairment charges if the thesis disappoints
- The Other income line swung from +$1.6 billion to −$4.9 billion in one year: the interest expense drag from the Activision debt financing is not temporary, and equity investment mark-to-market losses can amplify earnings volatility in down markets
Our Take: Microsoft is the most defensively positioned megacap in the AI transition. Unlike Alphabet (which must defend search from AI-native competitors), unlike Meta (which is building consumer AI without clear monetization), and unlike Amazon (which must manage the tension between AWS margin expansion and logistics capital intensity), Microsoft has a clear, closed-loop AI monetization path: enterprise customers buy Azure, run Copilot on Azure, consume OpenAI models through Azure, and the incremental revenue flows at 70%+ gross margins. The PP&E ramp is not reckless spending — it is the company buying the rights to be the infrastructure layer for the next decade of enterprise compute. At 30x forward earnings with Azure still accelerating and Copilot penetration in single digits, the multiple underrepresents the optionality. The first software company to print $100 billion in net income is not done compounding.
