Microsoft at 14x EBITDA: The Market Has Not Priced the Cloud Margin Inflection
A $2.8 trillion company at a sub-15x EBITDA multiple deserves a harder look than consensus is giving it.
Capex tripled to $64.6 billion in three years while free cash flow held steady. Understanding whether Microsoft's AI infrastructure bet is creating or consuming shareholder value.
Microsoft spent $64.6 billion in capital expenditures in FY2025, up from $23.9 billion in FY2022. That is a 170% increase in three years. Free cash flow over the same period moved from $65.1 billion to $71.6 billion, a gain of just 10%.
The arithmetic is stark. Operating cash flow grew from $89.0 billion to $136.2 billion, a 53% increase. But almost all of that gain was absorbed by the infrastructure buildout. The question for shareholders is whether that capital is being deployed at returns that justify the cost, or whether Microsoft is building the most expensive moat in corporate history and hoping the AI payoff arrives before investors lose patience.
Microsoft's capital expenditure has followed a clear trajectory. In FY2022 it spent $23.9 billion. In FY2023 it spent $28.1 billion. In FY2024 it spent $44.5 billion, and in FY2025 it reached $64.6 billion. Each year's total was the largest in the company's history.
The spending is almost entirely on data center infrastructure for Azure and the broader cloud and AI stack. Microsoft is building capacity ahead of demand, betting that AI workloads will require compute at a scale that is not yet fully visible in the revenue line.
For context, $64.6 billion in annual capex puts Microsoft in territory occupied by Aramco and Exxon Mobil as capital intensity goes. Those are capital-intensive commodity businesses. Software companies have historically delivered high returns precisely because they are not.
The Activision Blizzard acquisition in FY2023, which cost $68.7 billion, was a separate capital event. The capex trajectory above reflects infrastructure spending only. Combined, Microsoft has deployed extraordinary capital over a short period.
TickerXray Report
Get the complete Microsoft report with all 12 quantitative models, AI-generated investment thesis, and real-time data.
The following data is from Microsoft's annual cash flow statements.
FY2022: CapEx $23.9B, Operating Cash Flow $89.0B, Free Cash Flow $65.1B, CapEx as % of OpCF 26.8% FY2023: CapEx $28.1B, Operating Cash Flow $87.6B, Free Cash Flow $59.5B, CapEx as % of OpCF 32.1% FY2024: CapEx $44.5B, Operating Cash Flow $118.5B, Free Cash Flow $74.1B, CapEx as % of OpCF 37.5% FY2025: CapEx $64.6B, Operating Cash Flow $136.2B, Free Cash Flow $71.6B, CapEx as % of OpCF 47.4%
Capital expenditure as a percentage of operating cash flow has grown from 27% to 47% in three years. Less than half of operating cash flow now flows to shareholders or the balance sheet. In FY2022, three-quarters did.
Free cash flow declined in FY2023 despite revenue growth, then recovered in FY2024, then dipped slightly in FY2025. The pattern reflects capex consuming most of the incremental operating cash flow that Azure growth generates. If Azure's growth rate slows before capex normalizes, free cash flow could compress meaningfully.
The thesis behind Microsoft's capex surge is that AI-driven cloud demand is entering a sustained upcycle. Azure is growing at 31% year over year. Copilot and other AI products are embedding across the Microsoft 365 suite. The infrastructure being built today is the capacity constraint that, once relieved, enables further growth.
The key return metric is Azure's incremental revenue per dollar of infrastructure deployed. Data centers take 12-24 months from groundbreaking to revenue generation. The capex being spent in FY2025 will primarily generate revenue in FY2026 and FY2027. Investors are being asked to hold the position through a period where spending is peaking and revenue catch-up is still arriving.
Microsoft's operating margin has remained robust despite the capex surge, a critical differentiator. FY2025 operating margin was 45.6%, up from 42.1% in FY2022. Revenue grew from $198.3 billion to $281.7 billion over the same period. The top line is compounding even as infrastructure spending absorbs cash flow.
The risk scenario is not an operating collapse. It is a capex cycle that extends longer than expected while AI revenue adoption is slower than the infrastructure buildout implies. That creates a return profile that looks good eventually but disappoints in the near term.
Azure is the primary justification for every dollar of infrastructure spending. Quarterly revenue growth of 31% year over year, combined with a 16.7% overall revenue growth rate for the company, suggests the cloud business is in genuine structural acceleration.
Microsoft's intelligent cloud segment, which includes Azure, generated roughly $43 billion in quarterly run-rate revenue as of the latest report. If Azure sustains 25-30% growth for two more years, the incremental revenue will be substantial and will begin to absorb the capex deployed in FY2024 and FY2025.
The benchmark question is return on invested capital. Microsoft does not break out Azure segment returns explicitly, but the overall company ROIC of approximately 25-30% suggests the existing asset base earns well. Whether the new data center assets will earn at comparable rates depends on AI workload pricing holding up.
Large hyperscaler customers, including Meta and Alphabet, are also building massive infrastructure. That means Azure faces competition not just from Amazon Web Services and Google Cloud, but also from the possibility that large customers vertically integrate AI compute and reduce third-party spending.
Microsoft's capital return to shareholders has shifted materially over four years. In FY2022, the company returned $32.7 billion through buybacks. By FY2025, buybacks had fallen to $18.4 billion. That is a $14.3 billion annual reduction in direct shareholder return.
The share count has declined modestly but not dramatically. Outstanding shares fell from 7.55 billion in FY2021 to 7.46 billion in FY2025. At that pace of reduction, buybacks are barely keeping pace with stock-based compensation dilution. They are not a meaningful capital return driver at current levels.
The dividend has grown steadily and now pays $3.48 per share annually, yielding roughly 0.95% at current prices. That is a token yield in the context of the overall capital picture. Shareholders who want income are not the target audience for Microsoft at these prices.
The implicit trade-off management has made is: less cash returned to shareholders today, more infrastructure deployed for a larger revenue base tomorrow. Whether that trade-off is value-accretive depends entirely on what Azure earns on those assets over a five-to-ten-year horizon.
The reallocation of cash from shareholders to infrastructure is visible in the data.
FY2022: Buybacks $32.7B, CapEx $23.9B, Ratio (Buybacks/CapEx) 1.37x FY2023: Buybacks $22.2B, CapEx $28.1B, Ratio 0.79x FY2024: Buybacks $17.3B, CapEx $44.5B, Ratio 0.39x FY2025: Buybacks $18.4B, CapEx $64.6B, Ratio 0.28x
In FY2022, Microsoft returned 37% more to shareholders via buybacks than it spent on infrastructure. By FY2025, it spent 3.5x more on infrastructure than on buybacks. The company that was once famous for capital return discipline is now primarily a capital deployment vehicle.
This is not inherently bad. Amazon made a similar choice for years, suppressing near-term returns to build AWS. The question is whether Microsoft's AI infrastructure investment will generate AWS-scale returns. Amazon's answer took a decade to arrive. Microsoft investors may be signing up for a similar wait.
Microsoft's revenue grew from $198.3 billion in FY2022 to $281.7 billion in FY2025, a compound annual growth rate of roughly 12.4%. Operating income grew from $83.4 billion to $128.5 billion over the same period, a 54% increase. Net income grew from $72.7 billion to $101.8 billion.
Operating margin expansion from 42.1% to 45.6% is the most impressive element of the picture. Despite doubling R&D and tripling capex, Microsoft grew its operating margin. That is a testament to the operating leverage in the cloud business and disciplined cost management.
The balance sheet remains conservative. Cash and equivalents of $30.2 billion against $40.2 billion in long-term debt gives a near-net-cash position. Total assets have grown from $333 billion in FY2021 to $619 billion in FY2025, reflecting the Activision acquisition and infrastructure buildout.
Return on equity of 34.4% and return on assets of 14.9% are strong for the asset base. The critical test is whether these returns hold as the new infrastructure enters service. Higher assets with stable returns implies strong capital allocation. Higher assets with falling returns implies capital misallocation.
Microsoft trades at 22.3x trailing earnings and 20.1x forward earnings on consensus estimates of $17.20 in EPS for the current fiscal year. Wall Street's consensus target price is $589.90, with 41 strong buy ratings and 15 buy ratings from 61 analysts. Not a single analyst rates the stock a sell or strong sell.
At a $2.65 trillion market capitalization, Microsoft is the second largest company in the world by market cap. A 22x P/E on a company with 60% quarterly earnings growth and 16.7% revenue growth is arguably not expensive. The PEG ratio of 1.25 is close to fair value territory by traditional measures.
The EV/EBITDA of 14.9x is also reasonable for a high-quality compounder. Google trades near 12x EV/EBITDA. Apple is closer to 25x. Microsoft sits in a sensible range relative to peers given its growth profile.
The valuation risk is not that Microsoft is wildly overpriced today. It is that the capex cycle could suppress free cash flow for longer than consensus models assume, and that the AI revenue wave is partially priced into current multiples already.
The primary risk is that AI infrastructure spending continues to exceed AI revenue generation for another two to three years. If capex stays near $60-70 billion annually, free cash flow is capped near $70-80 billion regardless of operating income improvement. That limits the buyback program and dividend growth.
A second risk is hyperscaler disintermediation. Meta, Alphabet, Amazon, and now Microsoft are all building massive AI compute clusters. If large enterprise customers follow the hyperscaler pattern and build proprietary AI infrastructure, Azure's third-party compute opportunity shrinks precisely as Microsoft finishes building capacity for it.
Copilot adoption has been slower than Microsoft's pricing implied it would be. Enterprise customers have been selective about which Copilot SKUs they purchase. The $30-per-seat Copilot add-on has not achieved the penetration rates originally projected. That matters because high-margin software attach is the return model that makes the infrastructure spend rational.
Finally, competition from Google Cloud and AWS remains intense. Google's Gemini models are competitive with OpenAI's offerings. AWS has its own AI inference infrastructure and a loyal installed base. Microsoft's lead in enterprise AI is real but not unassailable.
Regulatory risk also deserves mention. The Activision acquisition cleared antitrust review, but Microsoft's expanding position in AI, cloud, and enterprise software continues to draw scrutiny from regulators in the U.S. and EU. Any forced divestitures or restrictions on AI bundling could affect Microsoft's ability to monetize its infrastructure investment through integrated products.
Microsoft's capital allocation story often focuses exclusively on Azure. But the company has three major segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing. Each allocates and generates capital differently.
Productivity and Business Processes includes Office 365, Teams, LinkedIn, and Dynamics. This segment generated approximately $77 billion in annual revenue in FY2025 and has been growing steadily. The Copilot add-on is primarily attached to this segment's customer base. If Copilot adoption accelerates here, it could generate high-margin incremental revenue without significant incremental capex.
LinkedIn is an underappreciated asset. It generates hiring, marketing, and premium subscription revenue with strong network effects. LinkedIn's data is also a unique training asset for AI models, giving Microsoft a proprietary data advantage that competitors cannot easily replicate.
The Gaming segment, anchored by the Activision acquisition, adds a large entertainment revenue base. Game Pass subscriptions provide recurring revenue. The integration of Activision titles into the Microsoft ecosystem is still early. The segment diversifies Microsoft's revenue mix but is not the core capex driver.
The key takeaway is that Azure's capex burden is being partially funded by very stable cash flows from Office, LinkedIn, and Windows. The infrastructure bet does not require Microsoft to bet the entire company. It requires patience from shareholders who own a fundamentally strong business making a large long-cycle investment.
Microsoft's capex surge is not reckless. It is a deliberate infrastructure bet on AI-driven cloud demand from a company with the balance sheet, operating margin, and market position to make that bet. The underlying business is compounding revenues at 12-16% annually with expanding margins. That is a strong foundation.
The tension is that free cash flow has effectively flatlined for three years while operating cash flow has grown 53%. Shareholders are being asked to wait while billions flow into data centers that will take years to earn their cost of capital. That is the Amazon AWS playbook. It worked for Amazon, but it required patient shareholders through a very long investment cycle.
At a 20x forward P/E, the market is broadly comfortable with this trade-off. Whether that comfort is warranted depends on whether Azure's AI revenue grows fast enough to absorb the infrastructure before capex discipline returns.
Full forensic analysis of Microsoft
+ 6 more models included
150,000+ stocks covered
Global coverage across 60+ exchanges. Every report includes all 12 quantitative models and AI analysis.
View plansEvery report runs 12 quantitative models and generates an AI investment thesis. From Piotroski scores to manipulation detection -- get the full picture in seconds.
12 forensic models
Piotroski, Altman, Beneish, DuPont & more
AI investment thesis
Synthesized outlook on every stock
Manipulation detection
Spot red flags before they hit the news
150,000+ tickers
Global coverage across 60+ exchanges
Expected return
Forward return projections for every stock
Real-time data
Live prices, insider trades, news sentiment
Free accounts get 1 report per month. Pro gets unlimited.
A $2.8 trillion company at a sub-15x EBITDA multiple deserves a harder look than consensus is giving it.
At 23x earnings, MSFT trades near a multi-year valuation low. The operating margins are historic. The free cash flow math is the whole debate.
At 23.4x trailing earnings with $71.6 billion in FCF, Microsoft looks like a fortress. Five underappreciated risks suggest the margin of safety is thinner than it appears.