MSFT · NASDAQ
Microsoft Corporation
The default operating layer of the enterprise — now compounding through AI.
A structured, long-term business thesis on Microsoft — analyzed through the six pillars of the Evergreen Framework.
Executive Summary
A generational compounder, re-underwritten for AI.
Microsoft is the closest thing the modern economy has to essential infrastructure. Its software runs the desk of nearly every knowledge worker on earth, and its cloud runs a growing share of the workloads those workers depend on.
What makes Microsoft interesting today is not that it is large — it is that a company of its size is still finding new S-curves. Azure remains early in AI workloads. Copilot attaches AI economics to hundreds of millions of paid seats. And the balance sheet, cash flow, and capital allocation discipline give management the rare ability to invest through a full cycle without breaking the model.
Investment Thesis
Four things that need to be true.
- 01
The enterprise substrate
Identity, productivity, and communications sit inside Microsoft. Removing them is a multi-year migration no CIO wants to run.
- 02
Azure is still early
The AI workload cycle is a fresh chapter — not a continuation of lift-and-shift — and Microsoft is a structural beneficiary.
- 03
AI as a monetization layer
Copilot converts AI from a cost center into a per-seat SKU across M365, Dynamics, GitHub, and Security.
- 04
Owner-operator capital allocation
Predictable buybacks, growing dividend, and disciplined M&A — capex is the swing factor, deployed into the highest-return surface of the company.
Business Overview
Three segments, one operating system.
Microsoft reports in three segments, but the business is best understood as one integrated stack: identity at the bottom, productivity and data in the middle, and AI increasingly wrapping the whole thing.
Productivity & Business Processes
Microsoft 365 (consumer + commercial), Dynamics 365, LinkedIn. The recurring per-seat engine.
Intelligent Cloud
Azure infrastructure and platform services, server products, GitHub, and enterprise support.
More Personal Computing
Windows OEM and commercial, Search & advertising (Bing), Devices, and Gaming (Xbox + Activision).
Business Driver Tree
Where value actually gets created.
Six drivers, one owner-operator lens. Each is a distinct compounding engine — and each reinforces the others.
Competitive Advantages
Four moats, reinforcing each other.
Enterprise distribution
A direct sales motion into essentially every Fortune 2000 company, with bundling leverage across Azure, M365, Security, and Dynamics.
Switching costs
Active Directory, Entra, Exchange, Teams, and the Office data layer make Microsoft the identity and productivity substrate of the enterprise.
AI platform position
OpenAI partnership, first-party models, and Copilot attach across every SKU turn AI from a threat into a per-seat monetization vector.
Cloud scale economics
Azure's global footprint and custom silicon (Maia, Cobalt) drive a unit-cost curve competitors cannot match without similar scale.
Capital Allocation
Where every incremental dollar goes.
Microsoft has been one of the most disciplined capital allocators in large-cap software. The mix today skews heavily to AI infrastructure — intentional, but the return profile is the single most important thing to monitor.
Reinvestment (Capex)
Elevated AI + Azure infrastructure spend; the largest single use of cash and the swing factor for near-term ROIC.
R&D
Consistent double-digit % of revenue — funds Copilot, first-party models, silicon (Maia, Cobalt), and platform software.
Buybacks
Steady, opportunistic repurchases — meaningful reduction in share count over time without dominating capital returns.
Dividends
Modest but reliably growing — signal of durability, not the primary return vehicle.
M&A
Selective and strategic (LinkedIn, GitHub, Nuance, Activision) — bolt-ons that extend distribution or IP, not empire-building.
Key Metrics & KPIs
What to actually watch each quarter.
Directional indicators, not price targets. Order-of-magnitude figures for reference — always cross-check against the latest 10-Q.
Azure growth (YoY, cc)
High-20s%
Constant currency
Commercial RPO
$300B+
Contracted backlog
Operating margin
~45%
Company-wide
FCF conversion
>30% of revenue
After capex
Capex intensity
Elevated
AI infra build-out
M365 seats
400M+ paid
Commercial
Risks
What could break the thesis.
AI capex return profile
Historic capex intensity — if Copilot attach and Azure AI revenue don't scale as fast as depreciation, near-term ROIC compresses.
Regulatory and antitrust
Bundling of Teams, Security, and AI features into M365 draws EU and FTC scrutiny; forced unbundling would erode pricing power.
OpenAI dependency & governance
Strategic and financial entanglement with a single frontier lab is both a moat and a concentration risk.
On-prem to cloud maturation
The easy cloud migration cohorts are largely done; future Azure growth depends on net-new AI workloads, not lift-and-shift.
Long-Term Outlook
Where Microsoft plausibly is in ten years.
Assume the AI capex cycle rationalizes and Copilot attach settles into the 15–25% range across the M365 base. Microsoft in 2035 looks like today's Microsoft with a materially larger cloud, an AI attach fee embedded in nearly every enterprise SKU, and a Gaming business finally earning its acquisition price.
The company does not need to win frontier AI outright to compound. It needs to remain the default distribution and infrastructure layer of the enterprise — which is a much shorter list of things to be right about.
Base case: continued high-single-digit to low-double-digit revenue growth, stable-to-expanding operating margins after the capex peak, and durable per-share value accretion.
What Would Change Our Mind?
The signals we're actively watching.
A thesis is only useful if it can be broken. If any of the following hold for more than a couple of quarters, the framework re-underwrites.
Sustained Azure deceleration to sub-20% growth without a corresponding step-up in operating leverage.
Copilot attach rates stalling below ~10% of eligible seats after two years of GA.
A durable break in the OpenAI relationship without a credible first-party frontier model.
Regulatory action forcing unbundling of Teams, Security, or Copilot from M365 in a major market.
Capital allocation shifting toward large, non-adjacent M&A instead of AI infrastructure and buybacks.