Microsoft delivered a strong fiscal Q2 2026, with revenue reaching $81.3 billion (up 17% constant currency) and EPS of $4.14 (up 24% constant currency, excluding OpenAI impact). The Microsoft Cloud surpassed $51.5 billion in revenue (up 26% CC), driven by Azure growth of 39% CC. Key performance drivers included the acceleration of AI adoption, with Microsoft 365 Copilot seats growing 160% year-over-year to 15 million paid seats and GitHub Copilot subscribers reaching 4.7 million. Despite significant capital expenditures of $37.5 billion focused on AI infrastructure, operating margins expanded to 47%. The company raised guidance for the full fiscal year, expecting operating margins to be up slightly, while projecting Q3 revenue growth of 15-17% and Azure growth of 37-38% CC.
| Metric | Value | Change |
|---|---|---|
| Revenue | $81.3 Billion | +17% (CC) |
| Earnings Per Share (EPS) | $4.14 | +24% (CC, Adj) |
| Microsoft Cloud Revenue | $51.5 Billion | +26% (CC) |
| Azure Growth | N/A | +39% (CC) |
| Capital Expenditures | $37.5 Billion | High Historical Level |
| Commercial RPO | $625 Billion | +10% YoY |
| Microsoft 365 Copilot Seats | 15 Million | +160% YoY |
| GitHub Copilot Subscribers | 4.7 Million | +75% YoY |
Microsoft is aggressively pivoting toward an 'Agent-first' architecture, treating agents as the new applications. The introduction of 'Agent 365' and the integration of 'WorkIQ' as a context layer for agents (connecting emails, meetings, and docs) suggests a strategic shift to lock users deeper into the Microsoft ecosystem by automating complex workflows. This is evidenced by the 160% year-over-year growth in Copilot seats and the claim that 80% of the Fortune 500 are actively building agents using their tools.
The company is vertically integrating its silicon strategy to drive superior unit economics and reduce reliance on third-party vendors. The launch of the 'Maya 200' accelerator, which delivers a 30% improvement in TCO compared to the latest generation hardware, signals a strategic intent to control the 'token factory' cost structure. This focus on 'tokens per watt per dollar' is critical for maintaining margins as AI consumption scales.
Microsoft is leveraging its unique 'data moat' via Microsoft 365 to differentiate its AI offerings. By grounding Copilot in 'WorkIQ'—the proprietary graph of enterprise data including communications, artifacts, and relationships—Microsoft claims to deliver 'unmatched' accuracy compared to competitors. This creates a defensive barrier where switching costs become prohibitively high as enterprises increasingly rely on this stateful agent for daily operations.
The company is prioritizing 'Sovereign AI' as a major growth vector, expanding data center presence into seven new countries this quarter. By offering the broadest selection of models (including GPT-5.0.2, Claude 4.5, and region-specific models) and local data residency solutions, Microsoft is positioning itself as the neutral, compliant infrastructure provider for global governments and regulated industries, mitigating geopolitical risk.
There is a growing disconnect between massive capital expenditures ($37.5B this quarter) and immediate Azure revenue recognition, which spooked investors despite the beat. Management admitted that a significant portion of GPU capacity is allocated to internal R&D and first-party products (Copilot, GitHub) rather than immediate external revenue generation. While they argue for long-term LTV, the short-term ROI on these assets remains opaque to investors.
Concentration risk surrounding OpenAI remains a tangible concern, with OpenAI-related contracts constituting 45% of the total commercial RPO ($625B). While management highlights that the remaining 55% is growing 28%, the heavy reliance on a single partner for a massive portion of future revenue introduces significant counterparty and regulatory risk.
The 'More Personal Computing' segment is showing signs of strain, with gaming revenue down 9% and Windows OEM revenue expected to decline by roughly 10% in the next quarter due to 'elevated inventory levels' and the normalization of Windows 10 end-of-support benefits. This indicates potential weakness in the consumer PC refresh cycle and execution challenges in the gaming division.
Management guided for a sequential decrease in capital expenditures but warned that the 'mix of short-lived assets' would remain high. This implies continued heavy cash burn without an immediate corresponding revenue ramp, potentially pressuring free cash flow in the near term as seen in the $5.9B FCF figure this quarter.
Overall: Management exhibited a highly confident and assured demeanor, particularly regarding the long-term viability of their AI stack investment. While Amy Hood remained disciplined and pragmatic on capital allocation, Satya Nadella was visionary, emphasizing the 'age of agents' and the company's unique position in the 'early innings' of AI diffusion. During Q&A, they became slightly defensive but remained firm when pressed on CapEx ROI, reframing the narrative from short-term Azure revenue to long-term ecosystem value.
Confidence: HIGH - Management demonstrated high confidence through specific language about 'demand exceeding supply' and the ability to monetize AI across the entire stack. They provided detailed technical metrics (tokens per watt per dollar) and concrete user numbers (15M Copilot seats) to substantiate their optimism.
$80.65 Billion - $81.75 Billion (Growth 15-17%)
37-38% (Constant Currency)
Down slightly year-over-year
Up slightly year-over-year
Decline roughly 10%
Hedging & Uncertainty: Management utilized temporal hedging to manage expectations regarding capacity and revenue recognition. Phrases like 'early innings,' 'beginning phases,' and 'normal variability' were used to frame current constraints as temporary growing pains. Amy Hood frequently used 'expect to' and 'should' when discussing Q3 guidance, particularly around Windows OEM declines and CapEx timing, to provide themselves room for execution variance. They also hedged the CapEx ROI question by shifting the timeframe from quarterly revenue to 'lifetime value' and 'six-year use life,' effectively deflecting short-term profitability concerns.
"We are in the beginning phases of AI diffusion and its broad GDP impact." - Satya Nadella, CEO
"The key metric we are optimizing for is tokens per watt per dollar." - Satya Nadella, CEO
"Our customer demand continues to exceed our supply." - Amy Hood, CFO
"Agents as the new apps." - Satya Nadella, CEO
"We don't wanna maximize just one business of ours. We wanna be able to allocate capacity... in a way that allows to essentially build the best LTV portfolio." - Satya Nadella, CEO
"Much of that risk... isn't there. Because they're already sold for the entirety of their useful life." - Amy Hood, CFO
"You get more and more and more efficient... the margins actually improved with time." - Amy Hood, CFO
Analyst Sentiment: Analysts were focused on the disconnect between surging capital expenditures and Azure growth rates, expressing skepticism about the return on investment for the massive infrastructure build-out. There was also notable interest in the concentration of revenue related to OpenAI and the durability of that partnership.
Management Responses: Management pushed back firmly against the notion that CapEx should be judged solely by Azure revenue, arguing for a holistic view that includes first-party AI products (Copilot, GitHub) and R&D. Amy Hood provided detailed technical explanations regarding the useful life of contracts and the efficiency gains of aging hardware to reassure investors about long-term margins.
Discussion on the correlation between CapEx spend and Azure revenue growth, with management clarifying that GPUs are allocated to internal products and R&D, not just external Azure customers.
Inquiry into the duration of RPO (2.5 years) versus the depreciation life of servers (6 years), with management explaining that contracts often cover the full useful life of the hardware.
Questions regarding the 45% RPO concentration from OpenAI, with management emphasizing the diversification and growth of the remaining 55% of the backlog.
Inquiries about the new Maya 200 silicon and its strategic importance for cost reduction and performance.
Discussion on enterprise AI adoption trends, specifically regarding 'frontier firms' and the compounding benefits of the Microsoft 365/GitHub/Security stack.
Microsoft remains the premier 'picks and shovels' play for the AI era, offering investors exposure to both the infrastructure layer (Azure) and the application layer (Copilot, GitHub). While the short-term optics of CapEx versus Azure growth may cause volatility, the company is successfully monetizing AI at scale, evidenced by 15 million Copilot seats and accelerating adoption in the enterprise. The expansion of gross margins despite heavy investment indicates strong pricing power and operational discipline. The shift to an 'agent-based' architecture creates a sticky ecosystem that deepens the moat around the Microsoft Cloud, making the stock a compelling long-term holding.
Management indicated that demand for AI compute continues to exceed supply, necessitating nearly 1 gigawatt of new capacity this quarter alone. This suggests a sustained, multi-year boom in data center construction and energy consumption driven by AI workloads.
The shift to 'agents' is driving a new wave of software spending. Microsoft noted that 80% of the Fortune 500 are building agents, and RPO is growing 28% outside of OpenAI, indicating that enterprises are budgeting aggressively for AI transformation.
The PC market is facing headwinds from 'elevated inventory levels' and the normalization of the Windows 10 refresh cycle. Management guided for a 10% decline in OEM revenue, signaling a potential trough in consumer and commercial PC hardware demand.
There is a rising trend toward 'Sovereign AI,' with customers demanding local data residency and specific model choices. Microsoft is responding by expanding data centers into seven new countries, but this trend increases complexity and cost for global cloud providers.