How to Analyze a 10-K Filing with AI: A 2026 Microsoft Example

Research cut-off: 3 November 2026. Figures are reported-period data, not live prices.

An AI-assisted 10-K analysis is useful only when its claims can be traced back to the filing. Start with the business model and risk factors, then reconcile the income statement, balance sheet and cash flow statement. Microsoft Corporation (Nasdaq: MSFT) provides a current example: its FY2026 annual report reports operating cash flow of $182.935 billion, capital expenditures of $115.948 billion and a simple cash-flow-less-capex measure of $66.987 billion. Those are three different measures and must not be collapsed into a claim about earnings.¹

Start with the latest filed annual report—not a search snippet

Open the original 10-K and record the fiscal year-end, filing date and reporting currency. Microsoft's FY2026 ended on 30 June 2026; its subsequent first-quarter FY2027 filing covers the three months ended 30 September 2026. An assistant that treats FY2027 Q1 as part of FY2026 is joining incompatible periods.¹³

Read Item 1 Business for what the company sells, Item 1A Risk Factors for what could go wrong, Item 7 Management's Discussion and Analysis (MD&A) for management's interpretation, and Item 8 for audited financial statements. The SEC's investor guide explains why the annual report is more than a list of earnings figures.⁴

Build an evidence ledger before asking for a summary

For every relevant figure, keep the metric name, unit, fiscal period, filing URL and the exact section where it appears. Distinguish reported financial statement lines from a ratio calculated by the analyst. This prevents a familiar failure mode: a neat AI narrative in which the number is plausible but its period or accounting definition is wrong.

Interpret the movement rather than just the ending number

Operating cash flow increased materially in FY2026, but capital expenditures grew even faster. On this narrow cash capex definition, the residual fell from $71.611 billion to $66.987 billion despite higher cash generated from operations. That is the research finding; it is more informative than a statement that 'Microsoft cash flow was strong.'¹²

A separate question is what assets the spending created and when they will generate returns. Microsoft's FY2026 risk factors discuss cloud and AI infrastructure investment and the possibility that demand, competition and technology change its economics. That is why an annual-report analysis should connect numbers with management's stated risks.¹

Use AI for retrieval, then require a second-pass reconciliation

A good first prompt asks an assistant to identify what changed in the latest FY2026 10-K relative to FY2025, cite each movement and mark calculated ratios. The next asks for counter-evidence: whether commitments, financing arrangements, cash taxes, accounting estimates or concentration risks complicate the conclusion. Finally, ask it to list what the report does not establish.

Financial statements are historical and report-specific; they cannot prove the ultimate return on new capital expenditure. Do not present a forward return-on-invested-capital estimate as a filing fact. If an AI answer cannot point to an original page or section for an asserted number, treat that statement as unverified and rerun the source search.

Read the 10-Q after the 10-K

The annual report establishes the base year, while the next quarterly report tests whether the trend continued. For Microsoft, the Q1 FY2027 10-Q is already available for the quarter ended 30 September 2026. It is a separate unaudited interim period, not an update to the audited FY2026 totals.³

November 2026 update: the latest 10-Q challenges a simple growth story

Microsoft's FY2027 Q1 cash-flow statement, covering 1 July to 30 September 2026, offers a useful test of the prior year's capital-investment thesis. Operating cash flow was $58.000 billion, compared with $45.057 billion in the year-earlier quarter; additions to property and equipment were $39.367 billion, versus $19.394 billion. The simple difference narrowed from $25.663 billion to $18.633 billion. In percentage terms, operating cash grew about 28.7%, while cash capex more than doubled. This is a stronger current-period test than merely repeating the FY2026 annual figures.^{Q1 FY2027}

That evidence supports two reasonable, competing interpretations. An optimistic analyst could argue the rise in capital expenditure reflects investment in future cloud and AI demand. A cautious analyst could reply that near-term incremental spending is consuming more of operating cash and that its return is not yet measurable. The filing supports both factual premises; it does not establish which valuation conclusion will ultimately prove correct.

How to test whether an AI summary understands the period

A practical audit is to ask the assistant to reproduce the $18.633 billion calculation and show the precise two source lines. Then ask for the comparable September 2025 quarter, not the June 2026 annual year-end. A result that answers with $66.987 billion is using the annual base year rather than the requested quarter. That is a report-period error even though the annual number itself is correctly extracted.

Next, check the 10-Q's notes on property and equipment. Microsoft disclosed $26.5 billion of purchases of property and equipment remaining in accounts payable at 30 September 2026, compared with $26.7 billion at 30 June 2026. Cash paid for additions and investments incurred or payable need not match period by period. That timing difference is why we should not silently equate cash capex with total AI infrastructure commitments.^{Q1 FY2027}

The resulting thesis is conditional: Microsoft is funding a large investment cycle from strong operating cash flows, but the most recent quarter reinforces the near-term cash-conversion pressure. An AI 10-K workflow becomes valuable when it detects that distinction, cites both report periods and states which future numbers would reverse the interpretation.