Microsoft: earnings and operating cash flow

This historical source review asks whether cash generation grew alongside earnings. It is not an AI execution benchmark or a current investment recommendation. Source checked: 2026-10-30.

Inputs and source locations

Use Microsoft’s 2025 Annual Report, Income Statements and Cash Flows Statements, for years ended June 30. Values are USD millions. Both columns come from one report; this is not independent corroboration. We reproduce limited factual figures and our calculations, not a copy of the report.

Observation	FY2024	FY2025	Calculated change
Revenue	245,122	281,724	+14.93%
Net income	88,136	101,832	+15.54%
Net cash from operations	118,548	136,162	+14.86%
Cash flow / net income	134.51%	133.71%	-0.79 percentage points

Growth is (current/prior-1)×100; ratios use each year’s cash flow and income. Round only for display.

Method and result

Before writing, record issuer, fiscal year, statement heading, unit and retrieval date for each original amount. A second reader should be able to locate the numbers from that evidence row. Then independently calculate changes.

For these periods, cash flow and earnings both increased; conversion declined slightly. This does not identify the causes, establish recurring free cash flow, or answer valuation. Check the reconciliation and investment spending before broadening the question.

Ask S01: “Compare these rows, retain original locations and separate reported figures from explanations. Do not infer future earnings.” Save a report, evidence table and follow-up list. Supply your actual old thesis separately for S02; financial statements cannot establish what you believed earlier.

Counterevidence and next check

Later collection weakness, a change in consolidation or a restatement could affect comparability. Next review the cash-flow reconciliation and the relevant later filing, keeping its own reporting period. Preserve this historical table when new figures arrive.

Limits and continuation

This page checks arithmetic and source correspondence; it does not certify model performance, installation or a comprehensive accounting review. The S01 package uses fictional DEMO-CONSUMER teaching inputs with different figures. Do not combine these datasets. Methods are free; your model and chosen data may cost extra.

Continue with earnings review and local files.