Explore more topics
AI trading task guides

Read filing evidence with Gemini CLI

Why does Amazon's free cash flow differ between the API and its annual report? Compare three filed periods, reproduce the calculation and inspect the accounting explanation.

As of 2026-10-11

Jump to a section

TL;DR

Amazon's FY2025 free cash flow differs because the definitions subtract different capital spending amounts. From $139.514 billion operating cash flow, the API subtracts gross PPE purchases to report $7.695 billion; Amazon subtracts purchases net of proceeds and incentives to report $11.194 billion. Reconcile both definitions against matching annual statements in Gemini CLI and retain the raw values. Operating income increased even as residual cash after investment fell.

Why does Amazon's free cash flow differ between the API and its annual report?

Amazon annual free cash flow definitions

Scale: 0 to 38.219 USD billions · zero baseline

FY2023

  • API FCF+32.22 USD billions
  • Issuer FCF+36.81 USD billions

FY2024

  • API FCF+32.88 USD billions
  • Issuer FCF+38.22 USD billions

FY2025

  • API FCF+7.7 USD billions
  • Issuer FCF+11.19 USD billions

The API subtracts gross PPE purchases; the issuer subtracts PPE net of proceeds and incentives. Both definitions retain their original values.

API FCF, gross PPE
$7.695B
Issuer FCF, net PPE
$11.194B
Show the table · 5 rows, 4 columns

Scroll sideways to compare all columns. Select a heading to sort.

CFO84.946115.877139.514
Gross PPE purchases52.72982.999131.819
PPE proceeds / incentives4.5965.3413.499
API: CFO − gross PPE32.21732.8787.695
Issuer: CFO − net PPE36.81338.21911.194

FY2025 operating cash flow was $139.514B. The API's $7.695B FCF subtracts gross PPE purchases; Amazon's $11.194B definition subtracts PPE purchases net of proceeds and incentives. Both can be reconciled without changing the raw record.

Reproduce this evidence in Gemini CLI

Inspect the request

This worked comparison was checked in NexusTrade and against the original statements. It is not an execution receipt from your external client. Use the connected client capability described in the setup guide, or supply the public statements directly when a tool is unavailable.

Compare three full fiscal periods

Scale: 0 to 716.92B USD · zero baseline

Revenue

  • AMZN · 2023-12-31$574.785B
  • AMZN · 2024-12-31$637.959B
  • AMZN · 2025-12-31$716.924B

Scale: 0 to 77.67B USD · zero baseline

Net income

  • AMZN · 2023-12-31$30.425B
  • AMZN · 2024-12-31$59.248B
  • AMZN · 2025-12-31$77.670B

Scale: 0 to 32.88B USD · zero baseline

Source FCF

  • AMZN · 2023-12-31$32.217B
  • AMZN · 2024-12-31$32.878B
  • AMZN · 2025-12-31$7.695B

Operating margin = operating income ÷ revenue. Not returned means missing, never zero. Source FCF definitions require the reconciliation below.

Show the table · 3 rows, 5 columns

Scroll sideways to compare all columns. Select a heading to sort.

AMZN · 2023-12-31$574.785B$36.852B / 6.41%$30.425B$32.217B
AMZN · 2024-12-31$637.959B$68.593B / 10.75%$59.248B$32.878B
AMZN · 2025-12-31$716.924B$79.975B / 11.16%$77.670B$7.695B

Operating margin = operating income ÷ revenue. Not returned means missing, never zero. Source FCF definitions require the reconciliation below.

Read-only NexusTrade sec_edgar fundamentals retrieved October 12, 2026 and checked against the original annual statements. Amounts below are USD billions; the retained raw response uses USD. These are historical fiscal periods, not a live quote or a forecast.

Follow the calculation into the business

  1. Evidence 1

    FY2025 gross bridge: 139.514 − 131.819 = $7.695B. Net bridge: 139.514 − (131.819 − 3.499) = $11.194B.

  2. Evidence 2

    Amazon's defined FCF: $36.813B → $38.219B → $11.194B for FY2023 → FY2024 → FY2025. Its operating income still increased to $79.975B.

  3. Evidence 3

    AWS revenue: $90.757B → $107.556B → $128.725B. Company revenue includes retail and other businesses; it is not all cloud revenue.

What this result leaves unanswered

  1. Accounting explanation

    FCF is a definition, not an income-statement line. State gross versus net PPE, cash proceeds and incentives before comparing vendors or competitors.

  2. Business and investment limits

    Operating cash flow is not free cash flow. FY2025 net income also includes nonoperating income, so faster net profit growth does not establish identical operating improvement.

Check fiscal periods and public availability

Inspect every input period, publication date and source

Select FY rows by periodEnd instead of assuming the array is latest first. Source statements may present comparative years in one filing. Their later publication does not prove those exact restated values were available at the earlier historical date.

Scroll sideways to compare all columns. Select a heading to sort.

AMZN20232023-12-312024-02-02Annual FY, USD
AMZN20242024-12-312025-02-07Annual FY, USD
AMZN20252025-12-312026-02-06Annual FY, USD

Finish with a claim you can audit

Show the table · 3 rows, 3 columns

Scroll sideways to compare all columns. Select a heading to sort.

The reported business improvedSame-period revenue and operating incomeSource definitions and segment mix
Profit turned into cashCFO reconciliation and PPE scopeWorking capital and noncash items
The stock is attractiveNot established by these statementsPrice, valuation assumptions and competing outcomes

AMZN: FY2025 reported result

Inspect the original amzn: fy2025 reported result

NexusTrade sec_edgar read recorded October 10, 2026. The reporting period ended 2025-12-31; the filing date was 2026-02-06.

Revenue
$716.924B716,924,000,000 USD
Operating income
$79.975B79,975,000,000 USD
Operating margin
11.16%Calculated as operating income ÷ revenue.

Interpret the reported figures

Inspect the original interpret the reported figures

Operating margin describes this reported period. Keep annual and quarterly values separate, select the matching period explicitly, and retain the source filing. The fundamentals array includes older observations and is not ordered latest first.

Request a bounded company brief

Inspect the original request a bounded company brief

Create research/amzn-cash-flow.md as the evidence record. Keep issuer identity, metric, unit, fiscal period, publication date, URL and the exact source passage together. The worked historical release below is a verification exercise, not the company's latest report.

Prompt
Read Amazon (AMZN)'s Q4 2024 release published February 6, 2025. Extract operating cash flow for the twelve months ended December 31, 2024 and December 31, 2023, with USD units and source links. Save research/amzn-cash-flow.md with the subtraction and growth calculation. Distinguish annual cash flow from quarterly earnings. Do not submit research_topic or other paid work before I approve it.

Recalculate Amazon cash-flow growth

Inspect the original recalculate amazon cash-flow growth

Amazon published its fourth-quarter 2024 results on February 6, 2025. Its cash-flow statement reports operating cash flow of $115,877 million for the twelve months ended December 31, 2024 and $84,946 million for 2023.

The difference is $30,931 million. 30,931 ÷ 84,946 × 100 is about 36.4%, consistent with the release's rounded 36% growth. This is operating cash flow, not free cash flow, and the annual period is not the quarter alone. It establishes a historical comparison, not an investment return.

Read filing references and statement periods

Inspect the original read filing references and statement periods

Call sec_edgar twice with the separate argument objects below. list_filings returns filing references; fundamentals reads inline XBRL metrics. Inspect filing dates, fiscal periods and units before comparing them with the historical release. The five recent filings may not include that older release. Missing evidence is a gap to resolve from the primary source, not permission to fill in a number.

Keep these reads separate from ingest_filings, which persists source text. For AMZN, annual periods preserve the twelve-month comparison.

JSON
{
  "ticker": "AMZN",
  "action": "list_filings",
  "forms": [
    "10-K",
    "10-Q"
  ],
  "limit": 5
}

{
  "ticker": "AMZN",
  "action": "fundamentals",
  "periods": "annual"
}

From a question to evidence

  1. Read filings

    NexusTrade tool

    sec_edgar

    Output to inspect

    Issuer, accession, reporting period and publication date

  2. Read relevant news

    NexusTrade tool

    search_stock_news

    Output to inspect

    Dated publisher links; distinguish news from filings

  3. Submit bounded research

    NexusTrade tool

    research_topic

    Output to inspect

    researchJobId and source-backed question

  4. Poll the accepted job

    NexusTrade tool

    get_research_status

    Output to inspect

    Same research_job_id, completed report and citations

Inspect the original from a question to evidence

Scroll sideways to compare all columns. Select a heading to sort.

Read filingssec_edgarIssuer, accession, reporting period and publication date
Read relevant newssearch_stock_newsDated publisher links; distinguish news from filings
Submit bounded researchresearch_topicresearchJobId and source-backed question
Poll the accepted jobget_research_statusSame research_job_id, completed report and citations

research_topic is asynchronous. Poll get_research_status with the same job ID instead of launching another billed research run for the same question.

Submit research only when the request is approved

Inspect the original submit research only when the request is approved

If the primary sources leave a specific question unresolved, show the exact message for research_topic first. The accepted response includes researchJobId and status; save them before ending the session. A completed inline response may already contain citations. Otherwise pass the saved researchJobId as research_job_id to get_research_status.

PROCESSING means poll the same job. COMPLETED returns the report content and citations. FAILED returns an error to investigate. A second research_topic call starts another billed run; a missing response is not permission to duplicate it. If no saved ID exists, inspect the original tool response before considering another submission.

Finish the brief and recover an interrupted session

Inspect the original finish the brief and recover an interrupted session

If the server is missing, check whether you selected user or project scope. Use /mcp reload to rediscover tools after configuration changes. For expired tokens, repeat /mcp auth nexustrade. Shell gemini mcp list can show connection errors. After an interrupted terminal session, reopen the saved run ledger or /resume and continue with the recorded job or deployment ID.

Open research/amzn-cash-flow.md and check each material claim against its link. Keep the original reported value and label each calculation. Leave absent/restated evidence explicit. For a newer-period brief repeat the source checks with the actual latest publication dates instead of carrying these historical figures forward.

The finished artifact should contain the source table, reproducible arithmetic, a bounded interpretation and the next unresolved question. It is a research brief, not a saved strategy or order instruction.

Costs and the next task

Inspect the original costs and the next task

The local brief records your sources and calculations. A research_topic submission can consume credits, so review its exact question and your plan before starting it. Reuse the accepted job ID when polling or recovering a session. Check each report claim against its cited source.

For live deployment, review the separate broker guide, exact trading account and permissions. Choose the broker guide that matches your trading account.

Actual NexusTrade filing-tool output

Inspect the exact returned fields and original requests

On October 10, 2026, we called sec_edgar with the two separate request objects below. The product returned a AMZN 10-K filed 2026-02-06, for the fiscal year ended 2025-12-31, and the FY2025 metric row shown here. We selected that row by fiscal year and period end, rather than assuming the first array entry was latest.

The amounts below are USD, while the linked issuer statement displays them in millions. Revenue and operating income answer different questions from operating cash flow. The compact fundamentals row does not supply operating cash flow; do not relabel freeCashFlow as that metric.

This receipt records a NexusTrade filing read on October 10, 2026. Optional historical exercises retain their own periods and units.

Returned fieldObserved value
TickerAMZN
PeriodFY2025
Period end2025-12-31
Filed2026-02-06
totalRevenue716,924,000,000 USD
operatingIncome79,975,000,000 USD
JSON
{
  "ticker": "AMZN",
  "action": "list_filings",
  "forms": [
    "10-K"
  ],
  "limit": 1
}

{
  "ticker": "AMZN",
  "action": "fundamentals",
  "periods": "annual"
}

Connect Gemini CLI in the intended environment

Documented setup flow for this client
Documented setup flow for this client
Connect Gemini CLI in the intended environment
Scope choice
This command uses user scope. Use project scope if the connection should belong only to this workspace.

With Gemini CLI installed, run the registration command below in your research workspace. Start gemini, enter /mcp auth nexustrade and finish NexusTrade browser sign-in. Use /mcp list to inspect the tools.

What you should see: The user-scoped server is available across projects. Google sign-in and NexusTrade account authorization remain separate.

Shell
gemini mcp add --transport http --scope user nexustrade https://nexustrade.io/api/mcp
gemini mcp list

Optional: check the account before owned-record changes

Optional: check the account before owned-record changes

Skip this step if you only want to inspect the public filing in the next step. Before reading or changing an owned portfolio, ask for a read-only list, inspect one returned ID with get_portfolio and compare its name, strategies and deployment type. An authenticated empty list is valid for a new account.

What you should see: For owned records, you have the intended account response or an authenticated empty response. Resolve an account mismatch before saving anything. Public filing inspection does not require an owned-portfolio read.

Identity boundary
A server list or public filing read does not verify access to your owned portfolios.
Prompt
Use only NexusTrade read tools. Call fetch_portfolios, then get_portfolio with one returned portfolio ID. Report the ID, name, strategies and deployment type. If the authenticated list is empty, report that result. Do not create, edit, backtest, deploy or place orders.

Continue exploring

Discussion

Sign in or create a free account to join the discussion.

No comments yet.

Aurora · AI research assistant

Try this in Aurora

Edit the request, then open it in Aurora. You choose when to send it.