Amazon.com (AMZN · Nasdaq)

Full report

1. Executive Summary

Key takeaways: We rate Amazon.com Underweight, with a fair value range of $110–190 per share against a price of $266.93, because the price requires about $323 billion of annual free cash flow after stock compensation from 2032, which is 1.5 times the $213 billion in our base scenario for 2031. AWS (Amazon Web Services, the cloud unit) grew 39% in Q3 2026 with a backlog of $578 billion, and our base scenario already assumes AWS revenue of $511 billion in 2031. The financing of that growth is visible in trailing free cash flow of −$42.4 billion and long-term debt of $181 billion, up from $69 billion at the end of 2025. Our valuation is sensitive to the discount rate: the price equals our bull scenario only at a discount rate of 8.7%, against the ABiS house rate of 10.5%.

Our three calls are set separately. Intrinsic value view: Expensive, because the price is 40% above the top of our fair value range. 1–3 month market trading direction: Sideways to lower, with high volatility, because the Q4 2026 report (expected 4 February 2027) falls inside that window and our companion event study found a median excess return of −3.7% against the S&P 500 over 21 trading days after past reports. Investment action: no new money, which maps to a Rating of Underweight, and our confidence is Medium. Earnings credibility is graded B, because operating income is confirmed by cash flow while reported net income includes $68.4 billion of non-operating gains over nine months.

Our view: the market is pricing Amazon as if returns on its $220 billion of 2026 capital spending will stay near the levels of the last decade, and our reverse DCF (the cash flows the price implies) shows that 72% of the price pays for growth that has not yet been earned. Over the next three months the variable we expect the market to trade first is the growth rate of AWS, followed by the 2027 capital spending plan, because the Alphabet decline of 12.6% on 29 October moved attention to those two variables. The expectations gap below is the core of our disagreement with consensus.

The strongest points for and against our call are these.

• Bull points. AWS growth has risen for five consecutive quarters to 39%, with a Q3 2026 operating margin of 38.7% and a backlog of $578 billion with a weighted remaining life of 6.1 years. Advertising revenue grew 19.8% in the first nine months of 2026, and the operating margin outside AWS rose to 6.7% from 5.8% in fiscal 2025. Stakes in Anthropic, OpenAI and other private companies carry at $294 billion, and we count them at half of that, or $13.5 per share.
• Bear points. Capital spending of $215 billion over twelve months was 1.3 times operating cash flow of $168 billion, and total purchase commitments of $884 billion exceed Amazon's entire operating cash flow for five years. Google Cloud revenue grew 103% to $30.7 billion in Q3 2026 against AWS growth of 39% on $46.0 billion, so we think AWS is losing share even as it accelerates. Reported net income overstates earning power, and the AI-lab investments make Amazon an investor in some of its largest customers.

The independent-view test is met. Our largest single disagreement with consensus is the price of the growth, and the market erred, in our view, by extrapolating the return on past capital spending to a build 2.7 times the size of 2024's. We would find out we were wrong first through the AWS operating margin holding at or above 38% through Q3 2027 while depreciation rises, which would show returns on the new capital above our assumptions.

2. Business Overview

Amazon earns 61% of its operating income from AWS, which is 22% of its revenue, and the remainder comes from a retail and advertising business with a thin margin and large scale. In the first nine months of 2026 revenue was $584.1 billion, up 16.0%, and operating income was $79.4 billion, up 44%. Amazon has 1.58 million employees (Robinhood, company profile) and is led by chief executive Andy Jassy.

By product line, online stores contributed $203.2 billion (34.8% of nine-month revenue), third-party seller services $134.9 billion (23.1%), AWS $125.8 billion (21.5%), advertising $56.7 billion (9.7%), subscriptions $41.2 billion (7.0%), and physical stores and other $22.4 billion (3.8%). Advertising and AWS are the profit engines, and both grew faster than the company, at 19.8% and 35.1%. Concentration data for AWS are partial. Anthropic has committed more than $100 billion over ten years to AWS (Reuters), and Goldman Sachs cited a total commitment from OpenAI of $138 billion over eight years, which together would equal about 41% of the $578 billion backlog if both are fully included, a point we could not confirm.

3. Business and Competitive Analysis

Cloud infrastructure demand is the dominant driver of Amazon's value, and AWS is growing quickly while its rivals grow faster. AWS revenue of $46.0 billion rose 39% in Q3 2026, the fastest growth in nearly five years, while Google Cloud revenue of $30.7 billion rose 103% (Alphabet Q3 2026 results, as reported by aiweekly.co), and Microsoft reported Azure growth of about 39% for its fiscal Q2 2026 (Fortune). We could not obtain Azure growth for the latest quarter. AWS has led in net new cloud dollars in some recent quarters (Quilter, October 2025), but its growth rate trailed Azure and Google Cloud in July 2025, which was the weakest day-one reaction in the set studied in our event study (−9.4% against the S&P 500).

Our moat assessment is a wide moat. Amazon's earnings power value (EPV, the value of current earnings capitalized with no growth) supports it: EPV is 1.87 times adjusted net assets, which is the financial signature of a franchise whose earnings are worth more than the capital invested in it.

Growth engines are the AWS backlog, advertising, and margin expansion in retail logistics. The point that would disprove our moat score is a sustained decline in AWS operating margin below 30% with growth above 35%, which would show that scale alone no longer earns excess returns. The main competitive risk is the capacity race: the four large cloud providers plan combined capital spending above $700 billion in 2026 (Reuters, cited by Investing.com), and returns on a build of that size are unproven.

4. Management, Governance, and Capital Allocation

Governance risk is moderate and mainly legal. We found no restatement, auditor issue or short-seller report in our searches, and we did not obtain the audit opinion or the internal-control assessment. The Federal Trade Commission (FTC) has a monopolization case set for trial on 29 March 2027 (MLex) and filed a separate complaint in August 2026 with 22 state attorneys general over reserve prices in sponsored advertising auctions (Amazon Form 10-Q, Q3 2026). Founder and chairman Jeff Bezos may sell up to 15 million shares under a plan that runs to 26 February 2027, which is about $4 billion at the 9 November close. Senior departures in AI included Rohit Prasad in December 2025 and David Luan in February 2026, and Amazon announced about 14,000 corporate job cuts in October 2025 and about 16,000 in January 2026.

5. Financial Analysis and Earnings Quality

5.1 Financial trend

Operating income more than doubled between fiscal 2023 and the latest twelve months, while free cash flow moved from $32 billion to −$47 billion as capital spending quadrupled. Operating margin rose from 6.4% to 13.1%, and the cost of that margin is now in capital spending, which reached 128% of operating cash flow over the latest twelve months.

Amazon does not split its capital spending between maintenance capex and growth capex. We use depreciation of $79.4 billion over twelve months (37% of capital spending) as a proxy for maintenance, which leaves about $136 billion of growth capital spending. Amazon also invests in the customers of its cloud unit, so free cash flow needs to be read before and after those outflows. In the first nine months of 2026 free cash flow was −$58.3 billion, and it was −$113.3 billion after $55.0 billion of purchases of OpenAI and Anthropic stock.

The balance sheet has changed quickly. Cash and marketable securities were $114.2 billion at 30 September 2026, and the face value of long-term debt was $180.9 billion, up from $68.8 billion at 31 December 2025 after $115.1 billion of debt issuance in nine months. Finance leases add $14.5 billion and operating leases $100.3 billion, which we exclude from debt. Net debt is $81.2 billion before our stakes, interest expense was $1.7 billion in Q3 2026 against $0.5 billion a year earlier, and operating income covers it 16.6 times. Gross debt is 1.1 times trailing EBITDA (earnings before interest, taxes, depreciation and amortization), which is manageable, and Moody's affirmed the A1 rating with a stable outlook in February 2026 while expecting external funding to be needed. Total commitments, including purchase obligations, are $883.6 billion. Peer comparison data (multiples and margins of Microsoft and Alphabet) were not obtained for this report.

5.2 Earnings quality

Amazon's operating earnings are confirmed by cash, and its reported net income is not. We grade earnings credibility B. Operating cash flow of $168.3 billion over twelve months is 1.6 times operating income, and the accrual ratio is negative (−7.6% for fiscal 2025, net income less operating cash flow over total assets). Net income, however, includes other income of $68.4 billion in the first nine months of 2026, of which $62.9 billion is upward adjustments to private investments after Anthropic funding rounds, so we value operating earnings only.

6. Valuation

Key takeaways: Four methods place Amazon's value below its price, and the fair value range we use is $110–190 per share. Our scenario DCF gives a weighted value of $144, with a bear scenario of $81, a base scenario of $140 and a bull scenario of $191, and no cell of the base-scenario sensitivity grid exceeds $187. The result depends on the house cost of capital of 10.5%: the price equals our base scenario at a discount rate of 7.3% and our bull scenario at 8.7%. All figures use a price of $266.93 (9 November 2026 close, Yahoo Finance), 10.929 billion diluted shares, and net cash of $65.9 billion after counting the AI stakes at half of carrying value.

6.1 What the price implies

At a discount rate (WACC, the weighted average cost of capital) of 10.5%, the price of $266.93 implies an enterprise value of $2,851 billion, which is the market value of $2,917 billion plus net debt of $81.2 billion less $147.1 billion for our half-value count of the AI stakes. After the $236 billion present value of our base-scenario free cash flow for 2027–31, the price requires steady free cash flow of about $323 billion a year from 2032, growing 3% a year. At an 18% free cash flow margin that is revenue of $1.8 trillion, or a revenue growth rate of 17% a year from fiscal 2026 (estimated at $827.6 billion); at 14% it is $2.3 trillion and 23%; and at 10% it is $3.2 trillion and 31%.

The price decomposition tells the same story. Normalized earnings (operating income of $104.4 billion over twelve months, taxed at 19%) are $7.74 per share, and capitalized at 10.5% without growth they are worth $73.67 per share. PVGO, the share of the price that pays for growth not yet earned, is therefore $193.26, or 72% of the price. Growth must come from AWS capacity that Amazon is paying for with debt, and our reverse DCF places the required growth beyond what the base-rate studies show for companies of this size, where more than 15% a year for five years has almost never been observed above $100 billion of revenue (empirical base rates, approximate).

6.2 Scenario DCF

The three scenarios are internally consistent sets of assumptions, built from AWS and non-AWS revenue, segment margins, capital spending, and depreciation that follows capital spending with a lag. Free cash flow is after stock compensation, taxes are 19%, and the explicit forecast covers fiscal 2027–31 (the stub of fiscal 2026 is excluded). From 2032 to 2036 growth fades to 3.0%, the operating margin is held, and returns on new capital converge to the 10.5% cost of capital. The bear scenario's revenue growth of 4.7% a year sits near the 30th percentile of the base-rate distribution, our base scenario's 11.5% sits above the 80th percentile, and the bull scenario's 14.4% is above the 90th, so the structural reasons for both are the $578 billion backlog and management's guidance of AWS growth above 40% in 2027 and 2028.

Scenario weights follow from our evidence grades. We grade the bull scenario Medium-strong, on five consecutive quarters of AWS acceleration to 39% with a 38.7% margin, the $578 billion backlog, and management's guide of more than 40% for 2027 and 2028. We grade the bear scenario Medium, on trailing free cash flow of −$42.4 billion, $883.6 billion of commitments, and peer declines of 10% to 14% when capital spending rose without a matching acceleration, but without direct evidence yet of falling AWS demand. The bull scenario is stronger by one step, which gives weights of 20% bear, 50% base and 30% bull. At the next weights on either side the weighted value is $138 (25 / 50 / 25) and $149 (15 / 50 / 35).

6.3 Earnings power and economic profit

Earnings power value (EPV) gives $79.70 per share with no growth and $87.07 with growth of 3% at a 14% return on new capital, and economic profit (EVA) gives $62.69. EVA is the value of invested capital plus the present value of returns above the cost of capital. Amazon's return on invested capital is 21.1% (normalized operating profit after tax of $84.5 billion on operating invested capital of $400.6 billion), 10.6 percentage points above the cost of capital, and we fade that spread to zero over ten years. Both checks sit below the scenario DCF because they give little credit for the growth that the backlog supports.

6.4 Summary and conclusion

We rely most on the scenario DCF because it models the capital spending and the backlog directly, and the other methods act as a floor. The methods agree on direction, and their range from $63 to $191 reflects the disagreement over whether returns on the new capital persist. We apply the pre-registered rule to the range of $110–190: the price is above 1.15 times the top of the range ($219) and below 1.5 times ($285), which gives the label Expensive. Street targets average $328 and are shown for reference only. We could not obtain peer multiples, so we did not run a relative valuation.

7. Analyst View Summary

The Street is overwhelmingly positive, and its average target sits 23% above the price. Of 71 analysts, 67 rate the shares Buy, 4 Hold and none Sell, with targets from $230 to $405 and a mean of $327.80 (Robinhood, updated 7 November 2026). After the Q3 2026 report several firms cut targets while keeping their ratings, including Citizens to $275 from $315 and Raymond James to $250 from $300, according to coverage we reviewed.

We disagree with consensus on three points. First, the lowest Street target ($230) is above the top of our fair value range, and reaching it requires a discount rate of about 9.4% in our bull scenario. Second, consensus earnings per share include gains on the Anthropic stake, which made reported EPS of $5.75 in Q2 2026 against an estimate of $1.82, so earnings per share is an unreliable anchor. Third, the Street frames the debate as AWS growth, which is not in doubt, while we frame it as the return on $220 billion of annual capital spending. We could not obtain consensus for Q4 revenue, AWS growth, or 2027 capital spending.

8. Recent News and Catalysts

The last three months produced four developments that bear on the thesis. Amazon reported Q3 2026 results on 29 October, the day after Alphabet's report took Alphabet down 12.6%, and Amazon fell 4.2% before its own report and closed 1.4% lower after it. The US midterm election on 3 November produced a Democratic Congress, and technology shares fell on 5 November on fears of AI regulation and paused data-center construction (Motley Fool). Amazon announced a Prime benefit with Uber Eats on 5 November. In July 2026 Amazon sold $25 billion of bonds, reportedly its last debt issue of 2026 (Investing.com).

8.1 Forecast register

The register lists forecasts that can be checked on a date. We record the original forecasts and will add results and error attribution at each validation date without rewriting them.

9. Investment Conclusion, Counter-Case, and Position Sizing

Key takeaways: We rate Amazon Underweight, and we would not buy the shares with new money at $266.93. Our fair value range of $110–190 and weighted value of $144 imply an expected return of −46%, and none of our three scenarios reaches the price. The case could fail if the market is right about the discount rate or if returns on the new AI capital stay above 20% for a decade, and we have weighted both through the bull scenario at 30%.

9.1 Conclusion

The three calls differ for a reason. Our intrinsic value label is Expensive because the price is above 1.15 times the top of the range. The direction label is Sideways to lower, because the 4 February report is an event with a thin and two-sided history, and Amazon has already recovered 5.5% since 27 October while the S&P 500 fell 2.2%. The rating is Underweight. Our action rule, applied to an Expensive label with high uncertainty, is no new money, and we would turn Neutral at a price of $218 or lower, where the price falls inside 1.15 times the top of the range, and Overweight below $93.

If today were a cash decision, we would not buy Amazon, because the price discounts our bull scenario at 8.7% against a cost of capital we set at 10.5%, and we see no margin of safety. We rate confidence Medium: the methods agree on direction, but their level depends on returns on capital that cannot be verified until 2028.

9.2 Counter-case and pre-mortem

If this call fails a year from now, these are the three most likely reasons. The first attacks our core thesis directly.

We have written the first failure into our bull scenario at a 30% weight. The third failure adds at most $13.5 per share at full carrying value of $294 billion, which cannot close a gap of more than $75 per share, and we weight it accordingly.

9.3 Position sizing

Our expected return is −46%, computed from the weighted value of $144, and all three scenarios are below the price (bear −70%, base −47%, bull −29%). No scenario shows a gain, so the upside-to-downside ratio is not meaningful, and the probability of loss inside our model is 100%, which overstates our certainty because the model itself may be wrong. We do not recommend short positions, because the Street is positive (67 of 71 analysts rate the shares Buy) and a short carries unlimited loss. The structure supports a position of zero for new money, and holders could reduce into strength.

9.4 Monitoring list

• Upside invalidation: AWS growth above 40% in both Q4 2026 and Q1 2027 with a 2027 capital spending guide at or below $250 billion and AWS operating margin above 38%. That would indicate returns above our base scenario.
• Downside invalidation: AWS growth below 33%, or a 2027 capital spending guide above $300 billion without backlog growth, either of which would shift weight toward our bear scenario of $81.
• Trailing free cash flow, with a flag below −$70 billion.
• Long-term debt face value, with a flag above $200 billion.
• Receivables days (DSO), with a flag above 50 days.

9.5 Confidence

Appendix: supporting exhibits

Method and data notes

Industry appendix: internet-platform, saas. The report date is 10 November 2026 and the price is $266.93 (9 November 2026 close, Yahoo Finance), confirmed against the exchange close. Financial data come from SEC filings (Form 10-K for fiscal 2025 and Form 10-Q for Q3 2026) and Amazon's Q3 2026 release; no peer data, consensus for Q4, or short interest were obtained. The explicit forecast covers fiscal 2027–31, and the stub of fiscal 2026 is excluded. Shares are the diluted weighted-average count for Q3 2026 (10.929 billion), held flat because Amazon has no repurchases and treats stock compensation as a cost in our free cash flow. Net debt is face value of long-term debt ($180.9 billion) plus finance leases ($14.5 billion) less cash and marketable securities ($114.2 billion), and we count the Anthropic notes ($146.2 billion), private equity investments ($143.7 billion) and warrants ($4.3 billion) at half of their combined carrying value of $294.2 billion, which is $147.1 billion. We did not adjust for up to $20 billion of further Anthropic investment on milestones. The cost of equity is 4.88% (10-year Treasury, 9 November 2026) plus beta 1.30 (two-year weekly regression, raw 1.46, Blume-adjusted) times an equity risk premium of 4.5%, which is 10.73%. Net debt is below 10% of market value, so WACC (weighted average cost of capital) equals the cost of equity and rounds to 10.5%. Terminal growth is 3.0%. Non-US revenue is 32% of the total, below the 50% threshold for a country risk premium.

Inputs: price $266.93 (9 November 2026 close, Yahoo Finance), risk-free 4.88%, beta 1.30, equity risk premium 4.5%, WACC 10.5%, terminal growth 3.0%.

Sources

Amazon.com Form 10-K for fiscal 2025 and Form 10-Q for Q3 2026 (SEC EDGAR); Amazon Q3 2026 release (aboutamazon.com, 29 October 2026); Robinhood fundamentals, earnings, analyst ratings and SEC filing data (9 and 7 November 2026); US Department of the Treasury daily par yield curve (9 November 2026); ABiS event study on Amazon.com (10 November 2026); Motley Fool, 5 November 2026; aiweekly.co on Alphabet Q3 2026; Investing.com on the July 2026 bond sale; MLex on the FTC trial; Reuters via The Star on Anthropic; Yahoo Finance on OpenAI; LetsDataScience on Moody's; Fortune on Microsoft; Quilter on AWS, October 2025; Fintool on fiscal 2025; Form 144 coverage of the Bezos plan; Mauboussin and McKinsey long-term studies for base rates (approximate).