Amazon.com, Inc. (AMZN), valuation method revision

Revision date: 27 August 2026. Frozen reference price: $258.63 at the 21 August 2026 close. Frozen research cutoff and filing set: 23 August 2026. The immutable delivered package is dated 23 August, despite the supplied path ending 21 August. No later evidence is introduced.

Answer first

Segment unlevered FCF SOTP replaces the prior multiples-led target reconciliation. Revised Bear, Base and Bull values are $93.797301, $256.411399 and $530.309014. Weighted value is $292.362983. Expected return is 13.04%. Conviction moves from 7.92 to 7.55. The decision remains FULL LONG under the formal sizing map, but the margin of safety is thinner after explicit capex, leases, central costs, and dilution.

Company specific valuation method

Segment unlevered free cash flow SOTP is primary. The method was selected before targets. It values AWS, residual North America retail, residual International retail, advertising, subscriptions, and other activities with one ten year DCF architecture. Residual retail excludes advertising and subscriptions. Central costs are deducted once. Net cash, operating and finance leases, and diluted shares complete the equity bridge.

Why this method fits

Amazon combines businesses with different margins, capital intensity, maturity, and disclosure. AWS is reported separately. Advertising and subscriptions are revenue groups embedded in geographic segments, so a consolidated multiple hides their economics. The SOTP makes utilization, fulfillment density, capex, working capital, tax, SBC dilution, leases, central costs, and duration explicit. Stock compensation remains an operating cost and projected dilution is separately captured in shares, avoiding an addback.

Why alternatives can mislead

A consolidated enterprise unlevered FCF DCF is a valid cross check and reconciles to the sum of the same segment cash flows. It is less diagnostic. Forward P/E can capitalize nonoperating investment marks and ignores cash construction timing. EV to sales ignores the gap between first-party gross sales and fee revenue. A standalone ad or Prime multiple added to total retail would double count. Reverse DCF diagnoses expectations but does not produce intrinsic value.

Mandatory current to forecast calibration

Base paths start with frozen Q2 2026 or FY2025 evidence. Growth, margin, reinvestment, working capital, tax, and dilution move together. Bull AWS growth requires higher early capex and working capital. Bear growth loses utilization and retail operating leverage. Advertising and subscriptions are subtracted from residual retail starts before being valued separately. Exact annual segment paths are in the workbook and assumptions manifest.

Bear, base, and bull construction

For each segment, revenue year t equals prior revenue times one plus interpolated growth. UFCF equals segment revenue times the interpolated after-tax unlevered FCF margin. Segment value equals the present value of ten annual UFCFs plus terminal UFCF times one plus terminal growth divided by WACC minus terminal growth. Equity value equals segment values less central-cost DCF plus net cash less lease liabilities. Per-share value equals equity value divided by diluted shares.

Bear transmits capacity overbuild into slower AWS growth, weaker utilization, lower margins, continued capex, working-capital use, and dilution. Base assumes backlog and capacity convert gradually while retail density and advertising mix improve. Bull requires faster AWS conversion, stronger ad economics, international scale, and buybacks only after cash conversion.

Cross check and reconciliation

The consolidated enterprise UFCF DCF is the exact aggregation of the segment cash flows and central costs, so it reconciles without averaging incompatible methods. The frozen workbook reverse model implied 7.857% owner-earnings growth for ten years. The revised Base makes the cash pathway explicit and deducts leases. The frozen 24.89 times forward P/E remains a loose secondary check only.

Probability weighted value and expected return

20% times $93.797301, plus 55% times $256.411399, plus 25% times $530.309014, equals $292.362983. Expected return from $258.63 is 13.04%. Annualized over three years it is 4.17%. No dividend or separate buyback credit is added.

Business Quality decomposition

Moat durability: $1 billion cannot recreate Amazon's buyer and seller network, Prime behavior, fulfillment density, purchase-intent data, or AWS platform within five years. The moat does not guarantee pricing discretion or capital returns.

Five pillar weighted decomposition

Decision card: prior 7.92, revised 7.55, delta -0.37. FULL LONG and full formal sizing remain because 7.55 stays in the 7.5 to 8.5 band. Key risk is that $220 billion of annualized investment does not earn durable per-share cash returns. R3: frozen operating-pillar average delta is 0.0; all-pillar average delta is -0.42. Null hypothesis: weaker conversion, margins, or cash returns would lower the score.

What this analysis does not assume

It does not use evidence after the frozen cutoff, count advertising or Prime twice, treat backlog as revenue, add back SBC, ignore lease liabilities, hold reinvestment or shares constant across cases, or create scenarios using multiples alone.

Amazon.com, Inc., deep teaching valuation revision

This additive report preserves the original teaching report and frozen operating evidence. Revised values below are current. Old values appear only in explicitly labeled historical comparisons.

Amazon.com, Inc. (AMZN), valuation method revision

Revision date: 27 August 2026. Frozen reference price: $258.63 at the 21 August 2026 close. Frozen research cutoff and filing set: 23 August 2026. The immutable delivered package is dated 23 August, despite the supplied path ending 21 August. No later evidence is introduced.

Answer first

Segment unlevered FCF SOTP replaces the prior multiples-led target reconciliation. Revised Bear, Base and Bull values are $93.797301, $256.411399 and $530.309014. Weighted value is $292.362983. Expected return is 13.04%. Conviction moves from 7.92 to 7.55. The decision remains FULL LONG under the formal sizing map, but the margin of safety is thinner after explicit capex, leases, central costs, and dilution.

Company specific valuation method

Segment unlevered free cash flow SOTP is primary. The method was selected before targets. It values AWS, residual North America retail, residual International retail, advertising, subscriptions, and other activities with one ten year DCF architecture. Residual retail excludes advertising and subscriptions. Central costs are deducted once. Net cash, operating and finance leases, and diluted shares complete the equity bridge.

Why this method fits

Amazon combines businesses with different margins, capital intensity, maturity, and disclosure. AWS is reported separately. Advertising and subscriptions are revenue groups embedded in geographic segments, so a consolidated multiple hides their economics. The SOTP makes utilization, fulfillment density, capex, working capital, tax, SBC dilution, leases, central costs, and duration explicit. Stock compensation remains an operating cost and projected dilution is separately captured in shares, avoiding an addback.

Why alternatives can mislead

A consolidated enterprise unlevered FCF DCF is a valid cross check and reconciles to the sum of the same segment cash flows. It is less diagnostic. Forward P/E can capitalize nonoperating investment marks and ignores cash construction timing. EV to sales ignores the gap between first-party gross sales and fee revenue. A standalone ad or Prime multiple added to total retail would double count. Reverse DCF diagnoses expectations but does not produce intrinsic value.

Mandatory current to forecast calibration

Base paths start with frozen Q2 2026 or FY2025 evidence. Growth, margin, reinvestment, working capital, tax, and dilution move together. Bull AWS growth requires higher early capex and working capital. Bear growth loses utilization and retail operating leverage. Advertising and subscriptions are subtracted from residual retail starts before being valued separately. Exact annual segment paths are in the workbook and assumptions manifest.

Bear, base, and bull construction

For each segment, revenue year t equals prior revenue times one plus interpolated growth. UFCF equals segment revenue times the interpolated after-tax unlevered FCF margin. Segment value equals the present value of ten annual UFCFs plus terminal UFCF times one plus terminal growth divided by WACC minus terminal growth. Equity value equals segment values less central-cost DCF plus net cash less lease liabilities. Per-share value equals equity value divided by diluted shares.

Bear transmits capacity overbuild into slower AWS growth, weaker utilization, lower margins, continued capex, working-capital use, and dilution. Base assumes backlog and capacity convert gradually while retail density and advertising mix improve. Bull requires faster AWS conversion, stronger ad economics, international scale, and buybacks only after cash conversion.

Cross check and reconciliation

The consolidated enterprise UFCF DCF is the exact aggregation of the segment cash flows and central costs, so it reconciles without averaging incompatible methods. The frozen workbook reverse model implied 7.857% owner-earnings growth for ten years. The revised Base makes the cash pathway explicit and deducts leases. The frozen 24.89 times forward P/E remains a loose secondary check only.

Probability weighted value and expected return

20% times $93.797301, plus 55% times $256.411399, plus 25% times $530.309014, equals $292.362983. Expected return from $258.63 is 13.04%. Annualized over three years it is 4.17%. No dividend or separate buyback credit is added.

Business Quality decomposition

Moat durability: $1 billion cannot recreate Amazon's buyer and seller network, Prime behavior, fulfillment density, purchase-intent data, or AWS platform within five years. The moat does not guarantee pricing discretion or capital returns.

Five pillar weighted decomposition

Decision card: prior 7.92, revised 7.55, delta -0.37. FULL LONG and full formal sizing remain because 7.55 stays in the 7.5 to 8.5 band. Key risk is that $220 billion of annualized investment does not earn durable per-share cash returns. R3: frozen operating-pillar average delta is 0.0; all-pillar average delta is -0.42. Null hypothesis: weaker conversion, margins, or cash returns would lower the score.

What this analysis does not assume

It does not use evidence after the frozen cutoff, count advertising or Prime twice, treat backlog as revenue, add back SBC, ignore lease liabilities, hold reinvestment or shares constant across cases, or create scenarios using multiples alone.

Frozen company and segment teaching, unaffected by methodology

Amazon sits across commerce, fulfillment, advertising, and computing

A customer sees a storefront or a cloud console. The economic system underneath is longer.

The commerce value chain

1. Brands, vendors, and third party merchants create selection.
2. Amazon attracts shoppers through price, convenience, search, content, and Prime benefits.
3. Amazon either owns inventory and records the full product sale, or hosts a third party seller and records commissions plus related service fees.
4. Fulfillment centers receive, store, pick, and pack goods.
5. Sortation centers and delivery stations organize the last mile.
6. Carriers, contractors, aircraft, vehicles, and local delivery routes move the parcel.
7. Amazon collects payment, handles customer service, and may process returns.
8. Sellers can buy sponsored placement at the moment a shopper has purchase intent.
9. Prime subscription revenue helps fund a bundle that increases repeat usage and makes the network denser.

The customer pays Amazon for products or Prime. Sellers pay commissions, fulfillment fees, shipping fees, and advertising charges. Vendors may sell inventory to Amazon. Advertisers pay for access to audience and purchase intent. The same visit can therefore support several revenue streams.

The cloud value chain

1. Semiconductor designers and manufacturers provide processors, memory, and networking components. Amazon also designs custom chips.
2. Data center developers, utilities, and equipment suppliers provide buildings, power, cooling, and electrical systems.
3. AWS assembles servers, storage, databases, networking, security, artificial intelligence infrastructure, and managed software services.
4. Enterprises, developers, governments, and startups consume those services.
5. Customers generally pay for usage, reserved capacity, support, or contractual commitments.
6. AWS turns installed capacity into revenue only when the capacity is available, useful, and consumed.

AWS sits between physical infrastructure and customer software. It absorbs large upfront investment so customers can buy flexible computing capacity. The economic promise is high utilization across shared infrastructure. The danger is building expensive capacity before demand arrives, or replacing chips and servers before the original investment earns an adequate return.

Where Amazon captures value

Amazon captures value at four bottlenecks.

• It controls a large shopping destination with immediate purchase intent.
• It coordinates sellers, inventory, fulfillment, delivery, payments, and customer trust.
• It bundles repeated consumer behavior through Prime.
• It operates a broad cloud platform with deep services and large installed customer relationships.

The company does not report advertising and subscriptions as operating segments. They are product and service revenue groups embedded mainly inside the geographic stores segments. That distinction matters. Advertising can improve North America and International segment profit without appearing as a separate segment profit line.

2. The economic engine, worked from operating activity to shareholder value

AWS worked example using reported Q2 2026 results

AWS reported $42.232 billion of Q2 2026 sales and $16.621 billion of segment operating income.[2]

The reported period economics were:

• Revenue: $42.232 billion.
• Segment operating expense: $25.611 billion.
• Segment operating income: $42.232 billion minus $25.611 billion, which equals $16.621 billion.
• Segment operating margin: $16.621 billion divided by $42.232 billion, which equals 39.36 percent.

Expressed per $100 of reported AWS revenue, the segment retained about $39.36 as operating income before corporate nonoperating items and tax. This is an average accounting relationship, not the margin on one new artificial intelligence workload.

The cash bridge is less direct:

1. Amazon spends cash to secure land, power, buildings, servers, chips, and networking.
2. Property and equipment enter service and begin depreciation over accounting useful lives.
3. Customer usage and contracts generate AWS revenue.
4. Revenue less operating costs and depreciation produces segment operating income.
5. Consolidated operating income is adjusted for interest, investment gains or losses, tax, and other items to reach net income.
6. Operating cash flow adds back noncash charges and reflects working capital.
7. Cash capital expenditures reduce free cash flow.
8. Only cash remaining after sustaining and growth investment can increase net cash, fund acquisitions, repurchase shares, or otherwise accrue to shareholders.

This explains the central tension. AWS operating income can rise while consolidated free cash flow falls because cash construction precedes revenue and depreciation.

Consolidated evidence shows the timing mismatch

Amazon revenue increased from $514.0 billion in 2022 to $716.9 billion in 2025. That is an 11.7 percent annual compound rate. Operating margin increased from 2.4 percent to 11.2 percent. Yet free cash flow moved from negative $16.9 billion in 2022 to $32.2 billion in 2023, $32.9 billion in 2024, and only $7.7 billion in 2025.[3]

The fall in 2025 free cash flow did not come from falling operating income. Operating income rose to $80.0 billion. Cash capital expenditure rose sharply. The annual filing reports $128.3 billion of cash capital expenditures in 2025, versus $77.7 billion in 2024, primarily for technology infrastructure supporting AWS and additional fulfillment capacity.[1]

In the first half of 2026, cash capital expenditures reached $96.3 billion, compared with $55.6 billion a year earlier.[2] This is why accounting margin expansion and shareholder cash conversion must be evaluated separately.

Retail worked example using reported segment averages

North America reported Q2 2026 sales of $116.177 billion and operating income of $9.123 billion.[2]

• Segment operating margin: $9.123 billion divided by $116.177 billion, which equals 7.85 percent.
• Per $100 of reported sales, about $7.85 remained as segment operating income.

That average blends different economics. First party product revenue is recorded gross and includes merchandise cost. Third party commissions and advertising are service revenue. A shift toward third party services and advertising can raise profit faster than gross merchandise volume because Amazon records less pass through merchandise cost and earns fees on the ecosystem.

A parcel also becomes cheaper to serve when more orders share nearby inventory, fulfillment buildings, line haul, delivery stations, and routes. This is density economics. The benefit can be retained as margin, reinvested in faster delivery, or passed to customers through lower prices.

3. Legal structure and operating structure are not the same thing

Amazon.com, Inc. is the public parent. Its consolidated financial statements include wholly owned subsidiaries and entities where Amazon has a variable interest and is the primary beneficiary. The filing specifically describes certain consolidated variable interest entities in India and entities supporting healthcare services and video content production and distribution.[2]

Intercompany balances and transactions are eliminated. This means consolidated revenue is not the sum of every legal entity invoice.

The three reportable segments

Amazon reports three operating segments:

• North America.
• International.
• AWS.

The segment measure is operating income. Costs associated with functions such as fulfillment and technology can be allocated based on usage. Corporate financing, investment gains and losses, income tax, and equity method activity sit below consolidated operating income rather than inside a clean fourth operating segment.

Revenue groups cut across the segment map

Amazon also discloses product and service groups. In 2025 these included online stores, physical stores, third party seller services, advertising services, subscription services, AWS, and other revenue.[1] These are not seven reportable segments.

The distinction prevents three common errors.

1. Advertising is strategically important, but Amazon does not publish its standalone operating profit.
2. Prime subscription revenue is visible, but Prime also changes shopping frequency and fulfillment costs across the stores system.
3. AWS is both a revenue group and a reportable segment. Advertising is a revenue group embedded in geographic segments.

Pass through revenue and gross versus net presentation

When Amazon owns inventory, it generally records product sales gross. When a third party seller is the merchant, Amazon records its net share of commissions and related services as service sales.[1] Two dollars of reported revenue can therefore represent very different gross merchandise value and profit economics.

This makes consolidated gross margin useful but incomplete. Segment operating income and cash conversion reveal more about economic quality.

Minority interests and parent costs

The preserved source package does not identify a material public minority interest that requires a separate deduction from the scenario targets. Consolidated variable interest entities are included when Amazon is the primary beneficiary. This report therefore makes no new minority interest adjustment.

Parent level financing and nonoperating investments still matter. Q2 2026 reported net income included a very large nonoperating gain. Operating income was $27.461 billion while total nonoperating income was $53.396 billion.[2] The preserved management critique attributes much of the distortion risk to artificial intelligence related investment accounting. Valuation should not capitalize such marks as recurring operating earnings.

Capital consuming activities

AWS infrastructure is the largest named capital consumer. The fulfillment network also consumes buildings, automation, aircraft, vehicles, equipment, and lease commitments. Satellite broadband, content, devices, healthcare, and artificial intelligence investments can consume cash before commercial evidence becomes visible. The segment presentation does not isolate every one of these activities.

4. AWS is the reported profit engine and the largest underwriting question

AWS sells computing, storage, databases, networking, security, analytics, and artificial intelligence infrastructure and services. Customers avoid building every layer themselves. AWS pools demand across many users and attempts to keep costly infrastructure highly utilized.

Q2 2026 AWS sales were $42.232 billion, up 37 percent year over year. AWS operating income was $16.621 billion, up from $10.160 billion.[2] The preserved assessment describes a $169 billion annualized revenue run rate and $496 billion backlog based on management's earnings call discussion.[4]

What makes AWS attractive

• Scale spreads engineering, procurement, network, and control plane costs.
• Broad services raise switching costs because applications depend on databases, security, data tools, and operating processes, not only raw servers.
• Custom silicon can lower workload cost or improve performance if customers adopt it.
• Large customer relationships and partner expertise make AWS a default choice for many workloads.
• Demand growth improves utilization when capacity arrives in the right location and configuration.

What can weaken AWS economics

• Capacity can arrive late because power, chips, networking, or construction are constrained.
• Capacity can arrive early and remain underutilized.
• Competitors can reduce price or bundle cloud with software.
• Custom chips require continuous design success and software support.
• Faster hardware cycles can shorten useful economic lives.
• Depreciation can rise after cash spending has already occurred.
• Large customers can negotiate price reductions through long term contracts.

The annual filing says 2025 AWS sales growth was partially offset by pricing changes primarily driven by long term customer contracts.[1] The moat therefore does not create unlimited pricing discretion.

KPI hierarchy

The most useful sequence is:

1. AWS revenue growth.
2. AWS operating margin.
3. Backlog conversion and capacity availability.
4. Cash capital expenditure.
5. Depreciation growth.
6. Consolidated free cash flow after capacity enters service.

A backlog headline without delivery timing and cancellation protection is weaker evidence than revenue from operating capacity.

5. North America converts network density into several revenue streams

North America combines first party retail, third party marketplace services, fulfillment, advertising, subscriptions, devices, content, and related activities.

In 2025 it produced $426.305 billion of sales and $29.619 billion of segment operating income, a 6.95 percent margin. In Q2 2026 it produced $116.177 billion of sales and $9.123 billion of segment operating income, a 7.85 percent margin.[1][2]

The flywheel in economic terms

More selection can attract more shoppers. More shoppers attract sellers. More sellers increase selection and advertising demand. More orders improve network density. Better density can reduce delivery cost or fund faster delivery. Faster delivery can increase shopping frequency and Prime value.

This is not automatic. Every additional order carries picking, packaging, transport, customer service, returns, payment, and inventory risk. Margin expands only if service and monetization improve faster than those costs.

Why third party services matter

Third party sellers expand selection without requiring Amazon to own every item. Amazon can earn commission, fulfillment, storage, shipping, and advertising revenue. The seller retains inventory and product risk unless Amazon provides related services.

The model creates network effects, but also regulatory and seller tension. Fees that maximize short term profit can encourage sellers to diversify to other channels. Enforcement mistakes can damage trust. Regulators can challenge marketplace conduct and self preference.

Why advertising matters

Advertising services generated $68.635 billion of revenue in 2025 and $19.809 billion in Q2 2026.[1][2] A sponsored result appears near a measurable purchase decision. That intent is valuable to sellers and brands.

Amazon does not report advertising operating income separately. Calling it a high margin engine is economically plausible, but a precise standalone margin would be invented. The evidence is that North America and International operating income improved alongside advertising growth. The allocation of shared traffic, technology, and fulfillment costs remains undisclosed.

6. International has improving scale with wider regulatory and competitive outcomes

International includes established and earlier stage markets with different consumer income, delivery infrastructure, regulation, and competition.

In 2025 it produced $161.894 billion of sales and $4.750 billion of operating income, a 2.93 percent margin. Q2 2026 sales were $42.197 billion and operating income was $1.717 billion, a 4.07 percent margin.[1][2]

The improving margin is evidence of scale and monetization. It remains well below North America.

Why outcomes differ by country

• Delivery density varies.
• Local competitors can have stronger brands and seller relationships.
• Cross border trade, taxes, labor rules, and product regulation differ.
• Currency can move reported revenue and income.
• Foreign ownership and marketplace rules can restrict operating structure.

The annual and quarterly filings describe local ownership, licensing, and cybersecurity constraints in China and marketplace ownership restrictions in India.[1][2] The legal structure can therefore be less direct than the customer interface suggests.

What must improve

International value creation requires sustained positive operating income, better delivery density, and advertising or seller service growth without regulatory damage. Revenue growth alone is insufficient.

7. Advertising and Prime deepen the ecosystem but require analytical restraint

Advertising and subscriptions reinforce the segments, but neither is a separately reported operating segment.

Advertising services rose from $46.906 billion in 2023 to $68.635 billion in 2025. Subscription services rose from $40.209 billion to $49.619 billion.[1]

Prime can increase order frequency and retention. It also bundles shipping, video, music, and other benefits with different cost structures. Subscription revenue cannot be treated as pure profit.

Advertising benefits from purchase intent, but excessive ad load can reduce customer trust and organic relevance. Sellers ultimately fund advertising from their own economics. Amazon must preserve enough seller return to sustain marketplace breadth.

The evidence supports strong monetization. It does not support a precise standalone margin or valuation for these activities. This report gives them credit through the preserved segment results and scenario targets, not through an additional sum of parts premium.

8. The moat is strong, but pricing power differs by activity

Franchise protection

Amazon's consolidated moat has four layers.

1. Scale economics. Fulfillment density, procurement, technology, and global infrastructure are difficult to reproduce.
2. Marketplace network effects. Buyers attract sellers. Sellers expand selection and advertising demand.
3. Behavioral switching costs. Prime, saved payment details, purchase history, trust, and habitual convenience support repeat use.
4. AWS ecosystem switching costs. Applications, data, security controls, developer skills, partners, and managed services make migration costly.

Pricing power

Pricing power is uneven.

• Retail consumers can compare prices instantly. Retail pricing power is limited.
• Sellers depend on Amazon traffic, but can use competing marketplaces or direct channels. Fee discretion exists, but overuse can weaken the ecosystem.
• Advertising pricing reflects auction demand and measurable conversion. This is stronger, but seller returns impose a ceiling.
• AWS customers face switching costs, yet large contracts can include price concessions. Competition from Microsoft, Google, and customer owned infrastructure limits discretion.

The moat protects volume, relevance, and utilization more clearly than it guarantees price increases.

Who captures the moat value

Shareholders capture value through higher utilization, fee revenue, advertising, and AWS operating income. Customers capture value through low prices, fast delivery, selection, and lower upfront technology investment. Sellers capture demand and fulfillment access. Employees, suppliers, utilities, chip vendors, landlords, and delivery partners also claim part of the economics.

Amazon has historically reinvested much of the moat value. The current capital cycle again shifts value from near term free cash flow toward future capacity.

What the moat does not protect

It does not protect Amazon from poor capital allocation, regulatory remedies, power shortages, obsolete chips, cybersecurity incidents, labor inflation, seller dissatisfaction, or a cloud price war. It does not ensure that a $220 billion capital program earns above the cost of capital.

New entrant test

Could a new entrant with $1 billion take meaningful consolidated share within five years? No. That amount cannot replicate the fulfillment footprint, shopper and seller network, Prime behavior, global cloud regions, software services, security certifications, and engineering ecosystem.

A well funded specialist can still take share in one layer, such as artificial intelligence models, advertising tools, commerce niches, delivery, or a cloud software category. The moat is broad, not absolute.

9. Evidence states prevent pipeline from becoming imaginary value

Backlog should be valued through demonstrated conversion, not multiplied by a headline margin. Capital expenditure should be evaluated after capacity becomes available and free cash flow has time to respond.