Snapshot: Executive Summary
Amazon is best understood not as a retailer but as a sum of three businesses with radically different economics: AWS (a cloud profit engine), a fast-growing advertising annuity, and a thin-margin retail base. In Q2 2026 the company printed $200.6B of revenue (+20%, its first $200B+ quarter) at a record 13.7% operating margin, with $27.5B of operating income (+43% YoY)1. The headline is the profit concentration: AWS earned $16.6B (60% of operating income) on ~21% of revenue2, and it reaccelerated to +37% YoY: its fastest in 18 quarters, on a ~$169B run-rate with a ~$364B backlog.
Three value pools, three valuation regimes:
- AWS (~$169B run-rate → ~$200B FY27E): the crown jewel: ~39% segment operating margin, the majority of company profit, and the AI flywheel (Trainium/Bedrock, a ~$100B Anthropic commitment). The adjustable premium multiple in the SOTP, and 58% of enterprise value.
- Advertising (~$75B+ TTM, +26%): a high-margin (~80% incremental) annuity carved out of the consumer segments. A premium-multiple asset the market under-prices because it is not reported standalone.
- Retail, North America + International (~$575B FY27E): the lower-margin base (NA ~8% / Intl ~3.5% operating margin). Profitable, improving, but a market-multiple ballast, not a re-rater.
At ~27x forward earnings a blended multiple obscures all of this. Our sum-of-the-parts values AWS at a premium cloud multiple, the advertising annuity at a high multiple, and retail at a market multiple, less ~$92B of net debt, a base fair value of ~$310 vs the $272 spot, ~14% upside. We maintain Buy, 12-month PT $310 (bull $360 / bear $235), in line with the Street's ~$318 consensus after the Q2 AWS reacceleration. The one thing that can break the call is the free-cash-flow drought: trailing FCF turned slightly negative to ~($7.6B) under a raised ~$220B capex program, 2026 is a "show-me" year for the AI payback.
Tactical: AMZN is trading at $272, ~14% below our $310 12-month SOTP target, upside to PT remains, anchored on the sum-of-the-parts (AWS at a premium cloud multiple + the advertising annuity + a market-multiple retail base, less net debt). Rating Buy; our $310 target is in line with the Street's ~$318 consensus, and the swing variable is the free-cash-flow trajectory under record AI capex.
Investment Thesis
Bull Case
- AWS re-rates as AI monetization shows: FY27E ~$225B at 21x EV/EBITDA on the Trainium/Bedrock/Anthropic flywheel
- Advertising compounds past $100B at premium multiples; retail margins keep expanding
- Free cash flow inflects in 2027 as operating cash flow outgrows a moderating capex base
- Net debt eases to ~$75B as the cash machine re-asserts itself
- Consolidated operating margin pushes toward the mid-to-high teens
Base Case
- AWS sustains mid-to-high-20s% growth to ~$215B FY27E at ~48% EBITDA margin, 20x EV/EBITDA
- Advertising ~$100B at 18x; Retail ~$575B at 8% EBITDA margin, 12x; ~$92B net debt
- FCF stays negative through 2026 under ~$220B capex, then begins to recover in 2027
- AWS = 58% of enterprise value; the parts still sum above the ~$3T whole
Bear Case
- The ~$220B capex never earns its return; the FCF deficit (~($7.6B) TTM) deepens and a depreciation wave hits margins
- AWS keeps ceding AI share at the margin; even after reaccelerating to +37% it still edges below Azure (+40%) and well below Google Cloud (+63%)
- AWS de-rates to 17x on ~$205B; ad and retail multiples compress; net debt climbs to ~$110B
- Regulatory drag (FTC trial currently scheduled ~Mar 2027) and tariff/consumer weakness pressure retail
Rating: Buy. The probability-weighted SOTP (45% base / 30% bull / 25% bear) blends to ~$306, above the $272 spot. We rate AMZN a Buy on ~14% base upside to a $310 SOTP target, the AWS reacceleration, and the advertising/FCF-inflection optionality. Our $310 base is now in line with the Street's ~$318 Strong-Buy consensus after we raised the AWS multiple on the Q2 reacceleration. Downgrade trigger: the free-cash-flow drought extending into 2027 with no visible AI payback, or AWS growth slipping below ~20%.
Business Overview
Amazon reports three segments, AWS, North America, and International, but the more useful frame for valuation collapses the business into three value pools, because the cloud profit engine, the advertising annuity, and the retail base earn radically different margins and deserve radically different multiples. The single most important fact: AWS earns the majority of operating income on a minority of revenue.
Revenue by segment (Q2 2026)
North America is the bulk of revenue; AWS is only ~21%. Advertising (~$19.8B/qtr) sits inside the NA + International consumer segments, not as a standalone line.
Operating income by segment (Q2 2026)
The mirror image: AWS is 60% of operating income ($16.6B of $27.5B). This profit concentration is why the SOTP values AWS at 58% of enterprise value.
Segment commentary
- AWS: infrastructure cloud (compute, storage, database) plus the AI stack (Bedrock model platform, SageMaker, and custom silicon Trainium/Inferentia/Graviton). ~$169B run-rate, ~39% segment operating margin, and the company's profit engine. The crown jewel.
- Advertising: sponsored products/brands on the retail surfaces, plus Prime Video ads and DSP. A ~$75B+ TTM business growing ~26% as reported at very high incremental margins. Reported inside the consumer segments, which is why it is easy to under-value.
- North America: the domestic e-commerce + physical stores + subscription (Prime) engine. ~$114B/qtr at an ~8% operating margin that has expanded sharply as the fulfilment network was re-architected regionally.
- International: e-commerce outside the US. ~$44B/qtr at a thinner ~3.5% margin; structurally improving as newer geographies mature.
AWS: The Profit Engine
The single most important section in this report. AWS is the business that makes Amazon a high-quality compounder rather than a thin-margin retailer. In Q2 2026 it grew +37% YoY: the fastest in 18 quarters, to a ~$169B run-rate, earning $16.6B of segment operating income at a 39.4% margin2. The forward signal is the backlog: ~$364B, underwritten by AI demand and a multi-year, up-to-5-GW / ~$100B compute commitment from Anthropic3.
AWS revenue trajectory ($B, annual)
FY25 ~$128B → FY27E ~$200B at sustained mid-to-high-20s% growth, the revenue engine of the SOTP. The ~$364B backlog gives the FY26–FY27 ramp a multi-year demand anchor.
Hyperscaler cloud growth: latest quarter (YoY %)
The honest read: AWS (+37%) has reaccelerated to its fastest in 18 quarters, narrowing the gap to Azure (+40%), though Google Cloud (+63%) still grows faster off a smaller base. AI share is the central bear question.
The AI flywheel
AWS's AI business is now a >$15B run-rate growing triple digits, Andy Jassy framed it as roughly 260× AWS's own $58M run-rate three years post-launch. Bedrock, the model platform, has processed more tokens over the past year than in all prior years combined, with customer spend up sharply and ~80% of the Fortune 100 now using it.
The bull-and-bear tension sits right here: that momentum is real, but it leans on the Anthropic relationship, which the bears flag as circular, Amazon is a major investor (the stake, carried ~$74.2B at Q1, was written up a further ~$53.4B in Q2 as Anthropic's mark re-rated toward its $965B Series H; up to ~$33B committed) and Anthropic is a major AWS/Trainium customer, so Amazon part-funds a customer that buys its cloud. (Anthropic raised a $65B Series H at a $965B post-money valuation in May 2026 and confidentially filed for an IPO in June 2026; that re-rating drove the ~$53.4B non-cash gain in Q2's net income.)
On differentiation, AWS leans on custom silicon: Trainium (training) and Inferentia/Graviton (inference/general compute) are a >$20B run-rate, Trainium3 is nearly fully subscribed, and a consistent majority of Bedrock workloads already run on Trainium; in April 2026 Meta signed a multi-year deal (billions over 3+ years) to deploy tens of millions of AWS Graviton5 cores, a notable validation given Meta is also a large Google Cloud customer. The bull case is that this cost-per-token advantage defends AWS's share even as Azure and GCP grow faster; the bear case is that the circularity and AWS's slower relative growth leave the premium multiple fragile.
Why AWS dominates the SOTP
AWS is ~21% of revenue but ~60% of operating income and, in our sum-of-the-parts, ~58% of enterprise value. At a ~48% EBITDA margin (39.4% segment operating margin plus the heavy data-center depreciation that runs through cost of revenue) and a 20x EV/EBITDA, a premium toward Alphabet's faster-growing cloud and above Microsoft (~16.5x), the AWS bucket is worth ~$2.1T on its own. The company's total enterprise value is ~$3.0T (the ~$2.92T equity cap plus ~$92B net debt), so with AWS ~$2.1T the market implies under ~$0.95T for the entire retail + advertising business, below our ~$1.36T sum-of-the-parts for those pieces, the parts are again worth more than the whole after we raised the AWS multiple on the reacceleration.
Advertising & Retail: The Annuity and the Base
Outside AWS, Amazon runs a high-margin advertising annuity hidden inside a large, improving retail base. In the SOTP these map to the premium-recurring "Advertising" bucket and the market-multiple "Retail" bucket.
Advertising: the hidden annuity
- Sponsored ads + Prime Video + DSP (~$75B+ TTM, +26% as reported): sponsored products and brands on Amazon's retail surfaces, where purchase intent is highest, plus the newer Prime Video ad tier. Incremental margins are very high (~80%), because the audience and infrastructure are already paid for by retail.
- Because advertising is reported inside the consumer segments rather than broken out, the market tends to under-price it. On a standalone basis it would be one of the largest, fastest-growing, highest-margin ad businesses in the world.
Retail: North America + International
- North America (~$114B/qtr, ~8% operating margin): domestic e-commerce, physical stores, and Prime. North America operating income rose to ~$9.3B as the regionalized fulfilment network and automation drove unit-cost gains.
- International (~$44B/qtr, ~3.5% margin): e-commerce outside the US, structurally less profitable but improving as newer geographies scale.
The risk lens
The retail base is improving but cyclically and politically exposed, tariffs, a soft consumer, and the FTC trial (currently scheduled ~March 2027) all sit on this part of the business. In the SOTP, retail is the ballast: a market multiple on thin margins. The re-rating optionality lives in AWS and advertising, not here.
Financial Health & Trends
Revenue by quarter ($B)
Consolidated operating margin (%)
FY16–FY26E: the margin transformation
Amazon's decade is the story of operating margin going from low-single-digits to a record 13.7% in Q2 2026, as AWS scaled, advertising grew, and the retail network was re-architected. Operating income hit $27.5B in the quarter (+43% YoY)1. Amazon has also cut ~30,000 roles across multiple units since late 2025 (~14,000 corporate roles formally announced; framed as removing management layers and reinvesting in AI rather than cost-cutting), a modest support to the record margins. The tension is below the operating line: trailing free cash flow turned slightly negative to ~($7.6B), as a raised ~$220B of AI/AWS capex outran even the $148.5B (+30%) operating cash flow4.
FY16–FY26E: revenue & operating margin
Revenue compounds while operating margin steps up with AWS + advertising mix and retail efficiency. The margin line, not free cash flow, is the cleaner read on the business while the capex cycle runs hot.
Q2 2026 print highlights
| Metric | Q2 2026 | YoY | Driver | Note |
|---|---|---|---|---|
| Total revenue | $200.6B | +20% | AWS + advertising | first $200B+ qtr |
| Operating income | $27.5B | +43% | 13.7% margin | record |
| AWS revenue | $42.2B | +37% | AI / backlog | fastest in 18 qtrs |
| AWS operating income | $16.6B | 39.4% margin | 60% of total | reported |
| Advertising revenue | $19.8B | +26% | sponsored + video | reported |
| Operating cash flow (TTM) | $148.5B | +30% | profit growth | reported |
| Free cash flow (TTM) | ~($7.6B) | turned negative | ~$220B capex | the debate |
Note: free cash flow is Amazon-defined and has now turned negative outright; the figure swings by definition and is the central bear input. Q2'26 net income ($62.6B, GAAP diluted EPS $5.75) was flattered by a ~$53.4B non-cash mark-to-market gain on the Anthropic investment (its $965B Series H re-rated the stake); excluding it, underlying EPS was ~$1.90. That mark inflates net income but not operating income or FCF, so we treat operating income, not net income, as the cleaner profit read.
Capital Allocation & Returns
Amazon's capital story is, right now, almost entirely about capex. The company is spending an estimated ~$220B in 2026 (raised from ~$200B on higher memory costs): among the largest annual capital programs of any company, on AWS data centers, custom AI silicon, and the fulfilment network. There is no dividend and buybacks are effectively paused; every dollar of operating cash flow is being reinvested. The bet is that this capacity earns a high return as AI demand monetizes; the risk is that it does not, and the free-cash-flow drought persists.
Amazon capex ($B)4
Capex has roughly tripled into the AI cycle, scaling toward ~$220B in 2026 (raised on higher memory costs), far above the prior ~$146B consensus. It has now outrun the trailing operating cash-flow base, which is why free cash flow has turned slightly negative despite record operating income. Sell-side sees the buildout extending, not moderating, UBS pegs aggregated 4Q25–4Q27 Amazon capex at ~$344B (AWS ~$260B), though that note predates the recent prints and Amazon has issued no formal 2027 figure.
Why the FCF drought is the whole debate
A company that earns record operating income but generates almost no free cash flow forces a question: is the capex a temporary investment that will normalize into a cash gusher (the 2024-style FCF of ~$25-30B), or a permanently higher level of intensity that depresses returns? The bull says operating cash flow ($148.5B, +30%) still outgrows capex over time and FCF inflects in 2027; the bear says ~$220B is the new floor and a depreciation wave is coming. Sell-side models make the timing concrete: full-year 2026 free cash flow is now running negative, consensus around ($18B), with estimates ranging from roughly ($17B) to ($28B), as the ~$220B capex program outruns even a fast-growing operating cash-flow base, with the inflection back to materially positive FCF generally modeled in 2027 (analyst modeling only; Amazon has issued no FCF target). Our SOTP sidesteps the timing by valuing the segments' operating economics, but the multiple the market pays is hostage to the FCF trajectory.
Valuation Overview
AMZN trades at ~27x forward earnings. Read at face value that looks like a premium to the market, but the headline multiple blends a ~48%-EBITDA-margin cloud business, a ~80%-incremental-margin ad annuity, and a thin-margin retail base into one number. Applying a single P/E to that mix is the wrong frame; it systematically under-values the high-quality parts. The right frame is sum-of-the-parts, and we develop the full mechanics in the next section.
AMZN forward P/E history
AMZN's multiple has compressed from the >50x of its lower-margin era toward ~27x as the earnings base grew, the re-rating is fundamental (more profit), not just sentiment.
Peer NTM EV/EBITDA
The SOTP anchors AWS between Microsoft (~16.5x) and Alphabet's faster-growing cloud (~20-23x), the ad annuity near the ads end, and retail at roughly half Walmart's (~23-25x) multiple, reflecting Amazon's thinner blended retail margin and a conglomerate discount.
Why the SOTP
A blended-P/E argument always loses to a fundamentals-aware sum-of-the-parts when the business mix is this heterogeneous. We separate AMZN into three buckets, AWS, Advertising, and Retail, apply scenario-appropriate EV/EBITDA multiples to each FY27E revenue × EBITDA margin, sum the enterprise values, subtract net debt, and divide by shares. That work, with an interactive scenario tool, is next, and it reconciles to the $310 base PT.
SOTP: Sum-of-the-Parts
The SOTP is the heart of the report. Each bucket is valued on its FY27E revenue, EBITDA margin, and EV/EBITDA multiple. Toggle between Bull, Base, Bear, and Reverse scenario anchors; drag the AWS and Advertising multiples to see the implied PT update. Amazon carries net debt (cash less debt and lease obligations), so the model subtracts it from enterprise value.
| Bucket | Rev ($B) | EBITDA margin | Multiple | EV ($B) |
|---|---|---|---|---|
| AWS (FY27E) | 215 | 48% | 20x | — |
| Retail, NA + Intl (FY27E) | 575 | 8% | 12x | — |
| Advertising (FY27E) | 100 | 45% | 18x | — |
| Total enterprise value | — | |||
| Less: net debt ($B) | ($92) | |||
| Equity value ($B) | — | |||
| Implied per-share PT (÷ 10.75B sh) | — | |||
| vs current market | — |
Revenues and EBITDA margins update from the scenario anchors; the AWS and Advertising multiples are user-editable via the sliders. Net debt (positive) is subtracted from enterprise value. The "Reverse" scenario uses base revenues/margins, drag the multiples until the implied PT equals the current market price to see what the market is pricing. AWS bucket margin includes the heavy data-center D&A that runs through cost of revenue (~48% EBITDA vs 39.4% segment operating margin), implying ~$20B of AWS D&A (~10% of FY27E revenue); note that this EBITDA-margin choice partly offsets the bear case's depreciation-wave risk.
Sensitivity grid: AWS multiple × Advertising multiple (PT in $)
PT calculator (alternative simple-multiple view)
Risk / Reward calculator
Note: The assistant reasons from the dashboard's data snapshot and thesis sections, it does not browse the web or access real-time fundamentals beyond what's in data.js. Treat its responses as scenario-modeling support, not primary research. Author judgments on rating, PT, and probabilities remain with the analyst.
Upcoming Catalysts
| Catalyst | Window | Why it matters |
|---|---|---|
| Q3 2026 print | ~Oct 29, 2026 | Guided $197–202B sales (+9–12%) / $22.5–26.5B operating income. The Prime Day pull-forward into Q2 sets up a tougher Q3 comp ('air-pocket' risk). The key read on whether AWS sustains its +37% reacceleration and whether the raised ~$220B capex deepens the FCF deficit. |
| Anthropic S-1 / IPO | 2H 2026 | Anthropic confidentially filed a draft S-1 (June 2026) for a possible fall-2026 Nasdaq listing at/above its $965B valuation; the public filing would disclose Amazon's exact stake for the first time and could drive further non-cash marks in reported net income. |
| AWS growth trajectory | Each quarter | Whether the +37% reacceleration sustains vs Azure / Google Cloud, the central read on AI share and the SOTP's biggest bucket. |
| Trainium3 + Anthropic capacity | 2H 2026 | ~1 GW of Trainium2+3 coming online by year-end; the proof the AI capex is converting to revenue. |
| Capex figure updates | Each call | Any moderation (or further increase) of the raised ~$220B full-year capex drives the FCF-inflection debate. |
| Advertising milestones | Each quarter | Surpassing the >$75B TTM run-rate and any incremental-margin disclosure re-rates a hidden asset. |
| FCF inflection signs | 2H 2026 – 2027 | Evidence operating cash flow ($148.5B, +30%) is outgrowing a moderating capex base, the bull's core proof point. |
Risk Factors
- The free-cash-flow drought. The pre-eminent risk. ~$220B of 2026 capex pushed trailing FCF (Amazon-defined) slightly negative to ~($7.6B), consuming all of operating cash flow and then some, with unproven payback. If FCF fails to inflect and capex creeps higher, the market de-rates the whole company regardless of operating income.
- AWS AI share loss. Even after reaccelerating to +37% (its fastest in 18 quarters), AWS still grows a touch slower than Azure (+40%) and well behind Google Cloud (+63%), which are taking AI workloads faster off smaller bases. If that gap persists, the premium AWS multiple that anchors 58% of the SOTP compresses.
- A depreciation wave. The AI buildout creates a large, lengthening depreciation tail that will pressure future operating margins and earnings quality even after capex moderates.
- Regulatory overhang. The core FTC monopolization trial is currently scheduled for ~March 2027 (rescheduled from February), and a $2.5B dark-patterns settlement is already paying out. An adverse outcome could force structural or behavioral remedies on the retail/marketplace business. Two newer, earlier-stage overhangs widen the surface onto the two profit pillars: on June 25, 2026 the EU preliminarily viewed AWS as a Digital Markets Act cloud "gatekeeper" (the first-ever cloud designation, not yet binding; a final decision is expected ~Q4 2026 and would bring interoperability / anti-self-preferencing obligations on the profit engine), and the FTC was reported in June 2026 to be drafting (not yet filing) a separate complaint alleging Amazon misled advertisers on sponsored-listings pricing.
- Tariffs and a soft consumer. Retail unit economics are exposed to tariff costs and discretionary-spend weakness, though the regime has shifted: the Supreme Court struck down the IEEPA "reciprocal" tariffs in February 2026 (ending the peak 104–125% China rates), leaving a lighter ~15% Section 122 global duty (imposed at 10%, raised to the 15% statutory max in late Feb 2026) that was set to sunset ~July 24, 2026 and remains tangled in ongoing litigation. Net tariff drag is lower than in 2025, but the post-sunset legal flux is a live swing factor for the thin-margin base.
- Earnings quality + Anthropic circularity. Q2 net income ($62.6B) was flattered by a ~$53.4B non-cash Anthropic mark-to-market gain (GAAP diluted EPS $5.75 vs ~$1.90 underlying), and the AWS-funds-Anthropic-which-buys-AWS loop makes part of the AI demand circular. A reversal of the gain or an unwind of the loop would dent both earnings and the AWS narrative.
Scenario Stress Tests
| Scenario | Mechanism | Anchor PT | Delta vs base $310 |
|---|---|---|---|
| Base | AWS ~$215B at 20x; ads 18x; retail 12x; ~$92B net debt | $310 | — |
| AWS decelerates below 20% | AWS FY27E $190B at 16x; other buckets unchanged | ~$254 | (18%) |
| FCF drought persists into 2027 | AWS de-rates to 15x; ads to 15x; net debt to $110B | ~$248 | (20%) |
| Advertising re-rates higher | Ads $105B at 20x; other buckets unchanged | ~$323 | +4% |
| Retail margin disappoints | Retail $560B at 10x; other buckets unchanged | ~$301 | (3%) |
| Full bear (drought + share loss) | AWS 17x / ads 15x / retail 10x on lower rev | ~$235 | (24%) |
| Bull: AI re-rating | AWS $225B at 21x; ads $105B at 19x; retail $605B at 13x | ~$360 | +16% |
All stress-test PTs are derived from the same SOTP framework used in the interactive tool, each shock changes only the inputs noted, holding the other buckets and the ~10.75B share count constant. Deltas are vs the $310 base PT.
Bull vs Bear Debate
| Issue | Bull view | Bear view |
|---|---|---|
| The free-cash-flow drought | FCF is depressed by choice, operating cash flow ($148.5B, +30%) is already outgrowing capex; FCF inflects materially in 2027 as capex moderates and AI capacity monetizes. | ~$220B is the new floor, not a spike. A depreciation wave is coming; negative FCF with unproven payback is a real risk, not a timing quirk. |
| AWS AI position | +37% is the fastest in 18 quarters on a ~$169B base, with a ~$364B backlog and a custom-silicon cost advantage (Trainium) that defends share. | Even at +37% AWS still grows a touch below Azure (+40%) and well below GCP (+63%); the relative-share question persists, which threatens the premium multiple. |
| Is the SOTP real? | AWS earns 60% of operating income on 21% of revenue; valuing it at a premium cloud multiple is just arithmetic, and it makes the rest of Amazon look cheap. | The parts are not separable, segment margins shift with allocation, and the FY27E revenues are analyst estimates, the SOTP is precise but not necessarily accurate. |
| Earnings quality | Operating income ($27.5B, +43%, record 13.7% margin) is clean and growing; the Anthropic mark is a footnote, not the thesis. | Net income ($62.6B) leaned on a ~$53.4B non-cash gain (GAAP EPS $5.75 vs ~$1.90 underlying), and the Anthropic loop makes part of AWS demand circular, headline earnings flatter the truth. |
| Premium multiple justified? | ~27x for a business compounding revenue, expanding margins to a record, and carrying a cloud + ad mix is reasonable, and the SOTP says the parts are worth more. | ~27x with negative free cash flow and still-slower-than-peers AWS prices a payback that has not been demonstrated; multiple risk is asymmetric. |
Technical Analysis
AMZN trailing-12-month closes
RSI (multi-timeframe)
Firm after the post-Q2 breakout toward the ~$287 high, momentum turning up as the market re-rated the stock, but not yet stretched into overbought.
MACD vs Signal
Turned back up on the post-Q2 rally; momentum re-accelerating as the stock broke out toward new highs.
Relative strength (2026 YTD)
AMZN has outpaced the Nasdaq and the broad market year-to-date as the AWS-reacceleration re-rating overtook the capex/FCF overhang.
EMA stack (current)
Trader's view
- Price broke out above its moving averages on the post-Q2 rally, pressing toward the ~$287 52-week high; the trend has turned up.
- Key support: the ~$240 breakout shelf; below it, the $205 area (the prior base) and then the $196 52-week low are the lines in the sand.
- Key resistance: the ~$287 all-time high; a decisive break needs continued AWS momentum or the first FCF-inflection print.
- Momentum: MACD turning up, RSI firm but not extended, a stock that has re-rated on the AWS reacceleration.
Glossary & Methodology Notes
- AWS (Amazon Web Services)
- Amazon's cloud-computing segment, compute, storage, database, and the AI stack (Bedrock, SageMaker, custom Trainium/Inferentia/Graviton silicon). The company's profit engine: ~21% of revenue but ~60% of operating income.
- Trainium / Inferentia / Graviton
- AWS's custom silicon, Trainium for AI training, Inferentia for inference, Graviton for general compute. The cost-per-token advantage that the bull case argues defends AWS's AI share against Nvidia-based rivals.
- Advertising (carve-out)
- Amazon's ad business (sponsored products/brands, Prime Video ads, DSP), reported inside the consumer segments rather than as a standalone line. In the SOTP it is carved out of Retail to avoid double-counting and valued separately at a premium multiple.
- Free cash flow (FCF) vs operating cash flow
- Operating cash flow is cash from operations; free cash flow subtracts capital expenditures. Amazon's FCF has turned slightly negative because a raised ~$220B of capex now exceeds its $148.5B operating cash flow, the central debate is whether that capex earns its return.
- EV/EBITDA
- Enterprise value divided by EBITDA (earnings before interest, taxes, depreciation, amortization). A capital-structure-neutral measure used to value and cross-compare the three buckets and the peers.
- SOTP (Sum-of-the-Parts)
- A valuation method that values each business separately, sums the enterprise values, adjusts for net debt or net cash, and divides by shares. Used when one blended multiple obscures different economics. Amazon carries net debt, so the adjustment is a subtraction.
- Hyperscaler
- The largest cloud operators, AWS (Amazon), Azure (Microsoft), and Google Cloud (Alphabet). Their relative growth rates are the key read on AI cloud share, where AWS is currently the largest but slowest-growing.
- Backlog
- AWS's contracted-but-unrecognized revenue (remaining performance obligations). At ~$364B it underwrites multiple years of AWS revenue and is a leading indicator of AI demand.
Methodology
- Snapshot anchor: August 7, 2026. Live price patches via the Cloudflare-Worker quote proxy on page load.
- All figures in USD. Q2 2026 segment data is reported; FY27E bucket revenues and all SOTP multiples/margins are author estimates, not company guidance, the largest swing factor in the SOTP.
- SOTP scenarios use FY27E revenues; advertising is carved out of the consumer segments to avoid double-counting with Retail. AWS bucket EBITDA margin includes data-center D&A.
- Conclusions are the author's view. Illustrative, not investment advice.
Sources & Citations
Inline citations
Superscripted numbers in the body link here. Click any N in the report to jump back to the source.
- Amazon.com, Inc., Second Quarter 2026 Results (reported July 30, 2026): total net sales $200.6B (+20%, first $200B+ quarter), operating income $27.5B (+43%, record 13.7% margin). Amazon Q2-2026 results. ↩ ↩
- Amazon Q2 2026 segment detail: AWS revenue $42.2B (+37%), operating income $16.6B (39.4% margin, ~60% of total); North America operating income ~$9.3B; backlog ~$364B. Amazon Q2-2026 earnings. ↩ ↩
- Anthropic–Amazon compute partnership: Anthropic commits >$100B to AWS over ~10 years (up to ~5 GW); Amazon committed $5B more in April 2026 (up to ~$33B total potential; stake carried ~$74.2B at Q1, written up a further ~$53.4B in Q2 as the mark re-rated toward the $965B Series H). Anthropic then raised a $65B Series H at a $965B post-money valuation (May 28, 2026, up from the ~$380B Series G) and confidentially filed a draft IPO S-1 (June 1, 2026). Anthropic–Amazon (Apr 2026); Anthropic Series H (May 2026). ↩
- Amazon capex (~$220B 2026 guide, raised from ~$200B on higher memory costs) and free cash flow: trailing operating cash flow $148.5B (+30%) less capex left ~($7.6B) of Amazon-defined free cash flow (turned negative from positive a year ago). Capex guide per management commentary on the Q2 2026 call. ↩ ↩
Background reading
- Amazon.com, Inc. 2025 Annual Report (Form 10-K), annual financials, segment disclosures, risk factors.
- Amazon quarterly results (Q1–Q4 2025, Q1 2026, Q2 2026), revenue, AWS, advertising, operating income, capex, cash flow.
- Amazon Q2 2026 earnings call + slides, AWS growth, backlog, capex outlook, Trainium/Anthropic commentary.
- Microsoft / Alphabet cloud disclosures (Azure, Google Cloud), the hyperscaler growth comparison.
- Walmart financials, the retail EV/Sales and EV/EBITDA comparison.
- Market data: stockanalysis.com (price, shares, market cap, 52-week range, forward P/E) as of August 7, 2026.
Disclaimer. This report is the author's institutional equity-research view, prepared for portfolio and educational purposes. It is not a recommendation to buy, sell, or hold any security. Forward-looking statements are subject to risk and uncertainty; past performance is not indicative of future results. Consult a licensed financial advisor before making investment decisions. All third-party trademarks are the property of their respective owners.