E-commerce, cloud & digital advertising

Amazon

AMZN

The conditional BXDC deal provides another way to finance AWS data centers in exchange for long-term leases guaranteed by Amazon. The net impact remains unproven, so business strength and operating scenarios are unchanged.

Analysis and ratings: ChatGPT·Business analysis 11/18/2026·Ratings & valuation 11/18/2026
THESIS STATUSAWS strengthens · cash economics weaken
CONFIDENCEMedium in the businessLow for price attractiveness; three-year cash flows, CAPEX normalization and refinancing are uncertain.
HORIZON3–5 years
WHAT MATTERS MOSTAI capacity returns, AWS margins and cash flow after CAPEX.

Business strength & price attractiveness

How AI assesses them

Business strength

Medium confidence
7.5/ 10

Previously 7.5 ·
Unchanged · Assessed

Business strength, independent of the stock price.

The conditional BXDC deal provides another way to finance AWS data centers in exchange for long-term leases guaranteed by Amazon.

Five areas & supporting evidence

Price Attractiveness

Low confidence
3.4/ 10

Previously 3.5 ·
↓ −0.1 · Assessed

At 257.89 USD · close
Three-year horizon, through .

Price attractiveness 3.4/10 at USD 257.89 (November 17, 2026; regular close). Three-year scenario returns: +20.4% in the base case and −67.1% in the adverse case.

Scenarios & supporting evidence

What changed in the latest update

New information: The conditional BXDC deal provides another way to finance AWS data centers in exchange for long-term leases guaranteed by Amazon. The net impact remains unproven, so business strength and operating scenarios are unchanged.

Strength 7.5 → 7.5Attractiveness 3.5 → 3.4Scores changed

The numerical comparison does not by itself confirm or weaken the investment thesis.

Dated prices & model comparison
Input / modelPreviousCurrent
Reference price250.69 USD257.89 USD
Base modelled return+23.9 %+20.4 %
Adverse modelled return-66.2 %-67.1 %

Quote dates: 11/3/2026 → 11/17/2026. Model horizons end: 11/4/2029 → 11/18/2029. Cumulative returns include modelled dividends; they are scenarios, not guarantees.

Terminal-price and cumulative-dividend assumptions are unchanged; reference-price inputs differ as shown above.

View rating history →

Assessed · Methodology 1.2. Analytical judgments by ChatGPT; price changes do not automatically update these scores.

MARKET OVERVIEW

AMZN · Price chart

Delayed data · English controls.TradingView
Analysis contents · 13 sections
01

Company Snapshot

Amazon.com Inc. — AMZN | NASDAQ Sector: e-commerce, cloud infrastructure and digital advertising Thesis status: AWS strengthens; the cash-flow and financing thesis has deteriorated Confidence: medium in business strength, low in price attractiveness Time horizon: 3–5 years Analysis: October 30, 2026 · latest results: Q3 2026, released October 29

Amazon combines a large commerce and logistics platform with AWS and advertising. The opportunity rests on a growing services mix and efficiency increasing profit faster than revenue. Current AI expansion, however, consumes more cash than the business generates after investment.

30-second view: AWS accelerated to 39.3%, but the group’s record 13.9% operating margin is not equivalent to shareholder cash. TTM FCF deteriorated from minus USD 7.6 billion to minus USD 42.4 billion, while management expects significantly more 2027 cash CAPEX after approximately USD 220 billion in 2026. ChatGPT’s independent assessment: cloud demand is stronger, but investment-cycle returns remain unproven. Q3 results, public Q3 call transcript

What matters most

  • How quickly new AWS capacity turns into revenue and cash.
  • Whether AWS maintains healthy margins as depreciation ramps up.
  • Whether advertising and logistics efficiency improve commerce-ecosystem returns.
  • Whether operating cash-flow growth eventually outpaces investment growth.

November 18, 2026: The conditional BXDC deal provides another way to finance AWS data centers in exchange for long-term leases guaranteed by Amazon. The net impact remains unproven, so business strength and operating scenarios are unchanged. Analysis and scenarios: ChatGPT.

02

Investment Thesis

Amazon can compound value over the long term through three engines: AWS, advertising and higher retail productivity. Cloud and advertising increase the ecosystem's economic value; Prime, the marketplace and logistics support purchase frequency, selection and customer relationships.

AWS can monetize AI at several layers: compute, proprietary chips, storage, databases, models through Bedrock and infrastructure for agents. It does not need a single particular model to win. Placing AI alongside customers' existing AWS data and applications may be an advantage. This is an investment interpretation of management's strategy, not a guarantee of beating Azure or Google Cloud. Q2 call

Condition for success: rapid growth must deliver higher free cash flow per share after equipment replacement, financing and dilution. Backlog, reported earnings and the size of CAPEX alone do not meet that condition.

The main debate around the stock: the bull case sees today's investment as advance funding for profitable future growth. The bear case assumes infrastructure replacement, price competition and financing costs will absorb cash for the long term. The current price alone does not establish which scenario the market has priced correctly.

03

The Business

BusinessHow value is createdWhat to monitor
North AmericaProduct sales, marketplace, seller services, Prime and advertisingCustomer activity and margins after logistics costs
InternationalA similar ecosystem outside North AmericaSustainable profitability against local competition
AWSCompute, storage, databases, cloud and AI servicesGrowth, capacity utilization and infrastructure returns
AdvertisingSponsored Products, video and other inventoryMonetizing purchase intent without harming customer experience

Amazon reports three operating segments: North America, International and AWS. Advertising is a separate revenue category, not a fourth operating segment. Its revenue or estimated profit cannot simply be added to retail segments again in valuation. Q3 2026 results

04

Competitive Advantages

  • Logistics scale and density. More orders, regional inventory placement and automation can shorten delivery times and lower cost per shipment. Faster delivery also expands the range of purchases for which customers use Amazon.
  • Prime and the marketplace. Broader selection attracts customers, customer demand attracts sellers, and sellers help expand selection further. This mechanism creates value only if fees and advertising do not undermine sellers' economics.
  • AWS and switching costs. Migrating connected applications, data and operating processes is often difficult. The installed customer base can help bring AI into production.
  • Proprietary chips and model choice. Trainium and Graviton may improve price-performance; Bedrock offers multiple models in one environment. Technology advantages must show up in actual economics for customers and Amazon.
  • Advertising close to the purchase. Amazon reaches customers at the moment of a purchase decision. This is a different source of value from advertising reach alone.

These advantages reinforce each other but do not remove competition or capital intensity. Q2 call

05

Growth Drivers

AWS and production AI. Q3 brought further acceleration, but token-volume growth alone establishes nothing certain about profit per token. In its results release, Amazon reports nearly 7× higher Fortune 500 Bedrock spending and nearly 17× higher token volume year over year. These are company operating metrics without separately disclosed Bedrock margins. Jassy’s expectation of AWS growth above 40% in each of 2027 and 2028 is management’s outlook, not a reported result or an automatic base-case input. AWS and PwC described Amazon Quick deployments at two large US enterprises; Classmethod announced a five-year AWS collaboration. Contract economics were not disclosed.

Trainium, Graviton and Bedrock. Proprietary chips can expand capacity and improve cost structure. TorchNeuron, announced October 19, 2026, allows existing PyTorch code to run on Trainium with minimal configuration changes and could reduce adoption friction. This is a company product claim and technical step, not disclosed evidence of new revenue, utilization or returns. Broad profitable adoption beyond the largest AI labs remains the key test of platform competitiveness. On the Q3 call, management newly planned external Trainium sales from late 2027. That is a future possibility, not an already-reported chip segment; it receives no separate value in the base valuation.

Advertising and Prime Video. Q3 advertising revenue was USD 19.637 billion, up 10.9% reported and 12% excluding FX, versus 26% excluding FX in Q2. Management mainly attributes the slowdown to Prime Day timing and other comparisons and expects “high teens” Q4 growth. The slowdown cannot be declared entirely structural, but subsequent results must confirm the expected recovery. Video and sports expand advertising inventory while adding content and rights costs. Q3 results and guidance

Retail productivity and purchase frequency. Regionalization, robotics, everyday essentials and groceries can improve network utilization. Fast delivery must be judged by order profitability, not speed alone. Q3 units sold rose 12% versus 11% shipping-cost growth; productivity therefore offsets inflation, but North America’s 7.7% margin and International’s 3.7% margin were slightly below Q2. Q3 tables

Other growth options. Pharmacy, Amazon Leo and Zoox offer additional opportunities. The core thesis does not depend on their success; their investment and losses belong in group economics. Figures and plans here draw on management’s Q2 call. An October 13, 2026 Amazon announcement described a successful Leo connectivity demonstration for US Army DEVCOM C5ISR. The October 26 Q3 announcements summary reports nearly 400 Leo satellites after 14 missions. These are company-announced technical and operating milestones, not quantified new contracts or demonstrated returns; they do not change the base cash-flow assumptions. On the Q3 call, the CFO said initial Leo service should start in 2026, but capitalization of some manufacturing and launch costs will begin only in Q1 2027, later than previously expected. Later accounting capitalization does not remove cash investment.

Transamerica expanded its AWS relationship on November 17 for AI use in financial services. Contract value and margin effects were not disclosed.

06

Economics & Financial Profile

Amounts are USD billions unless stated otherwise. Q3 is the quarter through September 30, 2026; TTM is the twelve months through that date. Reported data come from the results release; calculations are explicitly identified.

MetricQ3 2026 / latest TTMInvestment significance
Q3 revenue201.944; +12.1% reported, +13% excluding FXNear the top of original 197–202 guidance; Prime Day timing reduced growth by about 4 percentage points
Q3 operating income / margin28.076 / 13.9%Income 1.576 above the original 26.5 guidance ceiling; calculated +5.9%
Q3 AWS revenue / operating income45.984 / 17.773Revenue growth 39.3%; 38.7% margin
North America / International operating income8.723 / 1.5807.7% / 3.7% margins
TTM operating cash flow168.300; +29% year over yearGrowing, but insufficient for investment
TTM net property purchases210.657215.192 gross purchases less 4.535 proceeds and incentives
TTM free cash flow−42.357Amazon definition: OCF less net property purchases; Q2 TTM was −7.604
Q3 cash CAPEX75.878 net77.259 gross less 1.381; CFO rounds to 75.9
Standalone Q3 OCF / FCF42.422 / −33.456ChatGPT FCF calculation = 42.422 − 75.878; distinct from TTM
2026 cash CAPEX plan / 2027 commentaryApproximately 220 / significantly more2026 reaffirmed; no numerical 2027 amount given

Sources: Q3 results, statements and supplemental tables, released October 29, 2026; public Q3 call transcript, CFO and prepared remarks. Original Q3 guidance comes from the Q2 call. Comparison is against prior company guidance, not verified market consensus.

Operating income and margin comparability. Q3 AWS margin is 38.7% reported, calculated as 17.773 / 45.984. That is below Q2’s reported 39.4%, but Q2 included a favorable USD 0.551 billion energy revaluation; removing it yielded approximately 38.1%. Deterioration in ordinary AWS economics cannot therefore be inferred without adjustment. Q2 one-offs are not automatically carried into Q3; a complete adjusted Q3 calculation is outside this assessment. Q3 statements, historical Q2 10-Q

Net income is not operating earning power. Q3 GAAP diluted EPS of USD 1.89 fell from USD 1.95 a year earlier despite operating-income growth of 61%. Q3 net-income bridge: operating income 28.076 − nonoperating expenses 1.207 − tax 6.180 − after-tax equity-method investment result 0.004 = 20.685; divided by 10.929 billion weighted-average diluted shares gives USD 1.89. The prior-year quarter included USD 9.5 billion of pretax Anthropic revaluation gains; current TTM EPS still includes earlier investment gains. Without allocated tax, no “adjusted EPS” is invented, and headline P/E is not used as proof of cheapness. Q3 income statement

Cash flow is the decisive counterweight to accounting profit. The gap between TTM OCF and net investment widened further from Q2. Finance-lease and other financed-obligation principal payments added 1.681 + 0.333 = USD 2.014 billion TTM; deducting these gives indicative shareholder cash of minus USD 44.371 billion, not the company’s FCF definition. Operating leases are already included in OCF. Management says infrastructure payback is shortening but has not disclosed an audited return on the entire investment program. Q3 cash-flow tables, Q3 call

Q3 10-Q supplement: AWS remaining performance obligations reached USD 628.1 billion with a weighted remaining term of 6.1 years. This is a contractual base, not current cash or annual revenue. The Q3 unrealized energy gain of USD 0.091 billion was primarily in AWS; an illustrative adjustment by itself does not confirm deterioration in the underlying margin.

07

Key KPIs

KPILatest figureThesis implication
AWS year-over-year growth39.3% in Q3 versus 36.7% in Q2Further acceleration on a large base
AWS operating margin38.7% reported in Q3Distinguish AI mix and component inflation from efficiency
AWS backlogUSD 578 billion according to Jassy on the Q3 callCall-only figure; terms, duration and comparability with Q2 RPO of USD 496 billion need confirmation in the new 10-Q
Advertising revenueUSD 19.637 billion; +10.9% reported / +12% excluding FXVerify the promised return to “high teens” Q4 growth
Retail marginsNorth America 7.7%; International 3.7%Slightly below Q2’s 7.9% / 4.1%; monitor seasonality and shipping
OCF versus net CAPEXTTM 168.300 versus 210.657Cash-flow gap widened
Shares and awards10.811 billion shares + 0.223 billion stock awards at September 30The 11.034 sum is not GAAP diluted shares; Q3 weighted-average diluted shares were 10.929
Liquidity / long-term debtUSD 114.226 / 173.125 billion at September 30Debt and further-expansion financing; not complete net debt

Sources: Q3 results and statements and public Q3 call transcript. Backlog is management commentary; other current figures come from accounting and supplemental tables. Q2 RPO and its then-duration are documented by the historical Q2 10-Q; the old 6.4-year duration is not carried onto the new backlog.

Backlog is neither current cash nor an unquestionable guarantee of future revenue. Customer concentration, contract economics and demand from end users versus externally financed AI companies also matter.

08

Management & Capital Allocation

Andy Jassy combines AWS experience with group leadership. Further cloud acceleration and operating income above the original guidance ceiling demonstrate execution, rather than a reason for personal trust in every investment. Weaker FCF and higher debt require a separate capital-discipline test. Q3 results

The decisive capital decision is the scale of AI expansion. The 2026 cash CAPEX plan remains approximately USD 220 billion, but Jassy expects significantly more in 2027 without numerical guidance. The previous model’s investment decline starting in 2027 no longer fits current commentary and has been replaced. Q3 long-term-debt proceeds were USD 48.102 billion; capacity growth is not financed only through operating cash. Higher expected demand does not prove returns above the cost of capital. Q3 statements, Q3 call

New capital signals intensify the test. On October 20, Amazon announced over EUR 16 billion of German investment for 2027, including EUR 4.8 billion in AWS cloud and AI infrastructure; on October 19 it described financing development of a 960 MW target nuclear project in Indiana. 3M’s 8-K describes a seven-year Amazon warrant mainly tied to payments for data-center products, with that portion fully vesting after USD 6 billion cumulative payments. It does not establish an unconditional Amazon purchase commitment of that amount. These announcements demonstrate scale and bargaining power, not yet investment profitability.

Investment and commercial relationships: The Q2 10-Q described two USD 5 billion Anthropic investments, the then-remaining USD 15 billion conditional framework and a subsequent USD 21.3 billion OpenAI payment. These are historical positions, not verified September 30 balances. The OpenAI payment preceded the new balance sheet and is not deducted from its cash again. Q3’s combined acquisitions and nonmarketable-investments line was USD 23.986 billion, but without the new 10-Q notes it is not entirely assigned to one company. These links require separating end demand from customer financing. Q3 cash flow

Capital discipline means curtailing weaker projects, explaining investment economics transparently and increasing cash flow per share. Long contracts help, but can limit repricing flexibility when input costs rise.

Financing and capital allocation: the 10-Q confirms payment of the remaining USD 21.3 billion to OpenAI in Q3, bringing the total to USD 50 billion; the payment is already reflected in the September balance sheet. The filing also adds detail on current debt and infrastructure financing. Better financing disclosure does not establish that returns on the investment cycle will exceed the cost of capital.

The November 11 BXDC agreement conditionally transfers 75.1% of two joint ventures holding AWS campuses (13 data centers, 647 MW). The USD 5bn gross asset value is not Amazon’s net proceeds; AWS retains 24.9% and is to receive a distribution from BXDC equity and mortgage financing. AWS is to lease the entire campuses; the weighted initial term is 12 years with 2.75% annual rent escalation. Amazon guarantees tenant payments. Net effects are unproven and the model excludes the deal.

09

Risks / Bear Case

  • Investment exceeds profitable demand. New capacity has weaker utilization or lower pricing while depreciation, interest and replacement continue. Deteriorating FCF and planned further CAPEX growth raise the cost of mistakes despite strong current AWS demand.
  • AI-lab demand concentration. Large contracts can depend on customers’ ability to raise capital. Investment and commercial links may obscure independent end demand.
  • Competition and AI mix compress AWS economics. Azure, Google Cloud and specialized providers can win workloads or reduce prices. In Q3 Q&A, the CFO said AI currently has lower margins than core AWS; AI growth alone therefore does not guarantee segment-margin expansion. Proprietary-chip advantages may not last. Q3 call
  • AI changes shopping. External shopping agents may weaken direct customer relationships and some advertising inventory. Amazon’s own agents can mitigate the risk while changing existing monetization.
  • Retail margins remain sensitive. Energy, shipping, wages, tariffs, competition and regulation can consume automation savings.
  • Other projects and accounting optics. Leo, Zoox and content can consume cash; investment revaluations may suggest stronger earnings than ordinary operations support.
  • Valuation decline. Investors can lose money despite business growth if expected cash flow or market-required returns deteriorate.

These are analytical loss mechanisms, not predictions. Summit Research’s December 17, 2025 article on financed AI demand is a historical supplemental opinion. Its old targets and speculation are not current valuation inputs.

10

Valuation

ChatGPT produces the calculations and valuation scenarios. Share-price inputs are dated to the analysis; they are not continuously updated market data.

CHATGPT SCENARIOS · November 18, 2026 → November 18, 2029

Different outcomes. Different values.

Historical analysis input
257.89 USD · November 17, 2026 · regular close
Adverse
84.8 USD
Base
310.6 USD
Conditional scenarios from the analysis. The dashed line marks the historical share price used in the calculation. Values are neither current intrinsic value nor a share-price forecast.

Price attractiveness 3.4/10 at USD 257.89 (November 17, 2026; regular close). Three-year scenario returns: +20.4% in the base case and −67.1% in the adverse case.

Reference: Amazon common stock, Nasdaq AMZN, USD 257.89 regular close at 16:00 EST on November 17, 2026. Assessment dated November 18, 2026. September 30 balance-sheet and share data are used.

At USD 257.89 and 10.811bn shares, indicative capitalization is USD 2.79tn, or approximately USD 2.85tn including 0.223bn stock awards. The latter is a dilution sensitivity, not a GAAP diluted share count.

The September 30 balance sheet contains USD 114.226 billion cash and marketable securities (55.783 + 58.443), USD 173.125 billion long-term debt and USD 98.824 billion long-term lease liabilities. Long-term debt rose from USD 65.648 billion at 2025 year-end; this is not total debt including the current portion. The balance sheet therefore provides a large buffer but demonstrates substantially greater financing dependence. New Q3 10-Q maturities and current portions still need confirmation; old June balances are not presented as current. Q3 balance sheet

Preferred framework: value AWS and the commerce ecosystem through normalized cash flows or a consistent sum of the parts, including replacement investment, financing, dilution and loss-making projects. Advertising cannot be counted again within the segments that already contain it. Investment revaluations distort reported P/E, while negative current FCF cannot be valued as stable annual earnings.

How much cash the price must justify

The following is an arithmetic test of a hypothetical multiple of steady annual shareholder cash flow, not a DCF, target price or consensus. It uses indicative USD 2.443 trillion market capitalization at USD 226 and does not add debt separately. If that cash flow is achieved only several years later, discounting reduces its present value.

Assumed cash-flow multipleRequired annual cash flowCorresponding cash-flow yield
20×About USD 139bn5.0%
25×About USD 112bn4.0%
30×About USD 93bn3.3%

ChatGPT calculation: capitalization divided by the chosen multiple. Including all stated stock awards illustratively raises required amounts to approximately USD 125, 100 and 83 billion. Assumed multiples are not fair-value estimates.

Price Attractiveness

Both scenarios are independent ChatGPT assumptions, not guidance or consensus. The new starting base is TTM through September 30, 2026: USD 161.382 billion AWS revenue and USD 636.073 billion North America plus International revenue. Three growth steps approximate annual performance at the horizon’s end: a normalized run rate, not a precise quarterly-seasonality forecast. Advertising is inside retail segments and is not added separately. Starting TTM

ItemBase scenarioAdverse scenario
Business assumptionsAWS grows 30% annually and retail segments 10%; terminal revenue approximately USD 1,201 billion. AWS margin 36%, retail 9%; operating income approximately USD 204 billion. AWS growth is below management’s outlook and margin below the current reported level.AWS grows 14% and retail 7% annually; revenue approximately USD 1,018 billion. Weaker utilization and pricing reduce AWS margin to 30% and retail to 6%; operating income approximately USD 118 billion.
Annual operating cash flow at horizon endUSD 340 billion after interest and cash taxes; +26.4% annually from starting TTM.USD 265 billion; +16.3% annually, but high replacement investment and interest constrain shareholder cash.
Annual net cash CAPEX / financed principalUSD 220 / 3 billion; shareholder cash USD 117 billion.USD 220 / 4 billion; shareholder cash USD 41 billion.
Horizon-end shares11.3 billion; USD 10.35 cash flow per share.11.6 billion; USD 3.53 cash flow per share.
Future valuation30× shareholder cash flow; a 3.3% cash-flow yield requires continued growth and proven efficiency.24×; 4.2% yield. Lower multiple for weaker growth and returns, not a crisis minimum.
Three-year price, November 18, 2029USD 310.62USD 84.83
Three-year dividendsUSD 0 per share, ChatGPT assumption, without reinvestment.USD 0 per share, ChatGPT assumption, without reinvestment.
Three-year total return from USD 257.89+20.4%, about +6.4% annually−67.1%, about −31.0% annually

Terminal price equals operating cash flow less net cash capex and principal on financed obligations, divided by shares and multiplied by the valuation multiple. Three-year return uses USD 257.89 with no dividends.

OCF bridge and sensitivity. OCF is after interest, cash taxes and operating leases. The aggregate modeled terminal bridge is: base operating income 204 − cash interest 10 − cash taxes 40 + net noncash adjustments and working capital 186 = USD 340 billion; adverse 118 − 15 − 23 + 185 = USD 265 billion. Amounts are rounded ChatGPT assumptions, not a detailed depreciation model. For comparison, TTM OCF of 168.3 exceeds TTM operating income of approximately 104.4 by USD 63.9 billion; the model therefore requires a substantial depreciation ramp as assets expand. Neither depreciation nor SBC is an economic return: CAPEX captures replacement and share count captures dilution. Financed principal outside OCF was USD 2.014 billion TTM versus 3 / 4 billion modeled; operating rent is not deducted again. Starting statements

CAPEX and financing. Base 2027 / 2028 / 2029 OCF is 245 / 290 / 340, net cash CAPEX 270 / 250 / 220 and financed principal 3 / 3 / 3 USD billions, leaving −28 / 37 / 117. Adverse OCF is 205 / 235 / 265, CAPEX 285 / 255 / 220 and principal 4 / 4 / 4, leaving −84 / −24 / 41. Numerical 2027 CAPEX assumptions are not guidance: they convert “significantly more” commentary into a testable model. The previous CAPEX decline to 210 already in 2027 and terminal 175 / 180 has been replaced. Even the new base is not conservative in every respect: after investment growth, it assumes stabilization; sustained CAPEX of 270 would reduce terminal cash by USD 50 billion.

Financing bridge and liquidity limits. The starting point is USD 114.226 billion liquidity at September 30, not June liquidity less an already-paid OpenAI installment. Q4 liquidity developments are not modeled as known facts. The base funds the initial deficit through liquidity and operations, rolls maturing debt and assumes no further net debt expansion after 2027. The adverse case accumulates a USD 108 billion deficit in the first two years and permits up to USD 60 billion additional net debt to preserve an operating reserve; higher interest is already in OCF. New borrowing is not a shareholder return. Ordinary maturing bonds are refinanced; financed principal of 3 / 4 billion is a separate payment outside OCF, not all debt maturities. No private-stake monetization or unverified credit line is assumed. The new 10-Q must supplement contractual commitments and maturities; this is not a full liquidity test, and refinancing failure would materially worsen it.

Dilution and valuation. The initial 11.034 billion sum of shares and awards is not GAAP diluted shares. The model assumes further net dilution of approximately 0.8% / 1.7% annually to 11.3 / 11.6 billion, without buybacks. Multiples of 30× / 24× are retained for comparability and value shareholder cash after interest and financed principal, not pre-debt enterprise value. Debt is not deducted again, but financing must remain sustainable at assumed interest and refinancing terms. Terminal multiples preserve platform value, but even adverse 24× may be too high if confidence is lost. Private stakes, excess cash and retained interim cash flow are not added separately; reserves remain for financing and replacement. This does not imply those stakes have no value.

  • Main weakness: High sensitivity to future depreciation, replacement CAPEX, working capital and refinancing. Higher AWS demand can be real while cash returns on new capacity lag growth.
  • Assessment confidence: Low. The new results balance sheet and reference quote are verified, but management has not disclosed numerical 2027 CAPEX, and complete new 10-Q notes and a detailed depreciation schedule are not available.
  • Condition for change: A higher assessment requires documented cash-CAPEX stabilization and per-share cash growth, or a price with more protection under the same assumptions. At USD 200, identical scenarios return approximately +55% / −58%, prompting reassessment rather than automatic attractiveness. Persistently higher CAPEX without matching OCF, rising financing risk or lower margins would reduce the score.
  • Main sources: Q3 results and TTM through September 30, 2026, released October 29; public Q3 transcript, Quartr through StockAnalysis, October 29, especially CAPEX and Q&A; historical prices and Investing.com, verified October 30. The USD 226 post-results quote is October 29 at 19:59 EDT; last regular close USD 243.79. OCF, CAPEX, dilution and multiples are ChatGPT forecasts.

Valuation status: updated for methodology 1.2 and Q3 facts. More profitable AWS growth does not automatically outweigh the next investment round’s costs; old rapid-CAPEX-decline scenarios are no longer current valuation. Numerical scenarios are defensible but highly sensitive estimates, not DCF intrinsic value, consensus or management targets.

Price history.

11

What Would Change the Assessment?

The following thresholds are working analytical rules, not Amazon guidance or automatic trading instructions.

ObservationWhy it would change the view
AWS slows below 20% year over year for at least two quarters while CAPEX remains highThe link between investment and profitable demand weakens
AWS margin, adjusted for material one-offs, falls below 30% for at least two quarters without a credible return pathGrowth may have significantly weaker economics
FCF remains structurally weak after today's capacity ramps and debt keeps rising materiallyThe issue may be persistent capital intensity rather than timing
Performance of major AI contracts or key customers' financing deterioratesBacklog has less economic value
Retail margins retreat persistently without better customer activityInvestment in speed fails to earn adequate returns
Production AI demand broadens beyond the largest labs and FCF per share grows convincinglyConfidence in long-term returns strengthens and risk concentration falls

Assess each signal alongside service mix, accounting changes and investment timing. Short-term margin fluctuation alone does not break the thesis.

12

Current View

Q3 was mixed for the thesis, not an unambiguously weak quarter. Cloud demand and operating income improve, but advertising slowed and growth became much more financially demanding. ChatGPT’s independent conclusion: AWS strengthens the business story, while positive free cash flow per share is delayed and greater debt raises shareholder risk.

Working view: retain the thesis and scrutinize the capital cycle again. The post-results price decline alone does not establish undervaluation; converting higher AWS growth into cash after financing is decisive. The model increases growth potential and outcome dispersion simultaneously.

New guidance and next evidence: Q4 revenue guidance is USD 240–247 billion, operating income USD 28.5–33.5 billion, with 2026 CAPEX near USD 220 billion reaffirmed. Relative to original Q3 guidance, revenue was near the upper end and operating income above it; this is a company benchmark, not consensus comparison. Monitor advertising’s return to “high teens,” AWS margin as AI grows and a specific 2027 investment plan. Q3 guidance

Evidence: newly reported figures come from the results release; commentary and Q&A from the public transcript. Key call debates are incorporated into their relevant sections: Justin Post (BofA) → Jassy on growth and irregular capacity additions; Mark Mahaney (Evercore ISI) → Olsavsky on lower AI margins and no numerical AWS-margin guidance; Doug Anmuth (JPMorgan) → Jassy on expansion costs and returns without a disclosed numerical bridge. The bull test is further growth with margins and cash; the bear test is a continuing investment gap and debt. Public Q3 transcript

Business strength

Business strength: 7.5/10. Assessed October 30, 2026 (Europe/Prague). Methodology: 1.2.

AreaScoreMain evidence
Competitive advantage9.0/10Prime, marketplace, logistics, advertising and AWS integration still create a multisource moat; Q3 does not invalidate it. Bedrock expands model access, but technical promises do not guarantee lasting superiority.
Demand durability8.3/10Raised from 8.0: AWS accelerated to 39.3% on a large base with broad AI adoption. Concentration and AI-customer financing risks remain; advertising’s slowdown cannot be ignored.
Management execution8.0/10Retained: Q3 operating income exceeded the company guidance ceiling and both retail segments are profitable. Investment execution does not equal demonstrated capital returns.
Business economics5.5/10Reduced from 6.0: TTM FCF minus USD 42.357 billion versus minus 7.604 billion in Q2, weaker still after other financed principal. High AWS margins do not yet mean positive shareholder cash.
Financial resilience6.5/10Reduced from 7.0: USD 114.226 billion liquidity is large, but USD 173.125 billion long-term debt and further expansion reduce the cushion. New 10-Q maturities and commitments remain necessary.

Calculation: (9.0 + 8.3 + 8.0 + 5.5 + 6.5) / 5 = 7.46 → 7.5/10. Each area has 20% weight; rounding occurs only at the final average. Share price and ownership are excluded.

  • Brief rationale: Demand evidence improved, but new statements show materially worse cash economics and greater debt dependence. In the average, those negatives outweigh the demand-score increase. The modest 7.6-to-7.5 reduction reflects the investment gap and financing risk, not price movements or one weaker EPS figure.
  • Main weakness: The business must turn high CAPEX and long AI contracts into sustainable per-share cash after infrastructure replacement, financing and dilution.
  • Assessment confidence: Medium. New reported results and the public call support all five areas, but incomplete coverage of new commitments and maturities and uncertain capital returns prevent high confidence.
  • Condition for change: A higher score requires repeatedly positive FCF after financed principal, per-share growth and broader profitable AI adoption without a further material leverage increase. Adjusted AWS margin below 30% or growth below 20% for at least two quarters under high CAPEX, or weaker customer financing, would lower the score.
  • Main sources: Q3 results, released October 29, 2026, Q3 and TTM through September 30; public Q3 call dated October 29; historically Q2 results and Q2 10-Q. Assessed October 30; new statements document business changes, while management outlook is distinct from reported returns.

The conditional BXDC deal provides another way to finance AWS data centers in exchange for long-term leases guaranteed by Amazon. The net impact remains unproven, so business strength and operating scenarios are unchanged. Price attractiveness 3.4/10 at USD 257.89 (November 17, 2026; regular close). Three-year scenario returns: +20.4% in the base case and −67.1% in the adverse case. Company strength: 7.5/10.

13

Thesis History / Updates

PeriodImportant developmentThesis impact
Q3 2025AWS USD 33.0 billion; 20.2% growthRenewed cloud acceleration
Q4 2025AWS USD 35.6 billion; 24% growthAcceleration continues on a growing base
Q1 2026AWS USD 37.6 billion; approximately 28% growthFurther cloud and AI demand evidence
Q2 2026AWS USD 42.2 billion; 36.7% growth; CAPEX plan raised to approximately USD 220 billionStronger operating thesis with a greater requirement to prove returns
October 8, 2026Analysis created; TTM FCF minus USD 7.6 billionBusiness quality separated from stock price and accounting profit
October 10, 2026 — historical assessmentFirst business strength 7.6/10 and price attractiveness 4/10; methodology 1.0, USD 262.43 October 9, 2026 close. Added AI-customer investment links, more precise adjusted AWS margin and three-year scenarios.Operating thesis retained; scenario returns insufficiently compensate for risks.
October 14, 2026 — historical assessmentStrength 7.6/10 (9.0 / 8.0 / 8.0 / 6.0 / 7.0); attractiveness 4.0/10 at USD 258.58 October 13 close; methodology 1.1. Three-year prices 329.46 / 139.83, returns +27.4% / −45.9%. Added Leo milestone.Judgment retained: no new evidence of a business change and modest cheapening does not offset investment risks.
October 21, 2026 — historical assessmentStrength 7.6/10 (9.0 / 8.0 / 8.0 / 6.0 / 7.0); attractiveness 4.0/10 at USD 262.63 October 20 close; methodology 1.2. Unchanged three-year prices 329.46 / 139.83, returns +25.4% / −46.8%. Added TorchNeuron, Germany investment plans, Indiana energy project, 3M relationship and confirmed Q3 date.Judgment retained: product and investment scope documented, returns not yet proven. Higher price slightly weakened risk–return, insufficiently to change the score.
October 28, 2026 — historical assessmentStrength 7.6/10 (9.0 / 8.0 / 8.0 / 6.0 / 7.0); attractiveness 4.0/10 at USD 253.10 October 27 close; methodology 1.2. Unchanged three-year prices 329.46 / 139.83, returns +30.2% / −44.8%. Added Amazon Quick production examples, Classmethod collaboration, Mintlify scale and Leo status.Judgment retained: more adoption evidence, but no disclosed contract values or new results. Lower price modestly improved the profile, insufficiently to change the score.
Q3 2026AWS USD 45.984 billion; +39.3%; group operating margin 13.9%; TTM FCF minus USD 42.357 billionDemand strengthens while growth’s financial demands increase
October 30, 2026 — post-results assessmentStrength 7.5/10 (9.0 / 8.3 / 8.0 / 5.5 / 6.5); attractiveness 4.0/10 at USD 226.00, October 29 extended trading at 19:59 EDT / October 30 at 00:59 CET; methodology 1.2. Last regular close USD 243.79 on October 29. New three-year prices 310.62 / 84.83, returns +37.4% / −62.5%, dividends USD 0.Higher AWS demand raises that component, while weaker FCF and debt lower economics and resilience; the average falls from 7.6 to 7.5. Higher model CAPEX replaces the obsolete rapid-normalization assumption. Lower price improves entry, not return certainty.

Historical comparability: October 10 (1.0) and October 14 (1.1) assessments are historical and not directly comparable without aligning methodology; they were not recalculated. October 21, 28 and 30 reviews use 1.2. Current business strength fell because of new evidence, not price; attractiveness stayed 4.0 at a different quote and revised scenarios. Historical references are regular closes; current USD 226 is post-results extended trading, explicitly distinguished.

Sources

Analytical limitations: the model is not a complete test of commitments and maturities in the new Q3 10-Q. Without time- and definition-matched consensus, a market beat/miss cannot be assessed; reported results are compared with prior company guidance. Capital-return estimates and the financing bridge remain uncertain.

| 4 Nov 2026 | Added the Q3 10-Q, including RPO, energy revaluation, debt and infrastructure financing. Company strength remains 7.5/10; price attractiveness falls to 3.5/10 at USD 250.69. | Better coverage does not change the business score; the higher entry price worsens risk/reward. |

| November 18, 2026 | The conditional BXDC deal provides another way to finance AWS data centers in exchange for long-term leases guaranteed by Amazon. The net impact remains unproven, so business strength and operating scenarios are unchanged. | Price attractiveness 3.4/10 at USD 257.89 (November 17, 2026; regular close). Three-year scenario returns: +20.4% in the base case and −67.1% in the adverse case. |

Was this analysis useful to you?

Have a suggestion about this analysis?

Suggestions are not displayed publicly. They help inform future updates. No name or email is required.

Sources and scenarios reflect the date stated in the analysis.Back to all companies