Platform commerce and cloud. BABA / AMZN. Run 16 September 2026. Built from Alibaba results releases FY2021 to FY2026 and the June 2026 quarter (6-K, 20 Aug 2026), the FY2026 20-F; Amazon Q4 releases 2020 to 2025, the 2025 10-K segment data and the Q2 2026 release (30 Jul 2026); earnings-call transcripts; Omdia, Synergy, eMarketer, CIRP, Hugging Face; regulator releases (FTC, European Commission, Federal Register, US DoD). Events swept September 2025 to 16 September 2026; most recent events checked: Alibaba HK$80bn new-share placement (24 Aug 2026), Amazon GBP4.25bn sterling bond (9 Sep 2026). Figures in USD; Alibaba translated at RMB6.898 per US$ (company FY2026 convenience rate) for every year, ratios computed in RMB. Alibaba fiscal year ends 31 March, one quarter offset from Amazon's calendar year. Evidence is web primary sources; the project sources folder was not connected to this run. Not a valuation and not a recommendation.
The next five years ride on three Amazon cells with forward visibility. Global cloud: AWS growth reaccelerated to 37% in Q2 2026 with a $496bn backlog (up from $364bn a quarter earlier), and the power build targets doubling capacity by end-2027, so revenue is capacity-constrained rather than demand-constrained (Q2 2026 call). US advertising: $19.8bn, +26%, monetizing the same shopper traffic without new fulfilment cost. North America commerce: +16% in Q2 2026 on faster same-day delivery (over 40% more items same-day or overnight in H1 2026). Alibaba's fastest cells, China cloud (+45%, AI revenue in triple digits for 12 quarters) and quick commerce (+45%), are real but sit on a smaller base and one of them loses money until FY2029 guidance, while the China commerce core that funds both is flat (CMR -7% reported, +1% like-for-like, Jun-Q 2026 release).
This is where the lenses disagree. On a three-year average Alibaba still shows the fatter engine: 39.2% gross margin vs 33.2% and 10.4% EBIT margin vs 9.4%, because a take-rate marketplace carries no inventory and Amazon ships 40% of units first-party. But the averages describe a fortress that stopped compounding: China E-commerce adjusted EBITA fell 44% in FY2026 (RMB193.2bn to RMB107.5bn) as quick-commerce subsidies and merchant support ate the take rate, and CMR growth faded from +10% (Jun-Q 2025) to +1% like-for-like. Amazon's conversion mechanism runs the other way: the mix shift toward AWS (18% of 2025 revenue but 57% of operating income) and advertising lifts group margin every year, and AWS is raising AI capacity prices 15 to 20% while Alibaba cuts Qwen token prices 10 to 20%.
Alibaba's exposure is a funding loop: the China commerce profit pool pays for quick-commerce losses and a cloud capex run-rate that hit RMB67.7bn in one quarter (+75%), free cash flow was -RMB44.7bn that quarter, and the gap was closed with a HK$80bn new-share placement in August 2026. If CMR slips from +1% to negative, the loop has no internal funder. Amazon's exposure is concentration inside AI demand: 2026 capex of about $220bn against trailing free cash flow of -$7.6bn, with a large part of the backlog tied to OpenAI and Anthropic, both of which Amazon also finances. Amazon breaks later because its capex sits against signed commitments and a positive-margin retail business, while Alibaba's spend sits against uncontracted share defence in a price war regulators are already policing.
The two companies barely meet head-on: their cells are mostly separated by geography, so this table shows where each is strong rather than a cell-by-cell fight. The six rows carry essentially all revenue of both groups.
| Cell (product x region) | Alibaba | Amazon | Why (one clause, sourced) |
|---|---|---|---|
| E-commerce x US | 1 | 5 | Amazon about 40% of US e-commerce (eMarketer) with ads +26%; AliExpress lost the de minimis channel 29 Aug 2025 (Federal Register), Alibaba US share ND |
| Cloud x ex-China (global) | 1 | 5 | AWS 28% share, #1 (Synergy Q2 2026), backlog $496bn; Alibaba outside the global top three (inferred from Synergy ranking) |
| Cloud x China | 5 | 0 | Alibaba 37% of China cloud infrastructure, next Huawei 17% (Omdia Q4 2025); AWS not in Omdia's top four |
| E-commerce marketplace x China | 3 | 0 | Alibaba still the largest by CMR base but CMR +1% LFL and share vs PDD/Douyin ND; Amazon closed its China domestic marketplace in 2019 (approx., unverified) |
| Commerce x International (Europe, Japan, RoW) | 2 | 3 | Amazon International $161.9bn, +11%/yr at a 2.9% margin; AIDC $20.9bn near break-even with a EUR550m EU DSA fine (EC, Jul 2026) |
| Quick commerce x China | 3 | 0 | Revenue +47% in FY2026 but loss-making, profitability guided for FY2029 (Jun-Q 2026 call); share vs Meituan ND |
Scores are anchored to the exhibits in the three tabs. Scores are per cell and are not summed.
Normalization: Alibaba's China E-commerce Group less quick commerce is aligned to Amazon's North America segment as "home commerce"; AIDC to International; Cloud Intelligence to AWS. Amazon's North America figure includes its US advertising and subscriptions, Alibaba's China figure includes CMR (advertising plus commission) and direct sales. Alibaba resegmented again from the June 2026 quarter (E-commerce Group, AI Cloud & Compute, AI Labs & Applications, All Others), so FY2026 is the last year on the older basis used here.
The cell picture has one headline: all of Amazon's revenue sits in three cells growing 10 to 20% a year, and its most profitable cell is the fastest. At Alibaba the fastest cells (cloud, quick commerce) are about 23% of segment revenue, and the cell that earned nearly all group profit (China commerce, RMB193bn EBITA in FY2025) is the one being taxed to fund them.
| Cell | Alibaba rev | growth | Amazon rev | growth | Margin signal / leader |
|---|---|---|---|---|---|
| Home commerce | $69.0bn | CMR +5% FY26; +1% LFL Jun-Q26 | $426.3bn | +10%/yr; +16% Q2'26 | Alibaba China E-com EBITA 19.4% (FY25: 38.0%) vs Amazon NA 6.9%; Alibaba richer but falling |
| Cloud | $22.9bn | +34%; +45% Jun-Q26 | $128.7bn | +19%/yr; +37% Q2'26 | AWS 35.4% op. margin (39.4% Q2'26) vs Alibaba 9.0% EBITA (12% Jun-Q); Amazon |
| International commerce | $20.9bn | +9% | $161.9bn | +11%/yr | AIDC -1.4% vs Amazon Intl 2.9%; Amazon |
| Quick commerce, China | $11.4bn | +47% | 0 | n/a | Loss-making, loss ND (company does not disclose) |
| All other | $36.9bn | -25% | 0 | n/a | EBITA -RMB35.7bn; disposals of Sun Art, Intime |
| Memo: advertising engine | CMR $49.9bn | +5% | $68.6bn | +21%/yr | Subset of home commerce rows above; Amazon's is growing four times faster |
Cloud. AWS growth is supply-led: Jassy said capacity will be short in 2026 and 2027, capex guidance rose to about $220bn, and multi-year commitments from OpenAI ($38bn Nov 2025, expanded Feb 2026) and Anthropic ($100bn+ over 10 years, up to 5GW of Trainium, Apr 2026) pre-sell it. Alibaba Cloud growth is demand-led and model-led: AI product revenue RMB12.4bn in the quarter, MaaS ARR above RMB16bn with a RMB30bn year-end target, and 151,448 Qwen derivative models on Hugging Face (2.6x Meta's). Alibaba's driver is durable in China but chip-constrained; Amazon's is contracted.
Home commerce. Amazon's engine is speed and ads: rural same-day expansion ($4bn+) and Amazon Now in 80+ cities lift frequency, and ads monetize the traffic at no fulfilment cost. Alibaba's engine was take rate (0.6% software service fee from Sep 2024, Quanzhantui adoption), which lifted CMR to +10% and has now lapped; the new growth lever is quick commerce, which buys frequency with subsidies. All growth here is organic for both; Alibaba's revenue decline in All Others is disposal-driven, not operating.
Both own their demand channels (apps, membership). The difference is what they are doing with price. Amazon raised 2026 FBA fees about $0.08 per unit with no referral fee increase and added a 3.5% fuel surcharge in April 2026 (secondary source), while 3P share of units held at 60 to 61%, so sellers absorbed it without leaving. Alibaba reclassified merchant support into contra-revenue in FY2027, turning CMR from +1% to -7% reported: the platform is now paying merchants to stay active.
| Input | Alibaba | Amazon |
|---|---|---|
| AI accelerators | US H200 licences granted May 2026 but deals stalled on Beijing guidance; T-Head Zhenwu at scale, 650+ external customers | Nvidia GB200/GB300 plus own Trainium and Graviton5; memory costs pushed capex up |
| Logistics | Asset-light (Cainiao, third-party couriers) except quick-commerce riders | Owned network, 1.595m employees, NLRB joint-employer settlement May 2026 |
| Inventory | Marketplace, 1P share ND | About 40% of units 1P |
| Cross-border | AliExpress Choice hit by US de minimis end | Not importer of record for most items; about $600m tariff refunds Q2 2026 |
Global cloud: AWS 28%, Microsoft 20%, Google 15%, market $143.4bn growing 43% (Synergy Q2 2026); AWS is growing slower than the market, a share donor at the margin despite reacceleration. China cloud: Alibaba 37%, Huawei 17%, ByteDance about 13%, Tencent 10% (Omdia Q4 2025), Alibaba gaining. US e-commerce: Amazon about 40%, Amazon plus Walmart about 51% (eMarketer). China e-commerce shares for Alibaba, PDD, JD and Douyin: ND from a free credible source; CMR at +1% like-for-like is consistent with Alibaba donating share (inferred).
AWS: a large share of backlog comes from two model labs Amazon also funds, so a capital crunch at either would hit bookings and the depreciation base together. Alibaba Cloud: US export controls and Beijing's purchase guidance decide how fast capacity comes online. Amazon North America: the FTC antitrust trial (29 March 2027) targets seller-fee and Prime-eligibility practices at the core of the flywheel. Alibaba China commerce: SAMR's January 2026 subsidy investigation and June 2026 "Ten Rules" can end the price war, which helps margins, but could also freeze Alibaba's quick-commerce share below Meituan's. Alibaba is more exposed in every pair because its risks are access and policy, not cost.
China AI cloud leadership: 38.1% AI-cloud share (Omdia) and the largest open-model ecosystem (151k Qwen derivatives), durability High inside China because foreign clouds are excluded. Membership base: 64m 88VIP members growing double digits, durability Med (PDD and Douyin compete on price). Own silicon (T-Head), durability Med, years to replicate but not yet at Nvidia parity. Ant Group stake (about 33%, current % unconfirmed) adds payments data.
AWS installed base and backlog: 28% global share, $496bn backlog, durability High (migration and retraining costs, multi-year commits). Fulfilment density: same-day to 4,000+ small communities, over 40% more same-day items in H1 2026, durability High (years of build, not money alone). Advertising on first-party purchase data ($68.6bn in 2025), durability High. Prime about 203m US members (CIRP estimate; Amazon does not disclose).
Alibaba: no disclosed customer above 10% of revenue (ND, inferred none); merchants gained a 0.6% fee in 2024 but now receive merchant support booked as contra-revenue; cloud customers got 10 to 20% Qwen price cuts in August 2026 (secondary source). Amazon: no disclosed customer above 10%; sellers took a 2026 FBA increase and a surcharge while 3P unit share stayed at 60 to 61%; AI customers took 15% and 20% GPU reservation price rises in January and July 2026 (secondary source). Prime held at $139 since 2022, so consumer pricing power is exercised through ads, not the fee.
Alibaba: dual sourcing of AI chips is Low to Med (domestic T-Head plus restricted Nvidia); pass-through Weak, since cloud prices are falling while capex rose 75%. Amazon: dual sourcing High (Nvidia plus Trainium plus Graviton); pass-through Strong in cloud (GPU price increases) and Partial in retail (fee surcharge). Incident: Amazon's 2026 capex rise was blamed on memory prices, a cost shock it is passing on; Alibaba's May 2026 H200 approvals stalled on its own government's guidance, an access shock it cannot price.
| Cell | Route control | Pocket price | Continuity | Outcome (share / margin) | Confirming KPI |
|---|---|---|---|---|---|
| Global cloud (AWS) vs China cloud (Alibaba) | Both direct | AWS rising; Alibaba falling on AI tokens | AWS capacity-short; Alibaba chip-gated | AWS share flat, margin up; Alibaba share up, margin up slowly from 12% | Segment margins quarterly |
| US commerce (Amazon) | Amazon direct, Prime | Fees up modestly, ads +26% | FTC trial Mar 2027 | Share flat to up, margin up | NA operating margin at or above 8% in 2027 |
| China commerce (Alibaba) | Alibaba direct, 88VIP | Take rate rebated via contra-revenue | SAMR subsidy rules | Share down (inferred), margin down | CMR like-for-like growth above +5% |
The tension: Alibaba grows fastest exactly where it has the least pricing power (AI tokens, quick commerce), and Amazon grows fastest where it has the most (AI capacity, ads). That is the single most useful finding of the run.
1. Competitive intensity of the home market: Alibaba faces PDD, Douyin, JD and Meituan simultaneously; Amazon's nearest rival Walmart is a partial overlap. Impact Major. Alibaba cannot change the market; the fix is regulatory (subsidy caps) and would take 12 to 24 months to show in CMR. 2. Chip access: Impact Major for cloud; closing it means T-Head reaching frontier parity, a multi-year project money alone does not buy. 3. Fulfilment ownership: Amazon's owned network turns speed into share; Alibaba's asset-light model kept margins high but gave Meituan the instant-delivery beachhead. Impact Moderate; Alibaba is buying its way in, which is the cost showing in S&M.
Mapping. Alibaba reports by function, but its cost of revenue already includes logistics, payment processing and data-centre depreciation, so Amazon's COGS proxy is cost of sales plus fulfillment. R&D is not clean: Amazon's technology and infrastructure line includes AWS data-centre depreciation, so its 14.6% overstates true R&D against Alibaba's product development. SG&A is sales and marketing plus G&A for both. EBIT is GAAP operating income; Alibaba's includes intangible amortization and goodwill impairment (RMB14.6bn in FY2026). Alibaba FY2024 to FY2026 (to March), Amazon 2023 to 2025: a one-quarter offset. No group-level restatements found.
| % of sales, 3y avg | Alibaba | Amazon | Gap | What drives it |
|---|---|---|---|---|
| COGS | 60.8 | 66.8 | -6.0 | Marketplace take-rate vs 40% first-party units and owned fulfilment |
| R&D (Amazon: T&I) | 5.9 | 14.6 | -8.7 | Not comparable: Amazon line includes AWS infrastructure |
| SG&A | 20.9 | 8.9 | +12.0 | Alibaba buys traffic and quick-commerce frequency |
| Gross margin | 39.2 | 33.2 | +6.0 | Mirror of COGS |
| EBIT margin | 10.4 | 9.4 | +1.0 | Alibaba ahead on average, behind in the latest year (4.9 vs 11.2) |
| Year | COGS | R&D | SG&A | Gross margin | EBIT margin |
|---|---|---|---|---|---|
| Alibaba FY2024 | 62.3 | 5.6 | 16.7 | 37.7 | 12.0 |
| Alibaba FY2025 | 60.0 | 5.7 | 18.9 | 40.0 | 14.1 |
| Alibaba FY2026 | 60.2 | 6.5 | 27.2 | 39.8 | 4.9 |
| Amazon 2023 | 68.8 | 14.9 | 9.8 | 31.2 | 6.4 |
| Amazon 2024 | 66.6 | 13.9 | 8.7 | 33.4 | 10.8 |
| Amazon 2025 | 64.9 | 15.1 | 8.1 | 35.1 | 11.2 |
SG&A is the persistent gap: Alibaba is higher in all three years and the gap widened from 6.9 to 19.1 points, as Alibaba's sales and marketing rose from RMB115bn to RMB245bn while Amazon's stayed near $44 to 47bn on a revenue base growing 12% a year. The mechanism is the power map: Amazon's owned fulfilment and Prime retain demand, and its spend shows up as fulfilment and infrastructure, which are assets that also cut unit cost; Alibaba, without an owned instant-delivery network, pays in marketing and subsidies that leave no asset behind. The gap confirms the power map. The one contradiction is gross margin, where Alibaba leads by six points; that reflects a marketplace model, not pricing power, and it is being eroded from below the line rather than at it. A new cost discipline at Alibaba could close part of the gap once SAMR's subsidy rules bind, which is the main route to reversing this verdict.
| Metric | Threshold | By when | If it hits, it favors | Where published |
|---|---|---|---|---|
| AWS revenue growth and operating margin | Growth at or above 30% with margin at or above 35% | Q2 2027 (reported ~Jul 2027) | Amazon | Amazon quarterly release |
| Alibaba reported CMR growth | Back above +5% (like-for-like and reported) | Jun-2027 quarter | Alibaba | Alibaba quarterly release |
| Alibaba E-commerce Group adjusted EBITA | YoY growth positive for two straight quarters | Mar-2027 quarter | Alibaba | Alibaba quarterly release |
| Alibaba AI Cloud & Compute EBITA margin | At or above 15% with growth at or above 40% | Jun-2027 quarter | Alibaba | Alibaba quarterly release |
| Amazon trailing free cash flow | Positive again after the $220bn 2026 capex year | Q4 2027 | Amazon (negative with AWS growth below 25% favors neither) | Amazon quarterly release |