アマゾン：長期オーナーの視点から

優れた成長株の中から「10 年 5 倍」を探す——上振れ視点で問い詰める「もっと大きくなれるか？」

• QHow high is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?7/10

  A

  The ceiling is very high, but in essence it is "simultaneously expanding three existing pies" and creating incremental growth at their intersection, rather than carving out an entirely new market from nothing. Measured by Baillie Gifford's LTGG yardstick, Amazon's total addressable space is indeed among the largest in the world—but its height comes from the stacking of three mature arenas, not a single disruptive new category, so "high ceiling" must be distinguished from "the penetration curve is no longer wild."

  The report breaks this business precisely into three stacked industries: e-commerce retail, cloud infrastructure, and retail media advertising. All three are still growing, but all have passed their linear-expansion phase. On the e-commerce side, under the U.S. Census Bureau's measure, U.S. online retail was about $302.3 billion in Q1 2026, up about 9.7% year-over-year, accounting for 16.8% of total retail sales—still rising, but penetration has entered the middle stage and growth has fallen back to the single-digit to low-double-digit range. This is "expanding one's own share of an existing pie," not creating a new pie. The cloud side is the segment truly still growing fast: per Synergy Research Group, in Q1 2026 AWS held about 30% of the global cloud infrastructure market, still ranking first; AWS revenue grew 28% year-over-year in Amazon's official Q1 2026 results, reaching about $37.6 billion, an annualized run rate of about $150 billion—AI has pushed total cloud demand to a new plateau, and this part is closer to "expanding a pie that is itself swelling."

  The only segment that genuinely carries a "creating a new market" character is retail media advertising: it monetizes retail traffic and closed-loop data into a high-margin new category. The report discloses 2025 advertising revenue of $68.64 billion, while Amazon's official Q1 data shows Q1 2026 advertising already reached $17.2 billion, up 24% year-over-year, with TTM surpassing about $70 billion. Retail media is essentially a new monetization layer "grown" out of a closed-loop platform like Amazon, rather than taking share from an existing market—this is the part of its narrative most aligned with Baillie Gifford's preference for a "new market," but it still accounts for less than 10% of total revenue.

  To be honest, summing the three together, the height of Amazon's ceiling is beyond question—the total retail pool is in the ten-trillion-dollar range, global cloud is in the hundreds of billions and swelling, and retail media is just gaining momentum. But a "high" ceiling does not equal a "steep" slope: the bulk of its revenue (Online Stores + Physical Stores + Third-party Seller Services, totaling about $464 billion in 2025) has growth locked into the single-digit to low-double-digit range by mature e-commerce penetration. So the conclusion is: this is a composite that grafts low-return retail demand onto high-return cloud and advertising within a huge and still-growing market; it is an absolute winner at "expanding existing pies," and at "creating an entirely new market" only retail media stands as a truly valid branch.

  2026年6月11日
• QCan its revenue at least double over the next five years? Is growth driven mainly by volume, price, or new businesses?6/10

  A

  Doubling revenue over the next five years is highly difficult but not impossible—it requires about 15% compound annual growth, clearly higher than the actual trend in recent years, so it is not the base case. In terms of growth structure, it is driven mainly by "volume" (cloud compute consumption, the number of sellers and advertisers, Prime fulfillment scale), with "price" contributing only modestly in AWS / advertising / subscriptions, and new businesses (in-house chips, AI services) serving as an amplifier rather than an independent second pillar.

  First, let's nail down the hurdle: the report discloses 2025 revenue of $716.9 billion. Doubling to about $1.43 trillion over five years requires about 14.9% annualized compound growth. The historical trajectory the report itself provides is—about 19% compound over the eight years from 2017 to 2025, but only about 13% from 2020 to 2025, with the trend slowing. The latest Amazon Q1 2026 results show total revenue up 17% year-over-year, reaching $181.5 billion—if this growth rate can be sustained, five-year doubling is enough; but the 17% in Q1 includes a strong contribution from AWS's re-acceleration (up 28% year-over-year), and whether the whole can hold mid-double-digit growth for multiple years is precisely the uncertainty the report repeatedly emphasizes. It is worth noting that the report's internal valuation assumption for "growth over the next ten years" uses an 8%–12% owner-earnings growth rate, more conservative than revenue growth, which shows that even the report's authors do not treat "sustained 15% revenue growth" as a high-confidence premise.

  Breaking down the sources of growth, volume is the absolute mainstay:

  • Cloud (volume-led, price-assisted): AWS grew 28% year-over-year in Q1, an annualized run rate of about $150 billion (Amazon official), driven by a surge in compute consumption volume for AI training and inference, not by raising unit prices—in fact, cloud has a long-term price-cutting mindset. The roughly $244 billion AWS unfulfilled obligations with a weighted remaining 4.1 years disclosed by the report are the source of visibility into this "volume."
  • Advertising (both volume and price, steepest slope): advertising was $17.2 billion in Q1, up 24% year-over-year (Amazon Ads official measure), with both an increase in the number of advertisers and ad inventory (volume) and higher per-unit monetization from closed-loop data (price). This is the fastest-growing segment and the most efficient contributor to revenue doubling.
  • Retail (volume-led, slowest growth): dragged by mature e-commerce penetration, the growth of Online Stores disclosed by the report ($269.29 billion in 2025) has fallen back to single-digit to low-double-digit, relying only on faster Prime fulfillment and category expansion for marginal "volume" improvement, with almost no room to raise prices—the report explicitly says that on the retail side it more often passes efficiency gains to customers as discounts to maintain a low-price mindset.

  New businesses (in-house chips, AI platform services) are currently an amplifier rather than an independent second pillar: they amplify the cloud's "volume" by lowering AWS unit costs and increasing capacity to absorb AI workloads, rather than independently contributing a revenue curve that can stand alongside the three major segments.

  Conclusion: doubling revenue over five years requires AWS sustaining high growth + advertising continuing to grow fast + retail not stalling, all stacked together. It is an "optimistic but requires effort to deliver" scenario, not one that can be assumed by default; the main engine of growth is volume (cloud compute consumption and monetization of advertising inventory), while price's contribution is concentrated and limited.

  2026年6月11日
• QFive years from now, what will take over as the next growth engine? Does this "second curve" exist today?6/10

  A

  Amazon's "second curve" not only exists today but is already the main engine of current profit—but its "next baton" is different: the one taking over is not a new business still being incubated, but the already-grown AWS / advertising continuing to deliver deeper into AI and retail media. The genuinely not-yet-formed "third curve" that needs time to validate is AWS's internal in-house chips and AI platform layer.

  This differs from most companies' paradigm of "relying on a new story to take over." Amazon's distinctiveness lies in: its second curve (cloud + advertising) has long moved from incubation to harvest. The report notes that AWS accounts for only about 18% of revenue yet contributes about 57% of operating profit, and advertising already reached $68.64 billion in 2025—these two are the "second curve" that took over the retail baton over the past decade, and they are still accelerating now: Amazon's Q1 2026 results show AWS up 28% year-over-year, the fastest in 15 quarters, and advertising up 24% year-over-year to $17.2 billion (Amazon Ads official). So asking "what takes over in five years," for Amazon the more accurate framing is: the second curve has not yet peaked, and the real question is whether it can transition smoothly to the extension curve of "AI-driven cloud + retail media."

  There are three candidates for the "third curve" that can be seen today but have not yet delivered returns, all traceable in the report:

  • AWS in-house chips / AI infrastructure layer: the report repeatedly emphasizes that net capital expenditure surged to $128.3 billion in 2025, cash capital expenditure was about $43.2 billion in Q1 2026, and full-year guidance is on the order of about $200 billion (investor relations disclosure), with the core being to build the compute foundation for AI training / inference. This is the heaviest bet and the "greatest uncertainty" the report identifies—whether it can replicate AWS's historically high returns determines whether the third curve holds. The report honestly lists this as "to be determined," without advancing credit.
  • Deepening of retail media advertising: advertising TTM has already broken about $70 billion, expanding from search advertising into streaming (Prime Video advertising) and brand display—the most margin-friendly and highest-certainty candidate to take over.
  • AI investments such as Anthropic: the report discloses that Q1 2026 net profit included a $16.8 billion pre-tax gain from the Anthropic investment—but this is a change in investment fair value, a one-time gain or loss, and the report explicitly warns it "should not be simply treated as sustainable operating profit," so it is a strategic positioning, not operating revenue that can be counted into the second curve.

  An honest judgment: Amazon does not face the anxiety of a "successor that only vaguely appears five years from now" vacuum—its second curve (cloud + advertising) is alive and accelerating, with the deepest moat. Its real risk is not "whether there is a next baton," but "the next baton (the AI-ified cloud) must first swallow enormous capital expenditure"—the report discloses that Q1 2026 TTM free cash flow was squeezed down to just $1.2 billion. In other words, the second curve is very clear, and the seeds of the third curve (in-house chips / AI platform) are indeed already in the ground today, but the harvest must wait until the AI capital return is settled—which is precisely the core reason the report gives a "Watch" rather than a "Buy" rating.

  2026年6月11日
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?7/10

  A

  The core competitive advantage is a multi-layered flywheel of "retail traffic + seller network + Prime membership + closed-loop data + global fulfillment + public cloud," which almost no one can fully replicate; over the next three to five years the moat overall remains deep, but internally it diverges—the platform and fulfillment moats are widening, while AWS's relative-share moat is under pressure. This is a world-class moat, but not one that is "never disrupted."

  The report breaks the moat into ten items rated one by one, with the strongest being scale advantage, channel advantage, data advantage, and network effects. The quantitative support is solid: 2025 revenue of $716.9 billion, net property and equipment of $357 billion, and net capital expenditure of $128.3 billion that year—the report argues from this that to replicate the combination of "retail traffic + membership + seller network + advertising + global cloud + fulfillment" typically requires years of time and sustained investment on the order of tens of billions of dollars or more, and depends more on long-term operational accumulation than on financing alone. This still holds from a forward-looking perspective: even with money, competitors cannot buy the fulfillment density and data closed loop that Amazon has accumulated over twenty years.

  Looking at the moat's direction across time periods, we must separate "what history has proven" from "what still has to be fought for in the future":

  • The widening parts (history and forward look both favorable): retail logistics, Prime membership, third-party sellers, and advertising's data closed loop. The evidence is that advertising—the business most dependent on the closed-loop data—is still accelerating: Amazon official shows Q1 2026 advertising of $17.2 billion, up 24% year-over-year, with TTM about $70 billion; retail also recorded an all-time-high overall operating margin of 13.1% in Q1 (Amazon results), showing platform monetization efficiency is still improving. The "once strong" of this part of the moat is being extended into "now stronger."
  • The part under pressure (this is the key forward-looking risk): AWS's relative-share moat. The report states bluntly that in the AI cycle Azure and Google are catching up faster, and AWS's relative advantage is no longer as comfortable as before. Forward-looking data confirms this tension: per Synergy Research Group, in Q1 2026 AWS held about 30%, Azure about 25%, and Google about 13%—AWS still ranks first, but Azure's close distance is the nearest in history. At the same time, AWS operating margin fell back from about 39.1% to 37.7% (Q1 2026 measure), reflecting that early AI infrastructure depreciation is already eating into profit. So AWS's moat is not narrowing, but rather "the absolute advantage remains while relative comfort declines."

  To be honest, Baillie Gifford prefers moats that are "certainly widening"; Amazon's situation is more subtle—its moat depth is beyond question (the report gives 4.5/5), but whether it is "comprehensively widening" deserves a question mark. Over the forward-looking three to five years, the platform / fulfillment / advertising / data line will most likely keep widening, while the AWS line depends on whether it can hold share and margin in the AI era through in-house chips and engineering capability. The report's conclusion—"the platform and fulfillment moats are widening, the cloud relative-share moat is under pressure, but the overall moat remains very deep"—is a restrained and accurate judgment that does not gloss over AWS's pressure in order to fit a growth narrative.

  2026年6月11日
• QIf its core business were disrupted, does it have the gene to reinvent itself? How does it deal with mistakes and bad news?6/10

  A

  Amazon is one of the few companies repeatedly proven to have the "self-reinvention gene"—its history is itself a regeneration story of continually leaping from retail to higher dimensions; facing mistakes and bad news, management's disclosure is relatively candid and does not rely on whitewashing. This is precisely the implicit premise Baillie Gifford values most in its chain of follow-up questions—"whether it can reinvent itself when the core business is disrupted"—and Amazon scores very high on this point.

  The self-reinvention gene has a solid historical track record, not a slogan. From an online bookstore, Amazon successively grew a third-party platform (Marketplace), Prime membership, a self-built fulfillment network, AWS cloud computing, and retail media advertising—each time proactively opening up a higher-return new dimension before the existing business had peaked. The report confirms the fruits of this regeneration with the profit structure: AWS accounts for only about 18% of revenue yet contributes about 57% of operating profit, and the overall operating margin improved from 2.3% in 2017 all the way to 11.2% in 2025, setting an all-time high of 13.1% in Q1 2026—this is a structural metamorphosis "from commodity-retail-driven to service- and platform-driven," precisely proving its ability to graft a low-return core business onto high-return new engines. The current round of AI capital expenditure (about $43.2 billion cash capex in Q1, full-year guidance on the order of about $200 billion) is essentially another regeneration attempt of "proactively betting heavily before cloud is reshaped by AI"—success is to be determined, but the gene of "daring to bet on self-reinvention" is clearly visible.

  On how it deals with mistakes and bad news, the report gives a relatively high rating, with specific examples. The report explicitly states: the company directly discloses which parts of current-period profit come from the FTC settlement, layoffs, tax disputes, or investment gains, and also states plainly that the decline in free cash flow is mainly due to AI investment—this disclosure culture of "proactively laying out bad news and one-time factors" is not common among large tech companies. Specific examples include: the report discloses that Q1 2026 net profit included a $16.8 billion pre-tax gain from the Anthropic investment, and the company did not package it as operational improvement, with the report accordingly reminding that it "should not be simply treated as sustainable operating profit"; the company also disclosed that in 2025, changes in accounting estimates such as the useful life of servers / network equipment increased depreciation and amortization by $1.4 billion and reduced net profit by $1 billion—this practice of proactively revealing "unfavorable accounting changes" is a positive signal of honesty. The report also notes the company disclosed that internal controls were effective in 2025 and the auditor issued an unqualified opinion, with no obvious direct signals of financial fraud or aggressive accounting.

  The honest boundary: the report does not laud this item to the skies either. It points out that whether the acquisitions (One Medical, MGM) created value cannot be concluded with high confidence from the inadequate public segment disclosure, so it holds a reserved attitude of "excellent history, future to be determined" toward management's capital allocation "future chapters" (especially AI investment returns); on governance, it also flags that the company has transactions with Bezos-related enterprises (such as the 2022 $2.7 billion satellite launch agreement with Blue Origin, overseen by an independent board committee), which is a point requiring ongoing observation.

  Conclusion: self-reinvention gene—strong, backed by twenty years of multiple successful leaps; dealing with mistakes and bad news—candid, with a disclosure culture above the peer average. This is one of the dimensions in which Amazon best withstands scrutiny under the Baillie Gifford framework, and the fundamental reason the report acknowledges its "world-class enterprise quality" even while assigning a "Watch" rating.

  2026年6月11日
• QDoes management (especially the founder) have a long-term vision and interests deeply aligned with the company? Are they willing to sacrifice current profit for five to ten years out?7/10

  A

  Management's long-term vision is extremely pronounced, and it is proving this with hard cash in the way of "sacrificing current free cash flow for cloud dominance ten years out"; on interest alignment, Bezos is highly aligned with shareholders, and while Jassy's personal stake is not high, his compensation structure is highly long-term-oriented. This is one of Amazon's strengths under the Baillie Gifford framework—"willingness to sacrifice current profit for five to ten years out" is almost this company's founding principle.

  "Willing to sacrifice current profit for the long term"—Amazon gives a textbook-level affirmative answer, and it is happening right now. The report discloses that 2025 net capital expenditure surged to $128.3 billion, squeezing free cash flow from $38.2 billion in 2024 down to about $11.2 billion; Q1 2026 went further—TTM free cash flow was squeezed down to just $1.2 billion, single-quarter cash capex was about $43.2 billion, and the company explicitly keeps full-year capital expenditure on the order of about $200 billion (Q1 2026 capex guidance), with management stating plainly that it is mainly directed at AI infrastructure. The report cuts to the point: this is the reality of "accounting profit < operating cash profit > distributable free cash flow"—the company would rather pay almost no cash to shareholders now and instead pour money into the AI compute foundation to bet on returns ten years out. What Baillie Gifford most admires—"sacrificing the present for five to ten years out"—has no more straightforward example than this. Of course, the report also honestly lays out the other side of this coin: whether this sacrifice can yield AWS-historic-level returns is the "greatest uncertainty," so "willingness to sacrifice" is a definite strength, while "whether the sacrifice is worth it" is the to-be-determined risk.

  On interest alignment, look at two people:

  • Jeff Bezos (Executive Chairman): the report discloses he holds about 950 million shares, a stake of about 8.8%—based on the current market cap of about $2.56 trillion (close on June 10, 2026, share price about $238), this equity is worth about $225 billion, one of the individual interests most deeply tied to a single company on Earth. The report's judgment that "Bezos is highly aligned with shareholders" is beyond dispute.
  • Andy Jassy (CEO): the report discloses a stake of about 2.31 million shares, a stake of less than 1%, with the absolute value of personal holdings not high. But the report also notes his compensation is primarily long-term RSUs, typically vesting over five years or more, with no annual cash bonus, and that the committee did not grant new equity from the time it granted the CEO the large related RSUs in 2021 through 2025—this structure of "a one-time long-cycle grant, not relying on annual cash incentives" is more long-term-oriented than most professional managers, tying the executive's wallet to the five-year performance of the stock price.

  Cultural evidence of long-term vision: the report notes the company has long emphasized customer obsession, long-term thinking, and operational excellence, and lists this as a "culture / operational capability—very strong" moat element. On capital allocation, the report makes clear: the company's focus in recent years has been internal reinvestment rather than buybacks and dividends—no buybacks and no dividends from 2023 to 2025, with $6.1 billion of buyback authorization remaining unused as of year-end 2025. The report interprets this as "Amazon is not a company that boosts per-share value through buybacks, but one that drives per-share intrinsic value through reinvestment."

  The honest reservation: the report's overall assessment of management is "high marks in history, future to be determined" (giving 4/5), with the deducted points being precisely that the current AI investment return is not yet settled; on governance, it also flags Bezos-related transactions (such as the 2022 $2.7 billion satellite launch agreement with Blue Origin, overseen by an independent board committee) as points requiring ongoing observation. But on the core of this Baillie Gifford question—"long-term vision + interest alignment + willingness to sacrifice current profit for the long term"—Amazon's answer is a clear and strong affirmative.

  2026年6月11日
• QIf it disappeared tomorrow, how much would customers miss it? Is its way of growing sustainable and not reliant on harming society and regulation?6/10

  A

  If Amazon disappeared tomorrow, everyone from consumers to sellers to global enterprise IT would "miss" it intensely—its indispensability is extremely strong; but the "social and regulatory sustainability" of its way of growing is a part with real flaws, and antitrust and platform-rule risk are precisely what the report lists as the core threats that could reshape the business model. This question must be viewed in two ways: indispensability (full-marks level) and sustainability (with reservations), the two pointing in opposite directions.

  "How much would it be missed"—the answer is almost irreplaceable, viewed across three types of customers:

  • Consumers: the Prime membership system + self-built fulfillment network have become "default infrastructure." The report lists this as "brand advantage—strong" (Amazon, Prime, AWS have become one of the default options for consumers and enterprises) and "channel advantage—very strong." Lose it, and hundreds of millions of Prime users' next-day / same-day fulfillment capability cannot be filled by anyone in the short term.
  • Third-party sellers: the report discloses 2025 third-party seller services revenue of $172.16 billion, behind which are millions of sellers who survive by depending on Amazon's traffic, fulfillment (FBA), and ad inventory—for them, Amazon is not one channel among many, but their livelihood itself.
  • Enterprises / developers: AWS Q1 revenue was about $37.6 billion, an annualized run rate of about $150 billion (Amazon official), carrying a massive amount of the world's mission-critical workloads. The report discloses about $244 billion of AWS unfulfilled obligations with a weighted remaining 4.1 years, as well as high switching costs of "significant migration friction in cloud architecture, data, and toolchains"—meaning enterprises not only cannot leave it, but the cost of migrating away is also extremely high.

  The report's overall judgment on this is: at the intersection of e-commerce infrastructure, retail media, and public cloud, "there is almost no fully replaceable rival." So on the Baillie Gifford implicit premise of "indispensability," Amazon earns close to full marks.

  But on "whether the way of growing is sustainable and does not rely on harming society and regulation"—here we must honestly discount, and this is Amazon's most concrete soft spot. The report lists regulation / antitrust as one of the three most critical uncertainties, and specifically names the risk surfaces: antitrust, platform rules, Featured Offer ranking, seller data use, and the Prime structure could all be restricted. This is not an abstract worry—it points directly to the fact that part of Amazon's growth dividend comes from the structure of "the platform being both referee and player": using seller data to optimize its own offerings, using ranking weights to influence the Featured Offer, and using Prime bundling to strengthen lock-in. The report also discloses that the company has been involved in an FTC settlement in recent years (recorded in current-period gains and losses), showing that regulatory friction is already a real cost rather than a distant assumption. In other words, within Amazon's flywheel, part of the monetization efficiency is built on mechanisms that "could be deemed by regulators to harm competition"—once Marketplace ranking, seller data use, or key Prime terms are forced to change, the report explicitly warns it "would reshape Marketplace and advertising monetization logic," which is one source of valuation fragility.

  The spirit of this Baillie Gifford question is: a truly great growth stock's growth should make the world better and thus be sustainable, rather than exchanging harm to society or crossing regulatory red lines for it. On the side of "benefiting consumers and sellers" (faster, cheaper, more choices), Amazon does create real social value, but on the side of "whether platform power is excessively used for self-preferencing," there is a real possibility of regulatory tightening. So the conclusion is: indispensability—extremely strong, nearly irreplaceable; social and regulatory sustainability—overall sustainable but with clear flaws, antitrust being a sword hanging overhead. The report does not avoid this point, but rather lists it among the "signals that trigger reassessment" for ongoing tracking—this is responsible handling, and it also reminds investors: part of Amazon's moat depends on the premise that "regulation does not go badly wrong" holding true.

  2026年6月11日
• QWhat is the unit economics of this business (gross margin, incremental returns)? Does it get better or worse as scale grows? Where does the money earned go?6/10

  A

  Amazon's unit economics improve significantly with scale—but this is the improvement of a "hybrid": the low-margin retail pool has limited marginal improvement, and what truly lifts the overall unit economics is the rising share of high-margin AWS, advertising, and subscriptions; the money earned is almost entirely reinvested into AI compute and fulfillment infrastructure, rather than dividends and buybacks. Scale makes it better, but the current answer to "where did the money earned go" is—swallowed by AI capital expenditure.

  Gross margin and incremental returns improve with scale, with solid evidence. The report discloses that 2025 cost of sales was 49.7% of net sales, i.e., gross margin of about 50.3%; whereas in 2020 this ratio was still 60.4% (gross margin about 39.6%), 58.0% in 2021, and 56.2% in 2022—meaning gross margin improved by over ten percentage points over five years. The improvement in operating margin is steeper: rising from 2.3% in 2017 all the way to 11.2% in 2025, and setting an all-time high of 13.1% in Q1 2026. The report cuts to the point: this is not the trajectory commonly seen in the retail industry, but the result of a rising share of high-margin service revenue—essentially "scale + business-mix upgrade" jointly improving the unit economics.

  But to break it down honestly, the unit-economics improvement brought by scale is highly uneven:

  • AWS (best unit economics, and still at a high level): AWS Q1 operating profit was about $14.2 billion, with operating margin of about 37.7% (Q1 2026 measure)—a typical high-incremental-return business, where each additional unit of compute consumption brings high marginal profit. The report discloses that AWS accounts for only about 18% of revenue yet contributes about 57% of operating profit, ironclad proof of the disparity in unit economics. It is worth noting that AWS margin has slightly declined from about 39.1% to 37.7%, reflecting that early AI infrastructure depreciation is beginning to eat into profit—the unit economics remain excellent, but no longer move up one-sidedly.
  • Advertising (excellent unit economics): advertising was $17.2 billion in Q1, up 24% year-over-year (Amazon Ads official); retail media is a high-margin monetization layer driven by closed-loop data, almost pure incremental profit.
  • Retail (weakest unit economics, limited scale improvement): the report discloses that 2025 fulfillment was 15.2% of net sales and shipping costs were $102.7 billion, with the International segment's 2025 operating profit only $4.75 billion and a thin margin—even with enormous scale, retail's unit-economics improvement relies mainly on fulfillment density and seller-services penetration, with a ceiling far below AWS / advertising.

  "Where does the money earned go"—this is the current crux the report most emphasizes. The answer is extremely clear: reinvested, with almost nothing left for shareholders. The report discloses 2025 net capital expenditure of $128.3 billion, Q1 2026 single-quarter cash capital expenditure of about $43.2 billion, and full-year guidance on the order of about $200 billion, mainly directed at AI infrastructure; meanwhile no buybacks and no dividends from 2023 to 2025. The result is what the report repeatedly emphasizes: 2025 operating cash flow was as high as $139.5 billion, but free cash flow was only about $11.2 billion; Q1 TTM free cash flow was even squeezed down to just $1.2 billion. The report's precise characterization is—Amazon is currently a "high cash creation + high cash re-swallowing" enterprise.

  Conclusion: unit economics improve with scale (gross margin +10pct over five years, operating margin setting a 13.1% high), but the engine of the improvement is AWS and advertising rather than retail itself; the money earned is currently almost entirely fed to AI compute and fulfillment infrastructure. Baillie Gifford would like the story of "money reinvested into high-return production"—but the report honestly pins its judgment on one sentence: whether the return rate on this money can replicate AWS's history, no one can conclude today, which is precisely the financial foundation of the "Watch" rating.

  2026年6月11日
• QWhat conditions need to hold simultaneously for it to rise fivefold in ten years? Are these conditions realistic? What expectations does today's stock price imply?3/10

  A

  For Amazon to rise fivefold in ten years, four highly difficult conditions need to "hold simultaneously"—this combination is not impossible, but the probability is low, belonging to the optimistic end of the optimistic scenario; and today's stock price of about $238 already implies the strong expectation that "AI capital expenditure will eventually deliver AWS-historic-level returns," leaving investors with insufficient margin of safety. This is the core arithmetic behind the report's "Watch" rather than "Buy."

  First, quantify the hard hurdle of "fivefold." Calculating a ten-year fivefold from the current price requires about 17.5% annualized price return (5^(1/10)≈1.175). Yet the report's own projection of long-term annualized return from buying at the current price is: optimistic scenario only about 8%–11%/year, neutral about 3%–6%/year, conservative about -3% to 0%/year. In other words, even under the report's most optimistic scenario, the annualized return is only 8%–11%, far below the 17.5% needed for fivefold. The report's three-tier intrinsic value estimates equally illustrate the problem: conservative about $99/share, neutral about $174/share, optimistic about $285/share—the current about $238 (close on June 10, 2026) is already significantly above neutral and approaching the optimistic valuation. This means: buying at the current price for a ten-year fivefold requires not just "the company being excellent," but the stacking of two difficult things—"the company continually delivering beyond optimistic assumptions + the valuation multiple not contracting."

  "What conditions need to hold simultaneously"—the report has listed them, each a high hurdle:

  1. AWS maintaining high growth and high margins in the AI era: AWS grew 28% year-over-year in Q1, the fastest in 15 quarters (Amazon official), but margin has slightly declined from about 39.1% to 37.7% (Q1 2026 measure), and faces Azure's close pursuit (Synergy measure about 25%)—maintaining this for ten years requires continually winning the share war.
  2. AI capital expenditure on the order of about $200 billion/year ultimately proving to be high-return: this is the "most fragile assumption" the report names. The report discloses 2025 net capital expenditure of $128.3 billion, squeezing free cash flow down to about $11.2 billion, with Q1 TTM free cash flow only $1.2 billion—if this money returns ordinarily, Amazon will degrade from a "high-return compounding machine" into a "high-growth but capital-intensive platform company."
  3. Advertising, seller services, and subscriptions continually pushing up the overall margin: advertising was $17.2 billion in Q1, up 24% year-over-year (Amazon Ads), delivering well—the condition with the strongest evidence among the four.
  4. Regulation not reshaping the Marketplace / advertising / Prime monetization logic: the report lists it as one of the three uncertainties, exogenous and uncontrollable.

  The report's judgment is restrained: "as long as one or two of these are clearly discounted, the current valuation becomes fragile"—only if all four beat expectations and hold simultaneously is a ten-year fivefold possible, and this joint probability is low.

  "What expectations does today's stock price imply"—this is the core of the question. The report calculates it thoroughly: the current market cap of about $2.56 trillion (at the current price) corresponds to a static P/E of about 28–29x (stockanalysis measure, down from the report's 31.7x on May 19 because the share price fell back from $264.86 to about $238), and against 2025 free cash flow of about $11.2 billion it is even a P/FCF as high as about 200x or more. The report pinpoints what the market is paying for: "the market is already paying for future cash flows, not for current cash flows"—that is, the expectation implied by the stock price is "the free cash flow swallowed by AI today will strongly recover in the future due to high returns." The report's core thesis in one sentence: the current price is closer to "paying in advance for future excellence" than to "buying at a price that is also a good deal under conservative assumptions."

  An update reminder favorable to the report: the report did its arithmetic based on May 19's $264.86, while the current stock price has fallen back about 10% to about $238—this slightly improves the margin of safety (the current price has risen above the upper end of the report's "acceptable holding price range of 155–220," but is closer than $264.86), but it is still above the report's neutral intrinsic value (about $174) and the "ideal buy range of 120–155." The conclusion is unchanged: a ten-year fivefold requires four highly difficult conditions to hold simultaneously, with low joint probability; today's price implies a strong expectation of AI capital returns being delivered, with insufficient margin of safety. The report honestly does not inflate for a growth narrative—it acknowledges Amazon is a great company, but insists "at the current price you buy expensive excellence, not cheap greatness."

  2026年6月11日
• QWhy hasn't the market realized all this yet? Is it that it can't understand, looks down on it, or can't see far enough? What would become the "narrative inflection point"?3/10

  A

  For a super-heavyweight stock like Amazon, the most fully covered and most thoroughly researched globally, the premise that "the market hasn't realized" is largely invalid—it is neither misunderstood nor looked down upon; if anything, the market may "see too far and have already priced it in ahead of time." The real cognitive disagreement is not "the market underestimates it," but "whether the optimistic expectation the market is paying for AI capital returns is correct"; the narrative inflection point will be triggered by evidence of returns on AI capital expenditure.

  First, answer this question honestly in reverse. The premise of this Baillie Gifford question is "I discovered a greatness the market hasn't realized"—but Amazon does not fit this premise: it is a target with a market cap of about $2.56 trillion (June 10, 2026), covered by every institution in the world, with no cognitive gap of being "obscure or overlooked." The report's stance is exactly the opposite—it holds that the market is not underestimating, but has already fully or even slightly excessively priced in "future excellence." So checking item by item against "can't understand / looks down on it / can't see far":

  • Can't understand?—basically invalid. The report acknowledges Amazon "is not a simple business," with quite a bit of accounting noise (investment fair value, depreciation lifespans, one-time settlements disturbing profit), and gives business understandability only 4/5. There is indeed a "comprehension threshold caused by complexity" here—for example, the $30.3 billion Q1 net profit includes $16.8 billion of Anthropic investment gains (Q1 2026 results), and not breaking it apart would overestimate operating quality. But this complexity causes "short-term misreading," not "systematic failure to understand," and professional institutions have long seen through it.
  • Looks down on it?—invalid. The market gives AWS and advertising a full high-quality premium and does not treat it as a low-quality retailer.
  • Can't see far?—the direction is reversed. The report's judgment is that the market "sees too far": the current P/FCF of over about 200x (against 2025 free cash flow of about $11.2 billion) shows the market has already skipped the present and is paying for the cash flow recovery five to ten years out. This is not "failing to see far," but "discounting far-future optimistic assumptions into today's price ahead of time."

  So where is the real "cognitive gap"? Not in "the market hasn't discovered Amazon's greatness," but in a judgment disagreement that has no conclusion yet—whether AI capital expenditure on the order of about $200 billion/year (2026 guidance, Q1 capex measure) can ultimately replicate AWS's historically high returns. Bulls believe it can and have already paid a premium for it; cautious parties like the report believe "to be determined, the current price leaves no room for error." This is a genuine uncertainty about the future, not a market blind spot—both sides see the same facts, and the disagreement is in pricing the probability of returns.

  What would the narrative inflection point be? The report has actually listed bidirectional trigger signals, and the inflection point is hidden in this evidence:

  • Upside inflection point: AI investment clearly converting into higher AWS growth and stronger operating cash flow 2–3 years out, AWS share stabilizing or even rebounding with margins holding high, and free cash flow rapidly recovering even amid high capital expenditure. Once these data appear, the market will switch from "paying for expectations" to "expectations delivered," and the uncertainty discount on the valuation will disappear. AWS re-accelerating to 28% in Q1 (Amazon official) is already an early positive signal.
  • Downside inflection point: AWS share continually declining with margins falling below about 30%, capital expenditure continuing to surge but operating cash flow and backlog not rising in tandem 2–3 years out, and regulation substantively weakening Marketplace/Prime/advertising monetization. If any one appears, the market will re-price Amazon as a "high-quality but more capital-heavy" platform company, and the report warns it may be accompanied by a 35%–55% price drawdown.

  Conclusion: for Amazon, "why hasn't the market realized" is a not-very-applicable question—it is not a dust-covered pearl, but a giant that is fully priced, even optimistically priced. The report's honesty lies in: it does not hard-code a story of "the market can't understand so it's undervalued" to support a buy, but says plainly the current price already implies strong expectations and the margin of safety is insufficient. What is truly worth tracking is not "when the market discovers it," but "when the exam paper of AI capital returns gets its score"—that is the narrative inflection point that determines the stock price's direction.

  2026年6月11日