The Case for Amazon

The Setup

Amazon (NASDAQ: AMZN) reported $717 billion in revenue for 2025, a figure so large it is on the verge of dethroning Walmart as the highest-revenue company in the world. Operating income reached $80 billion. Operating cash flow hit $139.5 billion, up 20% year-over-year. And yet the stock returned only about 6% in 2025 while the broader market surged, largely because investors fixated on one number: $200 billion in planned 2026 capital expenditures.

The market looked at $200 billion in capex and saw risk. The data suggests something different: a company with so many growth vectors, each backed by real revenue and real demand signals, that it is investing across cloud infrastructure, AI chips, satellite internet, robotics, autonomous vehicles, healthcare, and grocery simultaneously, and generating enough cash flow to fund all of it.

This is not a company that does one thing well. Amazon is a collection of businesses, several of which would be among the most valuable companies in the world as standalone entities. The thesis is not about any single business unit. It is about the compounding effect of all of them operating together, feeding into the flywheel that has defined Amazon for 30 years, and the fact that the current valuation does not appear to reflect the full picture.

AWS: The Profit Engine That Keeps Accelerating

Amazon Web Services generated $128.7 billion in revenue for 2025, up 20% year-over-year. But the headline number understates the momentum. AWS growth accelerated throughout the year: 17% in Q1, 17.5% in Q2, 20% in Q3, and 24% in Q4. That Q4 number was the fastest growth rate in 13 quarters. By year-end, the annualized revenue run rate exceeded $142 billion.

The scale of AWS is worth pausing on. In Q4 alone, AWS added $2.6 billion in sequential revenue. It now generates more incremental quarterly revenue than Google Cloud, despite operating on a base that is roughly double the size. CEO Andy Jassy was direct about this on the earnings call: growing 24% on a $142 billion run rate is fundamentally different than growing at a higher percentage on a meaningfully smaller base.

AWS contributed 57% of Amazon's total operating income in 2025, with operating margins of 35%. That means the cloud business alone generated approximately $45 billion in operating income for the year. For context, that is more operating income than Coca-Cola, Nike, McDonald's, and Goldman Sachs earn individually.

The backlog tells the forward story. AWS disclosed $244 billion in committed backlog at year-end, growing 40% year-over-year. This backlog represents multi-year agreements from enterprise and government customers, including OpenAI, Visa, the NBA, BlackRock, Salesforce, the U.S. Air Force, Adobe, Thomson Reuters, AT&T, and United Airlines. These are not speculative commitments. They are contracted revenue that will convert over the coming years.

Amazon's custom chip business adds another dimension. The company's Graviton (compute), Trainium (AI training), and Nitro (networking) chips have reached a combined annual revenue run rate exceeding $20 billion, growing at triple-digit percentages year-over-year. Management suggested that if structured as a standalone chip vendor, the implied 2026 production run rate could approach $50 billion. Amazon is building its own silicon supply chain inside its own cloud, reducing dependency on third-party chip suppliers while improving margins.

AWS is not just Amazon's profit engine. It is one of the most valuable businesses on the planet, and it is growing faster today than it was a year ago.

The $68 Billion Advertising Business Nobody Saw Coming

Amazon's advertising segment generated approximately $68 billion in revenue in 2025, growing 22% year-over-year in Q4. The company added more than $12 billion in incremental advertising revenue in a single year. Five years ago, this business barely registered in earnings reports. Today it rivals YouTube's advertising revenue and operates at margins that are among the highest in the company.

The structural advantage is purchase intent. When someone searches on Google, they might be researching. When someone searches on Amazon, they are usually ready to buy. That difference in intent makes Amazon's advertising platform extraordinarily valuable to brands: the conversion rate on Amazon ads is meaningfully higher than on general search or social platforms because the user is already in a purchasing context.

The addition of Prime Video advertising in early 2024 expanded the opportunity further. Amazon now serves ads to an average of 315 million viewers globally across 16 countries through its streaming platform. Unlike traditional streaming competitors that are struggling to build ad businesses from scratch, Amazon had the ad tech infrastructure, the advertiser relationships, and the first-party purchase data already in place. Prime Video ads were revenue-additive from day one.

The advertising business is important to the investment thesis because it is almost entirely high-margin incremental revenue. The content, the eyeballs, and the platform already exist. Every advertising dollar is layered on top of existing infrastructure at margins that are significantly higher than retail. This business alone, valued on its own, would be one of the ten most valuable media companies in the world.

A Million Robots and Counting

In mid-2025, Amazon confirmed that it had surpassed one million robots deployed across its global operations network. That number has grown from approximately 200,000 in 2019 and 520,000 in 2022. Amazon now operates nearly as many robots as it does human employees, making it the largest operator of mobile robotics in the world.

The impact on operations is measurable. Amazon's most automated fulfillment center, SHV1 in Shreveport, Louisiana, operates with approximately 1,133 employees. Comparable traditional fulfillment centers require up to 3,000 workers. Pick rates jumped from roughly 100 items per hour to 300 to 400 per hour after the introduction of Kiva-based robotic systems. The Shreveport facility uses eight different robotic systems working in concert, including AI-powered inventory management (Sequoia), robotic arms with tactile sensing (Vulcan), and autonomous mobile robots (Proteus) that navigate the warehouse floor without cages or barriers.

The March 2026 acquisition of Rivr, a startup that builds autonomous delivery robots capable of climbing stairs, signals that Amazon's automation ambitions extend well beyond the warehouse. Combined with its Prime Air drone delivery program (now approved for beyond-visual-line-of-sight operations in the U.K.) and the Zoox autonomous vehicle subsidiary, Amazon is methodically building an end-to-end logistics chain where automation handles the package from warehouse shelf to front door.

Jassy's 2025 shareholder letter included a notable signal about the future: wherever Amazon can leverage the scale and real-time feedback loop from its million-robot fleet to build robotics solutions for other industrial and consumer customers, it will explore doing so. That language suggests Amazon may eventually sell its robotics capabilities externally, the same playbook it used to turn its internal cloud infrastructure into AWS.

Zoox: The Quiet Robotaxi Play

While Waymo (owned by Alphabet) dominates the autonomous driving conversation, Amazon's Zoox subsidiary has been quietly building toward commercial deployment with a fundamentally different approach: a purpose-built, bidirectional robotaxi designed from scratch for autonomous operation, with no steering wheel, no pedals, and seating that faces inward.

In June 2025, Zoox opened its first full-scale production facility in Hayward, California. The 220,000-square-foot factory is designed to eventually produce 10,000 robotaxis per year. Zoox launched its ride-hailing service in Las Vegas in September 2025, initially as a free early-access program, and is planning paid public rides in Las Vegas in 2026 with a San Francisco expansion to follow.

Zoox is far behind Waymo in terms of scale. With approximately 50 robotaxis in service compared to Waymo's 3,000, and no external valuation event comparable to Waymo's $126 billion fundraise, the market is not assigning meaningful value to Zoox within Amazon's stock price. That may be appropriate given the early stage of commercialization. But it also means that any acceleration in Zoox's deployment timeline represents pure upside optionality for Amazon shareholders.

The strategic logic is also different from Waymo's. For Alphabet, Waymo is primarily a transportation play. For Amazon, autonomous vehicles could eventually transform its entire logistics network, from last-mile delivery to long-haul trucking. A self-driving fleet that can move packages 24 hours a day without driver wages, rest stops, or turnover would fundamentally alter the cost structure of the $200 billion Amazon spends annually on fulfillment and shipping.

Amazon Leo: Internet From Space

Amazon Leo (formerly Project Kuiper) entered enterprise beta on April 8, 2026, with commercial availability targeted for mid-2026. The low-earth-orbit satellite internet service has approximately 241 satellites in orbit, with production facilities in Kirkland, Washington capable of building five satellites per day. The company has contracted 22 additional launches across United Launch Alliance, Arianespace, and Blue Origin rockets to build out the constellation.

The enterprise beta partner list is notable: Verizon, AT&T, Vodafone, JetBlue, NASA, and NBN Co (Australia), among others. These are not speculative partnerships. They represent revenue commitments from enterprises and governments. Andy Jassy confirmed the mid-2026 commercial launch in his most recent shareholder letter, positioning Leo alongside the $50 billion Trainium chip investment as one of Amazon's defining capital allocation bets.

Amazon Leo is competing directly with SpaceX's Starlink, which generated $10.6 billion in revenue in 2025 with more than 10 million paying subscribers. Amazon is behind on deployment (241 satellites vs. Starlink's 7,600+), but it has a structural advantage that Starlink does not: deep integration with AWS. For enterprise customers, the ability to move data from a satellite connection directly into AWS for storage, processing, and AI analytics creates a bundled value proposition that a standalone satellite provider cannot match.

Internal projections suggest Amazon Leo could generate $20 billion in annual revenue by 2030. Even if that target is optimistic, the satellite internet market is growing rapidly as demand for global connectivity outpaces terrestrial infrastructure deployment, particularly in rural and underserved areas. Amazon has already committed $4 billion to expanding its rural delivery network; satellite internet is the connectivity layer that makes rural customers as accessible as urban ones.

The Anthropic Stake: $8 Billion In, Multiples Out

Amazon has invested approximately $8 billion in Anthropic, the AI company behind the Claude assistant, making AWS Anthropic's primary cloud provider for mission-critical workloads. As of February 2026, Anthropic is valued at $380 billion following a $30 billion Series G round. Anthropic's annualized revenue run rate reached $19 billion by March 2026, up from $1 billion just 15 months earlier.

Amazon's Anthropic investment serves a dual strategic purpose. First, the equity appreciation is substantial: the $8 billion investment is now worth significantly more at Anthropic's current valuation. Amazon recorded a $9.5 billion pretax gain from the investment in Q3 2025 alone when Anthropic's valuation surged. Second, the partnership makes AWS the default infrastructure provider for one of the fastest-growing AI companies in the world, driving cloud revenue that compounds alongside Anthropic's growth.

With Anthropic reportedly targeting an IPO in the second half of 2026 at a valuation of $400 billion to $500 billion, Amazon's stake could become a material asset on the balance sheet in its own right.

The Retail Flywheel Still Spins

It is easy to overlook the retail business when AWS, advertising, robotics, and satellites are dominating the narrative. But Amazon's retail operation approaching $600 billion in annual revenue remains the foundation of the entire flywheel.

Prime membership now exceeds 240 million subscribers globally. Prime members spend roughly twice as much as non-Prime customers. Prime Day 2025 generated $24.1 billion in U.S. sales over four days, up 30% year-over-year and more than double Black Friday 2024. Worldwide paid units grew 12% year-over-year in Q4 2025, the highest quarterly growth rate of the year.

Amazon is also becoming a grocery destination for over 150 million Americans, with everyday essentials now representing one in three Amazon purchases. The healthcare expansion through One Medical and Amazon Pharmacy, integrated into Prime membership, positions Amazon to capture a share of the $4.5 trillion U.S. healthcare market over time.

The retail business matters to the thesis not because it is growing at 40%, but because it is the surface area that generates the data, the customer relationships, and the logistics infrastructure that make every other business segment more valuable. Advertising is high-margin because retail creates the purchase intent. AWS grew from internal infrastructure. Robotics evolved from fulfillment needs. The flywheel is not a metaphor. It is the actual operating model.

Valuation: Cheaper Than You Think

At a stock price around $250, Amazon trades at approximately 35x trailing earnings and 31x forward earnings. Those numbers might look expensive relative to the broader market, but they represent the cheapest Amazon has been by its own historical standards in over a decade. The 10-year median P/E for Amazon is approximately 81x. The 5-year average is 48x. The 3-year average is 47x. At 35x trailing, the stock is trading at less than half its historical average multiple.

The reason for the compression is that Amazon's profitability has improved dramatically. Operating income went from $37 billion in 2023 to $68 billion in 2024 to $80 billion in 2025. Net income reached $77.7 billion. These are not temporary gains from cost cuts; they reflect structural margin expansion driven by advertising growth, AWS operating leverage, and fulfillment network regionalization.

The PEG ratio of 1.9 suggests the stock is reasonably priced relative to its expected growth rate. Analysts project AWS revenue growth of 26% in 2026 and 29% in 2027 as committed backlog converts to recognized revenue. The average analyst price target is approximately $282, implying roughly 13% upside from current levels, with a consensus rating of Strong Buy (64 buy/strong buy ratings, 4 holds, zero sells).

A sum-of-the-parts analysis underscores the potential disconnect. AWS alone, valued at a reasonable cloud multiple on its operating income, would account for a significant portion of Amazon's entire market capitalization. Layer on the advertising business, the Anthropic stake, the Zoox optionality, Amazon Leo, and the retail flywheel, and the sum comfortably exceeds the current $2.7 trillion market cap. The market appears to be pricing the core retail and AWS businesses while assigning little value to the emerging segments that could each become multi-billion-dollar revenue contributors.

What Could Go Wrong

Capital expenditure risk. The $200 billion in planned 2026 capex is unprecedented for any company in history. If AI demand slows, if AWS backlog conversion disappoints, or if Amazon Leo faces deployment delays, the return on this investment could fall short of expectations. The counter-argument is that $244 billion in AWS backlog and accelerating growth suggest this spending is demand-driven, not speculative, but the sheer magnitude of the bet warrants caution.

FTC antitrust trial. Amazon faces an FTC antitrust trial scheduled for October 2026 that could impact marketplace practices. A severe ruling could theoretically force changes to how Amazon prioritizes its own products or even require divestiture of its logistics arm. While the most extreme outcomes are considered unlikely, regulatory uncertainty is a legitimate overhang.

Cloud competition. AWS remains the market leader, but Google Cloud grew 48% in Q4 2025 (vs. AWS's 24%) and Microsoft Azure grew 39%. Both competitors are investing aggressively in AI infrastructure. AWS's dominance is not guaranteed, and any loss of share in the high-margin cloud business would disproportionately impact profitability.

Amazon Leo execution risk. The satellite internet business faces a challenging FCC deadline: Amazon is required to have 1,618 satellites in orbit by July 2026 but currently has approximately 241. The company has applied for a two-year extension, but failure to meet the requirement could result in spectrum license revocation. Starlink has a multi-year head start with 7,600+ satellites and 10 million subscribers.

Retail margin pressure. Amazon's expansion into grocery, pharmacy, and same-day delivery is capital-intensive and operates at thinner margins than digital commerce. If the company over-invests in physical retail infrastructure without achieving sufficient scale, it could weigh on overall profitability.

Macro sensitivity. While Amazon has been resilient through economic cycles, a significant consumer spending slowdown driven by elevated oil prices, higher interest rates, or geopolitical disruption could slow retail growth and advertising spend simultaneously. The Iran conflict and sustained energy price increases represent near-term risks to consumer purchasing power.

The Thesis

Amazon is a $717 billion revenue business with $80 billion in operating income and $140 billion in operating cash flow. Its cloud division alone generates more profit than most Fortune 100 companies. Its advertising business, which barely existed five years ago, brings in $68 billion annually. It has a million robots in its warehouses, robotaxis in production, satellite internet in beta, a significant stake in one of the most valuable private AI companies in the world, and 240 million Prime subscribers whose spending habits fund the entire ecosystem.

The $200 billion in 2026 capex is not a warning sign. It is the cost of building moats across cloud infrastructure, AI compute, satellite connectivity, and physical logistics automation simultaneously, funded by cash flows that are growing 20% annually. Every major investment Amazon is making is backed by customer demand signals: $244 billion in AWS backlog, enterprise beta commitments for Leo, triple-digit growth in custom chips, and a robotics feedback loop from a million deployed units that no competitor can replicate.

At roughly 31 times forward earnings, Amazon is trading at less than half its historical average multiple, at a time when its profitability has never been higher and its growth vectors have never been more diverse. The market is pricing the company as if it is a mature retailer with a cloud business attached. The data says it is something far more ambitious: a technology conglomerate that is building infrastructure layers, from ground-level robotics to low-earth orbit, that will generate returns for decades.

At Wealth Engine Pro, the philosophy is to evaluate companies based on what they are, not what someone hopes they will become. In Amazon's case, what the company already is today, a diversified technology platform with demonstrated profitability across multiple high-growth segments, appears to be underpriced relative to the sum of its parts. The emerging businesses (Leo, Zoox, external robotics, healthcare) are not priced into the stock. That is the kind of asymmetric setup that data-driven investors should find worth examining.