Amazon.com, Inc.
Amazon is two businesses sharing a brand: a low-margin, high-turnover commerce and logistics network, and a high-margin cloud infrastructure franchise in AWS. A public radar that blends them without comment will mislead. Retail can look mediocre on operating margin while AWS carries group profits; in a cloud slowdown the reverse happens.
Prime, third-party marketplace fees, advertising on its own properties, and logistics density are the retail economics. AWS is capacity, software, and enterprise relationships. Capital expenditure is large in both warehouses and servers. Free cash flow is lumpy around those cycles.
The company reinvests aggressively. That is a feature for long-term owners who trust capital allocation and a bug for anyone screening on trailing GAAP margin. This page tries to describe that machine, not to grade Jeff Bezos-era folklore.
"Retail moat is a mix of network effects in the marketplace (more sellers attract more buyers) and cost advantage from fulfilment density. Prime raises switching costs for households. AWS moat is scale, breadth of services, and the operational cost of migrating production workloads — switching costs more than a legal exclusive. Advertising is an emerging high-margin layer on the retail graph. None of this makes Amazon immune to a consumption recession, a cloud digestion year, or regulatory limits on marketplace power."
Loading the Vaultflake…
—
Health depends on how you treat leases, working capital and capex; the engine can look levered in accounting terms while remaining cash-generative. Quality will be mixed if group ROIC is pulled down by retail reinvestment even as AWS ROIC is high — Vaultflake sees the consolidated payload, not a full segment DCF. Growth can stay solid while profitability mix shifts. Valuation often looks optically high on P/E because reinvestment suppresses earnings; FCF and EV/EBITDA are the better pair. Shareholder return has historically been reinvestment-first; buybacks appear when FCF allows, not as a staples-like dividend machine.
Do not compare Amazon's P/E to Visa's. Look at operating margin and FCF yield through the cycle, and at EV/EBITDA with leases in mind. ROIC is noisy because of capitalised infrastructure. Gross margin mix between retail, ads and AWS matters more than a single headline multiple. Missing cache values stay as em dashes.
Not a recommendation on AMZN. Segment quality is better in the 10-K than in a five-axis average. Yahoo data and Vaultflake will not separate AWS with institutional precision on this public page. Capex cycles can make FCF look terrible in a year when the franchise is fine. Do your own work on unit economics and cloud growth.
Retail moat is a mix of network effects in the marketplace (more sellers attract more buyers) and cost advantage from fulfilment density. Prime raises switching costs for households. AWS moat is scale, breadth of services, and the operational cost of migrating production workloads — switching costs more than a legal exclusive. Advertising is an emerging high-margin layer on the retail graph. None of this makes Amazon immune to a consumption recession, a cloud digestion year, or regulatory limits on marketplace power.
Health depends on how you treat leases, working capital and capex; the engine can look levered in accounting terms while remaining cash-generative. Quality will be mixed if group ROIC is pulled down by retail reinvestment even as AWS ROIC is high — Vaultflake sees the consolidated payload, not a full segment DCF. Growth can stay solid while profitability mix shifts. Valuation often looks optically high on P/E because reinvestment suppresses earnings; FCF and EV/EBITDA are the better pair. Shareholder return has historically been reinvestment-first; buybacks appear when FCF allows, not as a staples-like dividend machine.
Do not compare Amazon's P/E to Visa's. Look at operating margin and FCF yield through the cycle, and at EV/EBITDA with leases in mind. ROIC is noisy because of capitalised infrastructure. Gross margin mix between retail, ads and AWS matters more than a single headline multiple. Missing cache values stay as em dashes.
Not a recommendation on AMZN. Segment quality is better in the 10-K than in a five-axis average. Yahoo data and Vaultflake will not separate AWS with institutional precision on this public page. Capex cycles can make FCF look terrible in a year when the franchise is fine. Do your own work on unit economics and cloud growth.
Consumer Discretionary · Automotive
Consumer Discretionary · Luxury
Consumer Discretionary · Apparel Retail
Consumer Discretionary · Apparel Retail
Consumer Discretionary · Home Improvement
Consumer Discretionary · Home Improvement
A free account opens the interactive chart. The Vault assistant is Premium.