Financials

Financials — What the Numbers Say

Amazon today is two businesses fused: a $470B+ hyperscale retailer with thin operating margins and a $130B+ cloud + advertising engine that earns the profits. Revenue compounded at roughly 22% annually for the last decade and reached $716.9B in FY25, while operating margin expanded from sub-3% in 2018-2022 to 11.2% in FY25 as AWS scaled and the retail logistics network swung from over-built to under-leveraged. The catch: free cash flow collapsed from $32.9B in FY24 to $7.7B in FY25 because capex jumped 59% to $131.8B, with ~$200B of capex guided for 2026 to meet AI infrastructure demand. The company prints record GAAP earnings and ROIC near 15% while consuming most of its operating cash flow to build the next decade's data centers. Valuation at ~32x trailing EPS, ~17x EV/EBITDA and ~3.5x EV/Sales is below AMZN's own 10-year average, but the FCF picture decides the next 12 months. The single number that matters most is AWS operating margin — the swing factor that signals whether the AI capex cycle is accretive or value-destructive.

1. Financials in One Page

2. Revenue, Margins, and Earnings Power

Amazon's top line has compounded relentlessly — from $107B in FY2015 to $717B in FY2025 — but the profit story is where the real change happened. After a six-year stretch (2017-2022) where operating margin oscillated between 2% and 6%, the business stepped up to 11%+ in FY2024-2025, driven almost entirely by AWS scale, advertising margin (now an estimated $50B+ revenue stream embedded in retail), and the unwinding of the 2021-2022 logistics over-build.

The gross-margin trajectory tells the deeper story: 33% in 2015 to 50% in 2025. That is not a retailer's curve — that is mix shift toward AWS (cloud gross margin is structurally 60%+) and advertising (gross margin near 80%), which now contribute the majority of operating profit even though they are a minority of revenue. The 2022 dip is the post-pandemic over-investment hangover (extra capacity, extra headcount). What followed was a textbook fixed-cost absorption as revenue caught up to the bigger network, and that has continued into 2025.

The quarterly view confirms the pattern is durable, not a fluke. Operating income has run above $17B in every quarter since Q3 2024 and reached $25B in Q4 2025. Q1 FY2026 ($23.9B op income on $181.5B revenue) was a clean beat versus the $1.65 EPS consensus — actual EPS landed at $2.78. Earnings power is improving structurally, not normalizing to a peak. The risk is that AI training capex has to translate into AWS revenue acceleration to keep this curve bending up; flat-to-down AWS unit economics would freeze margin expansion.

3. Cash Flow and Earnings Quality

The earnings-quality test is straightforward: in most years, Amazon's reported net income converts to multiples of operating cash flow, because depreciation on $443B of property, plant and equipment is a real economic cost that is not yet a cash cost. Free cash flow — operating cash flow minus capital expenditures — is the right anchor for valuation.

Three things to take away. First, the OCF-to-net-income ratio is consistently above 1.5x — Amazon's earnings are real, not paper accruals. Second, capex intensity is the swing variable: 5-6% of revenue in 2018-2019, 18% in 2025. Every percentage point of capex above the depreciation rate is a bet that future cash flow will pay it back. Third, the cash conversion cycle stays negative around -30 to -42 days, meaning suppliers and customers fund Amazon's working capital — a moat advantage that lets the company self-finance growth that would bankrupt a smaller retailer.

4. Balance Sheet and Financial Resilience

Amazon ended FY2025 with $123B in cash and short-term investments against $66B of long-term debt — a net cash position of roughly $57B. Total assets stand at $818B; equity is $411B and has more than doubled in three years on the back of cumulative profits. This is one of the strongest balance sheets in megacap tech for a company of this scale and capital intensity.

Two non-obvious points. The capital lease component (recorded inside long-term debt and PP&E) deserves scrutiny — Amazon increasingly uses long-dated leases to add capacity, which adds reported debt and depreciation but not cash interest expense. Even so, the gross-debt-to-EBITDA ratio is roughly 0.45x and interest coverage is comfortably above 25x. The risk on the balance sheet is not solvency; it is opportunity cost — every dollar locked in PP&E ($443B and growing) is a dollar that cannot be returned to shareholders or redeployed if AWS demand softens.

5. Returns, Reinvestment, and Capital Allocation

Amazon's capital story is unusual for a megacap: management has a multi-decade preference for reinvestment over distribution. There is no dividend, share buybacks have been small and sporadic (~$6B in FY2022, none material since), and the share count has crept up about 1-2% per year from stock-based compensation.

ROIC has averaged roughly 14% over the last decade with a clear cycle: it spiked to 20%+ in 2018 and 2024 when revenue caught up to fixed costs, dipped to 2.5% in 2022 when over-investment outran earnings, and is now in the 15-18% zone. At 15% sustained ROIC, every dollar reinvested at scale is value-creative versus the company's ~9-10% cost of capital. That is the analytical justification for not returning cash today.

The honest counter-argument: SBC of $19-24B per year and rising share count mean per-share value is being diluted by a bit more than 1% annually. With a market cap of ~$2.9T, even a $50-100B buyback would be a meaningful per-share signal, and management's reluctance is a watch-item rather than a comfort.

6. Segment and Unit Economics

Amazon does not provide segment detail in the structured data file used here, but its three reporting segments are well known and disclosed in 10-K and earnings releases: North America retail, International retail, and AWS. Approximate FY2025 mix from filings and presentations: AWS ≈ $115B revenue at ~38% operating margin (the engine), North America retail ≈ $440B at ~6-7% margin, and International retail ≈ $160B at low single-digit margin. Advertising — embedded across segments — runs at an estimated $55-60B annual run-rate at high incremental margin and is the second profit engine.

The economic model in one line: retail is a working-capital efficient, single-digit-margin scale machine that funnels traffic and data into Prime, advertising, and AWS, which together carry ~80% of operating profit. AWS operating income ≈ $44B at ~38% margin in FY2025 means AWS alone explains more than half of consolidated operating income on roughly 16% of revenue. That is the single most important economic fact about Amazon.

7. Valuation and Market Expectations

At the current ~$275 share price (May 2026), Amazon trades at roughly 32× trailing EPS, 17× EV/EBITDA, and 3.5× EV/Sales, against a ~$2.9T market cap. On forward consensus ($8.58 EPS for FY2026, $9.87 for FY2027), the stock is at ~32× FY26E and ~28× FY27E.

Three observations matter. First, the current EV/EBITDA of ~17× is below the 10-year median around 27× — this is one of the cheapest setups for Amazon since 2018, despite materially better margins and ROIC than at any prior period. Second, Price/Book at 6× is the lowest level in a decade as the equity base has compounded faster than the share price. Third, P/E of 32× looks elevated only against the 2024-2025 earnings explosion; on a 2027 consensus basis (~28×), AMZN is in the same neighborhood as MSFT (30×) and below COST (54×), with a stronger EBITDA-margin trajectory than any peer in the cohort.

The Wall Street consensus 12-month reference price sits near $298 (mean of 56 analysts) with rating skewed Strong Buy at 4.62/5. Multiple-driven and DCF-driven estimates cluster in the $280–320 range, implying high-single-digit upside in the base case. Asymmetry is meaningful: the bull case (AWS reacceleration plus operating leverage) frames ~37% upside, while the bear case (AI capex with no incremental revenue) frames roughly 25-30% downside — partially reflected by the recent FCF collapse.

8. Peer Financial Comparison

Amazon's peer set spans two distinct economic models — the megacap tech cohort (MSFT, GOOGL, META) where margin and ROIC are structurally high and the retail cohort (WMT, COST) where scale is large but margins are thin. The table makes the contrast explicit.

Read this row by row. AMZN has the largest revenue base in the cohort, growth in line with the megacap-tech average, but operating margin and FCF margin that look more like a retailer than a tech platform — because consolidated AMZN is still 60%+ retail revenue. The peer gap that matters: AMZN trades at ~17× EV/EBITDA versus 22× for MSFT, GOOGL, and META, and the discount is justified by AMZN's lower consolidated margin profile, not by inferior unit economics inside AWS. The bull thesis is essentially "rerate the AWS-plus-Ads slice toward MSFT/GOOGL multiples and the retail piece is free."

The bubble plot shows that for every megacap-tech peer, EV/EBITDA correlates positively with operating margin. AMZN sits in the lower-left of the tech cohort because of its retail mix — but its trajectory (margin moving from 6% in 2023 to 11% in 2025 toward consensus 13-14% in 2027) suggests it is migrating up and to the right.

9. What to Watch in the Financials

The financials confirm three things: (1) margin expansion is structural, not a one-off; (2) the balance sheet can absorb a multi-year capex cycle without distress; (3) ROIC at 15% supports the company's reinvestment-over-distribution policy. They contradict the cleanest version of the bull narrative on cash returns — FCF in FY25 was barely positive, and FY26 likely worse before it gets better. The cleanest test of whether AI infrastructure is accretive or value-destroying at this scale is whether AWS operating margin holds or expands while data-center capex peaks.

The first financial metric to watch is AWS operating margin. Holding above 36% through the heavy 2026 capex year would be evidence the AI build is paying off in real time, with consolidated FCF returning to growth in 2027. Slipping below 33% would be evidence the capex commitment is destroying value, and would set up multiple compression.