Pairs with the Profit-Engine Map — a ready-to-use strategy tool, filled for Amazon. Included with a subscription, or $1.99.
For most of a decade, the smartest thing you could say about Amazon at a dinner party was that the store doesn't make money — the cloud does. Amazon sells you a vacuum cleaner at razor-thin margins, the argument went, and quietly pays for the whole circus with the obscene profits of renting servers to everyone else. It was true. It was the cleverest cross-subsidy in modern business. And by 2024, it had stopped being the story.
AWS quietly funds Amazon's money-losing retail business. The retail business stopped losing money. In FY2024, the North America segment earned $25.0B in operating income, and International — which had bled $7.7B as recently as 2022 — swung to a $3.8B profit.13 The store is no longer on life support. So the interesting question isn't whether AWS funds retail. It's what AWS does now that retail can stand on its own.
The cross-subsidy was real — and it was a phase, not a structure
The subsidy story wasn't a myth. It was a snapshot, taken at exactly the wrong moment and then frozen. Rewind to 2022, the year the popular framing calcified: North America posted an operating loss, and International lost $7.7B.3 In that year, the only thing standing between Amazon and a brutal earnings report was AWS. If you described Amazon then, 'a cloud company with an expensive retail hobby' was a fair sketch.
Then management did something un-Amazonian: it stopped spending like growth was free. More than 27,000 jobs were cut. The fulfillment network was torn up and rebuilt around 'regionalization' — moving inventory closer to customers so a package crosses fewer states and touches fewer trucks. The effect on the retail P&L was not subtle. North America went from a $2.8B loss in 2022 to $14.9B of profit in 2023 to $25.0B in 2024.31 That is not AWS rescuing the store. That is the store rescuing itself.
What AWS actually is now: the margin, not the lifeline
Here is the thesis, stated plainly. AWS isn't the thing that keeps Amazon alive — it's the thing that makes Amazon's profits look like a software company's instead of a grocer's. In 2024, AWS generated $39.8B of operating income on $107.6B of revenue, while contributing just 17% of Amazon's total $638B in sales.128 A sliver of the top line, a near-majority of the profit. That's not a lifeline. That's a margin transplant.
| Segment | Share of revenue | Operating income | What it really is |
|---|---|---|---|
| AWS | ~17% | $39.8B | The profit engine |
| North America retail | Majority | $25.0B | Now profitable on its own |
| International retail | Large | $3.8B | Swung from loss to profit |
| Total company | 100% | $68.6B | 10.8% operating margin |
Why does AWS throw off this much profit? Because it sells capacity, not goods. The data centers are a fixed cost built once; every additional hour of compute a customer rents costs almost nothing to provide, and the customer can't easily move their entire stack to a rival without rewriting it. In Q3 2024 that math hit a record 38.1% operating margin, up from 30.3% a year earlier — the highest since the segment was first disclosed.6 Retail, by contrast, runs on a ~6.4% margin even at its 2024 best, because every package is a real box with real shipping. Same company. Opposite physics.
Amazon's reported 10.8% operating margin2 is a weighted average. Retail supplies the enormous revenue base; AWS supplies the profit density; and a $56.2B advertising business — buried inside the retail segments, not broken out — quietly fattens those retail margins from the inside.87 The retail turnaround was never just cost-cutting. It was cost-cutting plus a near-pure-profit ad machine the cross-subsidy story forgot to count.
Isn't this still just AWS carrying the company?
The fair objection: 58% of operating income from one segment is not 'standing on its own.' Pull AWS out and Amazon's profit collapses by more than half — so isn't this the same subsidy in a nicer suit? Partly. AWS is unambiguously the most profitable thing Amazon does, and a sharp recession in cloud spend would still hurt enormously. The dependence is real.
But there's a difference between a profit engine and a life-support machine, and it matters. A subsidy means the funded business cannot survive without the transfer — that was true of retail in 2022. Today, strip AWS away entirely and the retail segments still post a combined ~$28.8B operating profit.1 The store no longer needs the transfer to live; it would simply be a lower-margin, perfectly solvent retailer. And the credit for that turn isn't AWS at all. As CFO Brian Olsavsky put it, the gains came from 'advertising growing, AWS being strong,' but also from 'cost controls' and 'regionalization efforts.'7 Three of those four levers are inside retail. The honest reading isn't 'AWS funds the store.' It's 'AWS makes a now-profitable store look extraordinary.'
Cross-subsidy narratives are sticky precisely because they were once true — and they outlive their facts by years. 'AWS funds retail' described 2021–2022 perfectly and 2024 not at all, because the funded business quietly fixed itself while everyone kept repeating the old line. When you inherit a 'X funds Y' story about any conglomerate, ask one question: would Y survive if you cut the transfer today? If yes, you're not looking at a subsidy anymore — you're looking at a blend, and the high-margin unit is doing something far more valuable than rescue. It's compounding the whole company's margin upward. Don't mistake a profit engine for a crutch; the strategy implications are opposite.
The cleverest cross-subsidy in modern business worked exactly as designed, and then it succeeded itself out of relevance. AWS bought Amazon the years it needed to make the store pay for itself — and once the store did, AWS stopped being the reason Amazon survives and became the reason Amazon's profits look like nothing else its size. The store no longer rides on the cloud. It walks beside it now. And a company that can do both at once isn't running one good business hidden behind a bad one. It's running two good businesses, and letting the world keep believing it has one.
Profit-Engine Map
A one-page map that pulls a business apart into the hook that gets the customer in the door and the engine that quietly earns the margin. Use it to see where the real profit lives, how the two halves are wired together, and what breaks if the link is cut. Blank to dissect your own P&L; filled as the worked example of a business whose advertised product is not where it makes its money.
Included with any subscription, or unlock this tool for $1.99. Get it → · See plans →
Sources
Where this comes from — the filings, records, and reporting behind it.
- 1In FY2024: AWS segment operating income was $39.8B; North America segment operating income was $25.0B; International segment operating income was $3.8B; total company operating income was $68.6B; AWS revenue was $107.6B on 19% YoY growth.
- 2Amazon's 2024 Annual Report (10-K) confirms: total revenue $638B, AWS revenue grew from $91B to $108B YoY (+19%), overall operating income improved 86% YoY from $36.9B to $68.6B (operating margin 10.8%).
- 3In FY2023 (the year before the turnaround completed): North America operating income was $14.9B (vs. a −$2.8B loss in 2022); International operating loss was $2.7B (vs. −$7.7B in 2022); AWS operating income was $24.6B.
- 4Beginning in Q1 2015, Amazon changed its reportable segments to North America, International, and AWS — this was the first time AWS financials were disclosed as a separate segment.
- 5Amazon's 2015 10-K (filed with SEC) confirms: 'Beginning in the first quarter of 2015, we changed our reportable segments to North America, International, and Amazon Web Services (AWS).' Technology infrastructure costs were allocated to AWS based on usage.
- 6Q3 2024 AWS operating margin reached 38.1%, a record since Amazon began separately reporting the segment in 2014/2015, up from 30.3% in Q3 2023.
- 7Amazon CFO Brian Olsavsky stated that operating income improvements were driven by 'advertising growing, AWS being strong,' but also by 'cost controls' and 'regionalization efforts' in logistics — meaning retail profitability improvement is not solely attributable to AWS cross-subsidy.
- 8In FY2024, Advertising Services generated $56.21B in revenue (8.81% of total), AWS $107.56B (16.86%), Online Stores $247.03B (38.72%), Third-Party Seller Services $156.15B (24.48%), and Subscription Services $44.37B (6.96%).