Amazon VRIO Analysis 2026: The Rival's Invoice
A VRIO analysis of Amazon in 2026. Seven resources scored on Q2 numbers, and why Amazon Leo must buy SpaceX rockets to catch up with Starlink.
In January 2026 Amazon told the FCC it could not put about 1,616 Amazon Leo satellites in orbit by its 30 July deadline, and blamed a "near-term shortage of available rockets." In the same month it disclosed that it had bought ten more Falcon 9 launches from SpaceX, the company whose Starlink network Leo exists to compete with. That makes October 2026 a useful moment to run a VRIO analysis of Amazon. Amazon is usually the incumbent that rivals cannot copy, and here it is the one doing the copying, so the analysis can look at imitation from both sides.
VRIO scores a company's resources one by one. For each it asks whether the resource is Valuable, Rare, costly to Imitate, and whether the firm is Organized to capture its value. Most Amazon VRIO write-ups answer the imitability question by assertion ("competitors would struggle to match Amazon's scale"). This page uses a checkable signal instead: who has to buy from whom. That is the anchor, The Rival's Invoice.
Position being analyzed
The numbers below come from Amazon's Q2 2026 results unless noted. Amazon reports Q3 in late October 2026.
| Metric | Q2 2026 | vs Q2 2025 |
|---|---|---|
| Net sales | $200.6bn | +20% (from $167.7bn) |
| Operating income | $27.5bn | +43% (from $19.2bn) |
| North America sales / operating income | $116.2bn / $9.1bn | +16% / from $7.5bn |
| International sales / operating income | $42.2bn / $1.7bn | +15% / from $1.5bn |
| AWS sales / operating income | $42.2bn / $16.6bn | +37% (fastest in 18 quarters) / from $10.2bn |
| Advertising services revenue | $19.8bn | Reported growth varies by source |
| Property and equipment purchases, trailing 12 months | $173.0bn gross | from $107.7bn |
| Free cash flow, trailing 12 months | −$7.6bn | from +$18.2bn |
| Non-operating income, mostly from the Anthropic stake | $53.4bn pre-tax | One quarter |
| Q3 2026 guidance | Sales $197–202bn, operating income $22.5–26.5bn | +9% to 12% sales growth |
The Amazon Leo timeline, from Amazon's filings and the FCC waiver:
| Date | Event |
|---|---|
| Jul 2020 | Amazon says it will invest $10bn in Project Kuiper, a 3,236-satellite constellation |
| Dec 2023 | Buys 3 Falcon 9 launches from SpaceX, on top of more than $10bn of launch contracts with ULA, Arianespace and Blue Origin |
| 12 Nov 2025 | Project Kuiper renamed Amazon Leo |
| Jan 2026 | Asks the FCC for 24 more months on the half-constellation milestone. Discloses 10 more Falcon 9 and 12 more New Glenn launches |
| Apr 2026 | Agrees to buy Globalstar for $11.57bn, with an Apple agreement to move iPhone and Apple Watch satellite features to Amazon Leo. Closing expected in 2027 |
| 5 Jun 2026 | FCC waives the July 2026 milestone. Satellites launched after it lose spectrum priority until 1,618 are launched, or until 30 Oct 2027 if Amazon certifies 1,618 have been built |
| Q2 2026 | Nearly 400 satellites in orbit, "enough to begin initial satellite internet service this year" |
VRIO applied, resource by resource
1. Fulfillment and delivery network: sustained advantage ✅
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | It is what makes the North American store profitable: $9.1bn of Q2 operating income on $116.2bn of sales, up from $7.5bn a year earlier |
| Rare | Yes | No other US retailer runs a parcel network of comparable density that it also sells to third-party sellers |
| Inimitable | Yes | The network was built over two decades of capex and is tied to warehouse locations near population centres. A rival with money still has to buy the land and wait for permits |
| Organized | Yes | Fulfillment by Amazon and Prime turn the network into fees and retention, not just lower costs |
2. AWS at scale: sustained advantage ✅
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | $42.2bn of revenue and $16.6bn of operating income in one quarter: a 39% margin, and more than half of Amazon's operating income |
| Rare | Yes | Three providers operate at this scale, and AWS is the largest |
| Inimitable | Yes | See The Rival's Invoice below. Rivals of Amazon's other businesses still buy from it |
| Organized | Yes | It is reported as its own segment, and AI workloads come in through Bedrock and Trainium |
3. Trainium custom silicon: temporary advantage ⏳
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | The chips business passed a $25bn annual run rate in Q2, and Anthropic and OpenAI have made multi-year, multi-gigawatt commitments |
| Rare | Yes | Only two hyperscalers sell in-house AI training silicon at this volume: Amazon and Google |
| Inimitable | No | Google's TPU shows it can be done. Microsoft and Meta are building their own chips |
| Organized | Yes | It is sold through AWS, so every chip comes with a cloud contract |
4. Advertising on first-party purchase data: sustained advantage ✅
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | $19.8bn of advertising revenue in Q2 2026 |
| Rare | Yes | Google and Meta know what people search for and look at. Amazon also knows what they bought, and it shows the ad on the page where they buy |
| Inimitable | Yes | The data comes from the store, so copying it means building the store first |
| Organized | Yes | Sponsored products sit inside search results and Prime Video carries ads by default |
5. The capital budget: competitive parity ⚖️
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | $173bn of property and equipment in 12 months, mostly for AI |
| Rare | No | Microsoft, Alphabet and Meta are each spending on the same order of magnitude |
| Inimitable | — | Not reached |
| Organized | — | Not reached |
A budget that every rival also has is the cost of staying in the race, not an advantage. And it has a price: trailing free cash flow went from +$18.2bn to −$7.6bn.
6. The Anthropic stake: owned, not operated ⚠️
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | It produced most of $53.4bn of pre-tax non-operating income in Q2 |
| Rare | Yes | Only Amazon and Google own large stakes in Anthropic |
| Inimitable | Yes | The stake can't be bought again at the same price |
| Organized | Partly | The gain appears below the operating line. Amazon captures the operating value only where Anthropic's commitments run on AWS and Trainium |
VRIO's fourth question exists for this case. A resource can pass the first three tests and still be an unused advantage if the company isn't organized to turn it into operating profit. Most of what the stake earned in Q2 is a mark-to-market gain, not AWS revenue.
7. Amazon Leo: competitive disadvantage ❌
| Test | Verdict | Reasoning |
|---|---|---|
| Valuable | Yes | Broadband for places without fibre, plus the Apple satellite-feature deal that comes with Globalstar |
| Rare | No | Starlink already has about 26 times as many satellites in orbit and more than 12 million subscribers |
| Inimitable | — | Not reached: Amazon is the imitator |
| Organized | Yes | Amazon is spending and launching heavily, and Globalstar adds spectrum and an anchor customer |
Scorecard
| Resource | V | R | I | O | Result |
|---|---|---|---|---|---|
| Fulfillment and delivery network | ✅ | ✅ | ✅ | ✅ | Sustained advantage |
| AWS at scale | ✅ | ✅ | ✅ | ✅ | Sustained advantage |
| First-party ad data | ✅ | ✅ | ✅ | ✅ | Sustained advantage |
| Trainium silicon | ✅ | ✅ | ❌ | ✅ | Temporary advantage |
| Capital budget | ✅ | ❌ | — | — | Competitive parity |
| Anthropic stake | ✅ | ✅ | ✅ | Partly | Unused advantage |
| Amazon Leo | ✅ | ❌ | — | ✅ | Competitive disadvantage |
The Rival's Invoice
VRIO's third question, "is it costly to imitate?", is usually answered with adjectives. Amazon in 2026 offers a better signal, because it sits on both sides of the same supplier relationship.
The Rival's Invoice: if copying a rival's resource requires buying an input from that rival, the resource is costly to imitate for a reason money cannot fix, because the rival controls the price and the pace of your catch-up. Read in reverse, if rivals have to buy an input from you, that input is very likely your inimitable resource.
| Who pays whom | The input | What it shows |
|---|---|---|
| Amazon pays SpaceX | 3 Falcon 9 launches bought in Dec 2023, 10 more disclosed in Jan 2026 | Launch capacity is the input under Starlink that Amazon can't build on its schedule. Leo is at a disadvantage |
| OpenAI pays AWS | A $38bn, seven-year infrastructure agreement | OpenAI is the main partner of Microsoft, AWS's largest cloud rival, and still buys AWS capacity. AWS is the advantage |
| Netflix pays AWS | Its streaming runs on AWS | Netflix competes with Prime Video and still rents Amazon's infrastructure. See our Netflix Five Forces on why that gives Amazon some leverage |
| Apple will pay Amazon Leo | iPhone and Apple Watch satellite features, under the Globalstar agreement | Amazon used an $11.57bn acquisition to get an invoice it can send, not just one it pays |
The SpaceX row is the one that matters for Leo. Amazon has the money: it pledged $10bn in 2020, its launch contracts alone exceed that, and Globalstar adds $11.57bn. It still missed its FCC milestone, and its own filing named rockets, not money, as the reason. The table below puts the gap in numbers.
| Amazon Leo | Starlink | |
|---|---|---|
| Satellites in orbit | Nearly 400 (Q2 2026) | About 10,400 (June 2026) |
| Subscribers | Initial service due to start in 2026 | More than 12 million (June 2026) |
| Who launches them | ULA, Arianespace, Blue Origin, and SpaceX | SpaceX |
| FCC half-constellation milestone (about 1,616 by 30 Jul 2026) | Missed, waived on 5 Jun 2026 | Not applicable |
This is related to the Auction Test on our Netflix page, but it reads a different number. The Auction Test reads the price a firm refused to pay for a resource. The Rival's Invoice reads who a firm has to pay to get one. Netflix decided it could substitute for what it was bidding on. Amazon can't substitute for launch capacity. It can only wait for Blue Origin to reach cadence or keep paying SpaceX.
Pre-registered: the launch rate, not the budget, sets Leo's pace
| Placement | Amazon Leo stays a competitive disadvantage while launch capacity, not money, is the constraint |
| Prediction | Amazon Leo has launched fewer than 1,618 satellites by 30 October 2027, so it regains spectrum priority by certifying satellites as built, not as launched |
| Falsifier | 1,618 or more Leo satellites launched by 30 October 2027 |
| Scored on | Amazon's Q3 2027 release and the public Leo launch record (late October 2027) |
Where this misleads
- Buying from a rival is not always a weakness. Apple bought displays from Samsung for years without losing anything that mattered. The test only bites when the input is scarce, and launch capacity was scarce: Amazon's own FCC filing said so.
- The invoice can shrink. Blue Origin is owned by Jeff Bezos, and Amazon has booked 12 more New Glenn launches. If New Glenn reaches a high launch rate, Amazon stops depending on SpaceX and the Leo row could move to parity. The score above is a snapshot.
- Paying AWS doesn't mean AWS can't be copied. OpenAI also buys from Microsoft and others. The AWS row shows that AWS is hard to avoid, not that it is the only option.
Signals to watch
| Signal | Resource | Date |
|---|---|---|
| AWS growth vs Q2's 37%, and Q3 operating income vs the $22.5–26.5bn guidance | AWS, capital budget | Q3 2026 release, late Oct 2026 |
| Trailing free cash flow turning positive again | Capital budget | Each quarterly release |
| Amazon Leo commercial service starting, and first paying customers | Amazon Leo | Promised for 2026 |
| New Glenn launches carrying Leo satellites | Amazon Leo (fewer invoices to SpaceX) | 2026–27 |
| FCC approval and closing of Globalstar | Amazon Leo | Expected 2027 |
| Leo satellites launched by 30 Oct 2027 | The pre-registered prediction | 30 Oct 2027 |
Key takeaway
VRIO on Amazon in 2026 finds three moats (fulfillment, AWS and first-party ad data), a capital budget that has become the cost of entry rather than an advantage, and one area where Amazon is the imitator. Amazon Leo is not short of money: Amazon's launch contracts alone exceed the $10bn it pledged in 2020, and Globalstar adds $11.57bn. It is short of rockets, and some of those rockets come from the rival it is chasing. When a VRIO analysis can't settle imitability by argument, follow the invoices. A resource your rivals have to pay you for is hard to imitate. A resource you can only build by paying a rival is one you are still behind on.
Want to run VRIO on your own company's resources? Framework for iPhone & iPad walks the four questions with AI assistance.
Related
- VRIO framework: the four questions and how to score them
- Netflix VRIO Analysis 2026: The Auction Test, which reads the price a company refused to pay for a resource
- Tesla VRIO Analysis 2026: the same framework on a company whose moats are industrial
- Amazon PESTEL Analysis 2026: the macro forces around the same company
- SpaceX BCG Matrix 2026: Starlink seen from SpaceX's side
If you came here looking for a SWOT, our sister site SWOTPal has an Amazon SWOT analysis.
Sources
- Amazon — Q2 2026 results (SEC Form 8-K, Exhibit 99.1)
- Amazon — Q2 2026 earnings release
- CNBC — "Amazon asks FCC to extend satellite limit as it buys more SpaceX rides" (30 Jan 2026)
- Fierce Network — "FCC waives Amazon Leo's July 2026 satellite launch deadline"
- Benton Institute — "FCC Grants Kuiper's Request to Extend or Waive Interim Milestone"
- Fierce Network — "Amazon to acquire Globalstar for $11.57 billion"
- Amazon — AWS and OpenAI multi-year strategic partnership
- Wikipedia — Amazon Leo
- Wikipedia — Starlink
Frequently asked questions
What is a VRIO analysis of Amazon?
VRIO tests each resource a company has against four questions: is it Valuable, Rare, costly to Imitate, and is the company Organized to capture its value. Applied to Amazon in 2026, it separates durable advantages (the fulfillment network, AWS, first-party ad data) from temporary ones (Trainium chips), from resources at parity (the $173bn trailing capex budget, which other hyperscalers match), and from the one area where Amazon is the imitator rather than the incumbent: satellite broadband, where Amazon Leo trails Starlink.
Is AWS a sustained competitive advantage under VRIO?
Yes. AWS had $42.2bn of Q2 2026 revenue, up 37% year on year, and $16.6bn of operating income. The strongest VRIO evidence is who buys from it. OpenAI, the main partner of AWS's largest rival Microsoft, signed a $38bn seven-year AWS agreement, and Netflix, which competes with Prime Video, runs on AWS. When rivals of your other businesses still have to buy from you, the resource is hard to imitate.
Why is Amazon Leo a competitive disadvantage?
Because Amazon is the imitator. In July 2026 Amazon Leo had nearly 400 production satellites in orbit against roughly 10,400 for Starlink, and Starlink had more than 12 million subscribers while Leo had not yet started commercial service. Amazon missed its FCC milestone of about 1,616 satellites by 30 July 2026 (the FCC waived it on 5 June 2026), citing a shortage of rockets, and has bought 13 Falcon 9 launches from SpaceX, the company that owns Starlink.
What is The Rival's Invoice?
The Rival's Invoice is a test for VRIO's imitability question. If copying a rival's resource requires buying an input from that rival, the resource is costly to imitate for reasons money cannot fix, because the rival controls the price and the pace of your catch-up. Read the other way, if rivals have to buy an input from you, that input is likely your inimitable resource.