AI-generated analysis, not investment advice. The articles are written by AI, edited, and checked against company filings — but the judgements are opinions and the figures go stale. How this is made · Terms

Amazon Leo: The Broadband Rival SpaceX Puts Into OrbitNarrow moat

SpaceX (SPCX) — moat facet

The only well-funded broadband rival still buys its rides from the company it is trying to beat, and that company grew its own broadband revenue 69% while it waited.

Amazon's Leo constellation is the only well-capitalised attempt to build a Starlink competitor, with more than $10 billion committed, commercial service begun and carrier partnerships attached1. It is also a SpaceX customer: Falcon 9 flew three of its missions in 2025 and Amazon bought ten more launches2.

Satellites in orbit~11,000Starlink654OneWeb~394Amazon LeoAmazon bought 10 more Falcon 9 launches from SpaceX to accelerate deployment.
SpaceX is paid to put its only serious broadband rival's satellites into orbit.

Taking that business is not a lapse of judgment. Launch capacity is the binding constraint on Amazon's programme, Amazon would buy it somewhere else if SpaceX declined, and the money is better here than at a rival provider. It remains a peculiar sight.

What SpaceX did with the time is the point. Connectivity revenue reached $5,248m in the September quarter, up 69% on a year earlier and 22% on the June quarter, and its income from operations roughly doubled over the same twelve months3. Consumer subscribers reached 13.6 million, "up 1.6 million sequentially" and "driven by growth in developed markets"4. Developed markets are exactly where fibre and 5G already exist, which makes those additions the harder kind to win.

One caution about the scoreboard itself, and it is the sort of thing worth saying plainly. Starlink has roughly 11,000 satellites in orbit against Amazon's few hundred and OneWeb's 6545, but SpaceX changed how it counts this quarter. Its satellite total now "includes both operational and de-orbit approved satellites"6, and the company says the old definition would have left roughly 400 satellites out of the September figure. A business widening a headline metric in its best quarter is making a small withdrawal from its own credibility, and SpaceX is right that the change does not touch the accounts. It does mean about 400 satellites of this comparison moved by definition rather than by launch.

That SpaceX is taking the rivalry seriously shows up in its costs. Connectivity's total expenses rose $974m year over year, "driven primarily by our growing satellite constellation", Starlink hardware deliveries and "marketing initiatives to support revenue growth"7. Marketing is what a business starts paying for when the customer has a choice, and in developed markets the customer does.

Amazon faces an FCC milestone of roughly 1,600 satellites in orbit by mid-20268, which the current count does not obviously meet. Count satellites actually in orbit rather than announcements, and read Connectivity's growth rate alongside them: 69% is not the number of a business losing a market.

Moat trajectory: Holding steady

Amazon is in commercial service with carrier partners and over $10bn committed, which is further than any previous Starlink challenger reached, and it is still buying its launches from SpaceX. Meanwhile Connectivity grew 69% and added 1.6 million consumers in a quarter, mostly in developed markets. A real competitor, structurally constrained by the thing SpaceX sells it.

The number that tests this moat
Reported
Connectivity revenue growth, latest quarter
69% year over year, to $5,248m

A broadband rival taking share shows up as decelerating growth at the incumbent long before it shows up in satellite counts, which is just as well, because SpaceX has just widened how it counts satellites. Growth falling toward the low twenties while Amazon deploys on schedule would be the first real evidence that Leo is biting.

Source: SpaceX Q3 2026 results release (Form 8-K exhibit 99.1, 20 October 2026) — Connectivity segment operating and financial data and business highlights ↗
References
  1. Third-party estimateAmazon's Leo constellation is the only well-capitalised attempt to build a Starlink competitor, with more than $10 billion committed, commercial service begun and carrier partnerships attached.
    CNBC and constellation tracking — Amazon's Leo constellation (formerly Project Kuiper) entered commercial service with beta partnerships including Verizon, AT&T, Vodafone and NASA, backed by more than $10 billion of committed investment; Starlink has roughly 11,000 satellites in orbit against Amazon's few hundred and OneWeb's 654; the FCC requires Amazon to have about 1,600 satellites in orbit by the end of July 2026 and the full 3,236-satellite constellation by July 2029 — 2026 · publ. July 2026 · source ↗
  2. ReportedIt is also a SpaceX customer: Falcon 9 flew three of its missions in 2025 and Amazon bought ten more launches.
    SpaceNews — Amazon purchased 10 additional Falcon 9 launches from SpaceX to accelerate deployment of its broadband constellation, following a 2023 contract for three Falcon 9 launches that flew in July, August and October 2025, each carrying 24 satellites — 2023-2026 · publ. 2026 · source ↗
  3. ReportedConnectivity revenue reached $5,248m in the September quarter, up 69% on a year earlier and 22% on the June quarter, and its income from operations roughly doubled over the same twelve months.
    SpaceX Q3 2026 results release (Form 8-K exhibit 99.1, 20 October 2026) — Connectivity segment operating and financial data and business highlights — Q3 2026 — quarter ended 30 September 2026 · publ. 20 October 2026 · source ↗
  4. ReportedConsumer subscribers reached 13.6 million, "up 1.6 million sequentially" and "driven by growth in developed markets".
    SpaceX Q3 2026 results release (Form 8-K exhibit 99.1, 20 October 2026) — Connectivity segment operating and financial data and business highlights — Q3 2026 — quarter ended 30 September 2026 · publ. 20 October 2026 · source ↗
  5. Third-party estimateStarlink has roughly 11,000 satellites in orbit against Amazon's few hundred and OneWeb's 654, but SpaceX changed how it counts this quarter.
    Third-party analysis of Starlink economics — Starlink posted $4.29 billion of revenue in Q2 2026, up 66% year on year, with subscribers doubling to about 12 million; average revenue per customer fell from roughly $99 to about $66 over three years — 2023-2026 · publ. 2026 · source ↗
  6. ReportedIts satellite total now "includes both operational and de-orbit approved satellites", and the company says the old definition would have left roughly 400 satellites out of the September figure.
    SpaceX Q3 2026 results release (Form 8-K exhibit 99.1, 20 October 2026) — Note 2, Other Definitions: the SI, subscriber and satellite definition changes — Q3 2026 — quarter ended 30 September 2026 · publ. 20 October 2026 · source ↗
  7. ReportedConnectivity's total expenses rose $974m year over year, "driven primarily by our growing satellite constellation", Starlink hardware deliveries and "marketing initiatives to support revenue growth".
    SpaceX Q3 2026 results release (Form 8-K exhibit 99.1, 20 October 2026) — Connectivity segment operating and financial data and business highlights — Q3 2026 — quarter ended 30 September 2026 · publ. 20 October 2026 · source ↗
  8. Third-party estimateAmazon faces an FCC milestone of roughly 1,600 satellites in orbit by mid-2026, which the current count does not obviously meet.
    CNBC and constellation tracking — Amazon's Leo constellation (formerly Project Kuiper) entered commercial service with beta partnerships including Verizon, AT&T, Vodafone and NASA, backed by more than $10 billion of committed investment; Starlink has roughly 11,000 satellites in orbit against Amazon's few hundred and OneWeb's 654; the FCC requires Amazon to have about 1,600 satellites in orbit by the end of July 2026 and the full 3,236-satellite constellation by July 2029 — 2026 · publ. July 2026 · source ↗
Sources
Generated October 21, 2026