Side-by-Side Comparison
| Feature | Starlink (SpaceX) | Amazon Leo (Amazon) |
|---|---|---|
| Constellation Size | 19,408 authorised (42,000 applied) | 3,232 authorised (7,736 with next-gen) |
| Active Satellites | 9,898 | 392 |
| Orbital Altitude | 480–570 km | 590–630 km (3 shells) |
| Orbital Planes | Many, across several shells | 98 across 3 altitude shells |
| Inclination | 43°–97.6° | 30°–51.9° |
| Download Speed | 25–220 Mbps (real-world) | Up to 400 Mbps (Leo Pro) and 1 Gbps (Leo Ultra) |
| Latency | 25–60 ms | ~20–40 ms (expected) |
| Inter-Satellite Links | Laser links (v1.5+) | OISL — 100 Gbps infrared lasers |
| Baseband Chip | Custom ASIC | Prometheus (up to 1 Tbps/sat) |
| Commercial Launch | 2020 | Not open yet (planned for late 2026) |
| Subscribers | 13.6 million (100+ countries) | None yet (waitlist open) |
| Customer Terminals | Standard, Standard Actuated, Business | Leo Nano, Leo Pro, Leo Ultra |
| Dish Cost | $349 (US) · £299 (UK) | Not announced yet |
| Monthly Price | From $55/mo (US) · £40/mo (UK) | Not announced yet |
| Launch Vehicles | Falcon 9, Starship | Atlas V, Falcon 9, Vulcan, Ariane 6, New Glenn |
| Architecture | OISL mesh (v2 satellites) | OISL mesh from day one |
| Brightness Mitigation | VisorSat, DarkSat | Dielectric mirror coating |
| FCC Deployment Deadline | Met (operational) | 1,616 by 30 July 2026: missed, waived · 3,232 by 30 July 2029 |
| Operator | SpaceX | Amazon / Kuiper Systems LLC |
Track both constellations live: Starlink Tracker · Amazon Leo Tracker
Satellite counts auto-updated from live orbital data · FCC Gen1 deadline: all 3,232 satellites by 30 July 2029
Constellation Architecture
The two constellations take fundamentally different architectural approaches. Starlink uses a massive number of satellites (targeting 42,000) at relatively low altitudes (480–570 km) in several orbital shells. This brute-force approach delivers excellent coverage and low latency but requires constant replenishment due to atmospheric drag at lower altitudes.
Amazon Leo distributes 3,232 satellites across three concentric orbital shells — 590 km, 610 km, and 630 km — spread over 98 orbital planes. This multi-shell architecture provides layered, overlapping coverage from day one and simplifies constellation management. The higher altitude means longer satellite lifespans but slightly higher baseline latency.
A key difference is inclination. Starlink's primary shell at 53° plus polar shells at 97.6° provides near-global coverage including polar regions. Amazon Leo's inclinations range from 30° to 51.9°, focusing coverage between approximately 56°S and 56°N — where the vast majority of the world's population lives, but providing weaker polar coverage than Starlink or OneWeb.
Inter-Satellite Links: The OISL Factor
Both constellations use optical inter-satellite links (laser links between satellites), but with different adoption timelines. Starlink began adding laser links with v1.5 satellites in 2021, and all current v2 Mini satellites include them. However, Starlink's early v1.0 satellites had no laser links at all, and SpaceX has been retiring them as newer satellites take over.
Amazon Leo has OISL built into every satellite from the first production unit. Each link operates at up to 100 Gbps using infrared lasers, creating a complete mesh network in space. This means Amazon Leo's architecture is OISL-native rather than retrofit — potentially offering more consistent mesh routing. See the OISL mesh visualization on our tracker.
For long-distance connections, OISL routing through space can be faster than terrestrial fibre because light travels ~47% faster in vacuum than through glass. Both providers can leverage this for reduced latency on intercontinental links.
Starlink's Advantages
SpaceX's biggest advantage is time. With 9,800+ satellites in orbit, 13.6 million subscribers, and infrastructure in 100+ countries, Starlink is a proven, revenue-generating service. SpaceX also controls its own launch capability — Falcon 9 and eventually Starship — dramatically reducing deployment costs and removing third-party scheduling dependencies.
Starlink's v2 Mini satellites offer 4× the capacity of earlier versions, and the upcoming v3 satellites launching on Starship will further increase per-satellite throughput. SpaceX's ability to iterate rapidly on satellite design — having launched over 13,000 spacecraft — gives it an operational maturity that Amazon Leo has yet to demonstrate.
Broader orbital coverage is another advantage. Starlink's polar shells mean it can serve high-latitude customers (Arctic research stations, polar shipping routes) that Amazon Leo's 51.9° maximum inclination cannot reach. This matters for government, maritime, and aviation contracts in northern regions.
Amazon Leo's Advantages
Amazon brings massive financial resources — over $10 billion committed to the programme — and an existing global customer base through AWS and Prime. The integration with AWS cloud infrastructure is a unique differentiator: Amazon Leo ground stations connect directly to AWS regions, enabling edge computing at the gateway level for enterprise customers.
Custom silicon is a significant technical advantage. The Prometheus baseband chip powers satellites, customer terminals, and ground gateways alike — processing up to 1 Tbps of data per satellite. This unified chip architecture could reduce costs and improve performance compared to off-the-shelf components.
Amazon has announced three customer terminal types — Leo Nano (portable/IoT), Leo Pro (residential), and Leo Ultra (enterprise/aviation) — targeting a broader range of use cases from day one. Amazon has said the Leo Pro terminal costs under $400 to build, and analysts expect aggressive pricing, potentially bundling with Prime memberships.
Amazon Leo's use of five different launch vehicles (Atlas V, Falcon 9, Vulcan Centaur, Ariane 6, New Glenn) diversifies launch risk — though it also adds scheduling complexity. No other mega-constellation has used this many vehicle types.
The FCC Deployment Deadline
Amazon did not reach the FCC's interim milestone of 1,616 satellites by 30 July 2026. On 5 June 2026 the FCC granted a conditional waiver that keeps the licence in force, and all 3,232 first-generation satellites are still due by 30 July 2029. Amazon had asked the FCC for more time in January 2026, citing launch vehicle availability, particularly delays with Vulcan Centaur and New Glenn.
In February 2026 the FCC authorised Amazon's next-generation systems, 3,212 Gen2 and 1,292 polar satellites, bringing the total planned constellation to 7,736. They carry their own deadlines: half by 10 February 2032 and all by 10 February 2035.
Track the deadline countdown and deployment progress live on our Amazon Leo Tracker.
Launch Vehicle Comparison
SpaceX has an enormous structural advantage in launch costs: it builds and operates its own rockets. Each Falcon 9 launch costs SpaceX an estimated $15–20 million (internal cost), and Starship promises to reduce this further. Amazon must purchase launches from five different providers at commercial rates, making per-satellite deployment significantly more expensive.
However, Amazon's multi-provider strategy provides redundancy. If one vehicle experiences delays (as New Glenn and Vulcan have), others can partially compensate. See the full launch vehicle scorecard on our tracker.
Coverage and Market Focus
Starlink targets consumers first, with a self-install dish and straightforward pricing. It has expanded into business, maritime, aviation (Starlink Aviation), and government (Starshield) segments.
Amazon Leo is expected to pursue a multi-segment strategy from launch — residential broadband, enterprise WAN via AWS, aviation (JetBlue partnership announced for 2027 deployment), and government connectivity. The Prime membership integration could be a powerful consumer acquisition tool in markets where Starlink is already established.
Amazon has also secured preliminary grants from the US BEAD broadband expansion programme, positioning Leo as a rural connectivity solution alongside its commercial offering.
Which Should You Choose?
As of November 2026, Starlink is the only option with widespread commercial availability. Amazon Leo has not opened consumer service yet. Amazon says initial service will start before the end of 2026, first in the US, UK, France, Germany and Canada, and a free waitlist is open at leo.amazon.com. Selected businesses have been testing it in an enterprise preview since November 2025.
For anyone who needs satellite internet today, Starlink is the clear choice. When Amazon Leo launches commercially, competition should drive prices down and improve service quality for both providers — a win for consumers regardless of which service they choose.
For enterprise customers evaluating future connectivity, Amazon Leo's AWS integration and Prometheus chip architecture may offer compelling advantages for cloud-heavy workloads. For polar and high-latitude coverage, Starlink and OneWeb remain the better options due to their higher-inclination orbits.