Amazon's answer to Starlink
Amazon Leo — the low Earth orbit broadband network Amazon spent six years building under the code name Project Kuiper — is the company's multi-billion-dollar attempt to become the second serious player in a market SpaceX's Starlink has all but owned. The pitch is straightforward: reliable, high-speed internet for the roughly hundreds of millions of households, enterprises, and vehicles that terrestrial fiber and cellular networks still cannot reach economically. What makes it more than a copy of Starlink is who is building it. Amazon is not a launch company or a telecom; it is a retailer and a cloud provider, and Amazon Leo is engineered to plug directly into both.
The strategic logic is defensive as much as offensive. Connectivity is the on-ramp to everything Amazon sells — retail, Prime, devices, and above all AWS. A world in which Starlink is the default pipe for remote and mobile connectivity is a world in which a rival controls a growing slice of that on-ramp. Amazon Leo is the hedge: a network Amazon owns end to end, from the silicon in the customer terminal to the AWS region the traffic lands in.
From Project Kuiper to Amazon Leo
Amazon announced Project Kuiper in 2019 and won FCC authorization for an initial constellation of 3,236 satellites in 2020, committing more than $10 billion to build the network and its ground infrastructure. Two prototype satellites, KuiperSat-1 and KuiperSat-2, flew in October 2023 and validated the system design, and full-scale production deployment began on April 28, 2025 with the first 27 operational satellites aboard a ULA Atlas V.
In November 2025 Amazon retired the Kuiper code name, rebranded the service Amazon Leo, and revealed a three-terminal consumer and enterprise hardware lineup, signaling the shift from an engineering program to a product with a go-to-market plan. The enterprise beta went live on April 8, 2026, with a marquee roster of test partners — Verizon, AT&T, Vodafone, JetBlue, NBN Co, NASA, and others — and CEO Andy Jassy initially told shareholders to expect broader commercial launch in mid-2026.
The satellites, the terminals, and the factory
Amazon Leo satellites operate in low Earth orbit at roughly 590–630 km, low enough to deliver the sub-second latency that geostationary satellite internet never could. The customer side is a three-tier terminal family: the compact Leo Nano (up to 100 Mbps), the mainstream Leo Pro (up to 400 Mbps, expected to retail for under $400 and measuring under 11 inches square), and the enterprise-grade Leo Ultra (up to 1 Gbps). Driving terminal cost down is central to the strategy — hardware price is the single biggest barrier to consumer satellite-internet adoption.
The satellites themselves are built in-house at a high-volume factory in Kirkland, Washington, designed to produce on the order of four to five spacecraft a day, or up to roughly 30 a week. Vertical manufacturing at that scale is a genuine capability few competitors can match — but through 2026 the bottleneck has not been building satellites. It has been getting them off the ground.
The AWS moat and the revenue map
Amazon Leo's clearest structural advantage over Starlink is its integration with AWS. Because the network's ground stations and orchestration run on Amazon's cloud, enterprise and government customers can route satellite traffic directly into AWS regions — collapsing latency and network complexity for edge computing, disaster recovery, IoT, and private-network use cases. No other satellite operator can offer that native cloud path, and it explains why Amazon has led with an enterprise-first beta rather than a consumer land grab.
The revenue map reflects that breadth: residential and small-business broadband via Leo Nano and Pro; enterprise and government connectivity on Leo Ultra; telecom backhaul for carriers such as AT&T and rural-Africa operator Vanu; maritime service through reseller partners; and aviation, where Amazon has landed two of the largest U.S. airlines. Delta Air Lines signed in March 2026 for an in-flight Wi-Fi rollout across an initial 500 aircraft beginning in 2028 — choosing Amazon Leo over Starlink — and JetBlue committed to an installation program starting 2027.
The deployment race and the FCC crunch
The defining challenge of 2026 has been velocity. Amazon's FCC license requires half of the 3,236-satellite constellation — 1,618 satellites — to be operational by July 30, 2026, and the full fleet by July 30, 2029. By early July 2026 Amazon had only about 396 satellites in orbit, far short of the halfway mark. Launch-vehicle availability, not satellite supply, was the constraint: Blue Origin's New Glenn was temporarily grounded, ULA retired the Atlas V, and the manifest depends on rockets that are themselves still ramping.
On June 5, 2026 the FCC granted a waiver letting Amazon miss the 50% deadline, but with a penalty — any satellite launched after July 30 carries reduced spectrum priority until March 2028 or until Amazon reaches the 50% milestone, whichever comes first. To close the gap Amazon has assembled an unusually diversified launch program of 100-plus missions across five vehicles: ULA's Atlas V and Vulcan Centaur, Blue Origin's New Glenn, Arianespace's Ariane 6, and even rival SpaceX's Falcon 9. Recent flights show both the scale and the strain: an Ariane 64 lofted a record 36 satellites (LE-03) on June 17, 2026, and ULA flew its final Atlas V 551 for Leo (LA-08, 29 satellites) on July 2, 2026.
The capital gap with Starlink
Amazon has committed more than $10 billion to Amazon Leo, funded off its own balance sheet rather than through outside rounds — a luxury almost no other new-space venture enjoys, and one that lets Amazon absorb years of losses that would sink a standalone company. But the gap it is trying to close is enormous. Starlink already operates on the order of 8,000–10,000 satellites and serves well over ten million paying customers, generating billions in annual revenue at healthy margins, while Amazon Leo has yet to open commercial service and is still counting satellites in the hundreds.
That asymmetry frames the whole venture. Amazon is not racing to overtake Starlink in the near term; it is racing to reach commercial viability and defend a strategic position before Starlink's lead becomes unassailable. The enterprise, government, and aviation deals signed during the beta are the early evidence that customers want a credible second supplier — but converting letters of intent into recurring revenue depends entirely on getting the constellation built.
What to watch next
The nearest catalyst is the first commercial service launch, now expected in late 2026 after the mid-2026 target slipped. Amazon says it has enough satellites for continuous coverage across initial latitudes, with the first markets — the U.S., Canada, UK, France, and Germany — focused on enterprise, government, and mobility customers before coverage expands toward the equator. The second thing to watch is launch cadence: reaching the 50% milestone before March 2028 would restore the spectrum priority forfeited by missing the July 2026 deadline, so whether Vulcan and New Glenn can scale to a high tempo is the pivotal variable.
Two longer-dated storylines round out the picture. Amazon's planned acquisition of Globalstar, expected to close in 2027, would fold in spectrum licenses across 100-plus countries and direct-to-device capability — Amazon's answer to Starlink's satellite-to-phone push. And the Delta and JetBlue aviation deployments, ramping in 2027 and 2028, will be the first large-scale, brand-name proof of whether Amazon Leo can deliver at operational scale outside the lab.
Sources
- Amazon — Everything you need to know about Amazon Leo
- Via Satellite — Amazon rebrands Project Kuiper to Amazon Leo, shares terminal lineup
- Payload — FCC gives Amazon Leo more time to launch (waiver & spectrum conditions)
- CNBC — Amazon has deployed enough satellites to launch Leo service this year
- Delta — Delta and Amazon Leo sign 500-aircraft in-flight Wi-Fi agreement
- Amazon — Amazon to acquire Globalstar and expand Amazon Leo network