Amazon's new aviation antenna puts up impressive numbers on paper: 1 gigabit per second down, 400 megabits per second up. Starlink's comparable offering maxes out at 310 Mbps down and 44 Mbps up. On specs alone, Amazon Leo looks like it could lap the competition.
There is one problem. Amazon has 241 satellites in orbit. It needs 1,616 to meet a licensing milestone due this summer.
The company unveiled its Amazon Leo Aviation Antenna on Monday, a purpose-built phased-array unit designed for commercial aircraft, per a company blog post. The company has deals with Delta for 500 planes and JetBlue for roughly a quarter of its fleet, with service targeted for 2027 and 2028 respectively. The antenna shares core technology with Amazon's Leo Ultra ground terminal, repackaged for aviation durability requirements: no moving parts, electronically steered, measuring 147 by 76 by 6.6 centimeters. A single unit is rated to cover an entire aircraft.
The performance numbers are real and the hardware looks solid. What Amazon is short on is satellites.
The company's 3,232-satellite first-generation constellation is years behind schedule, as SpaceNews reported. Launch capacity has been the constraint. Amazon has relied heavily on United Launch Alliance's Atlas 5 and SpaceX's Falcon 9, having purchased ten additional Falcon 9 launches on top of three remaining Atlas 5 flights. Blue Origin's New Glenn and ULA's Vulcan Centaur, both contracted for Amazon Leo missions, have yet to launch a single Amazon satellite. Vulcan is currently grounded pending investigation of a February solid rocket booster anomaly. Ariane 6, the European vehicle Amazon also contracted, managed one launch in February and has another scheduled for late April. The Atlas 5's April 4 mission carried 29 satellites, a new record for Amazon Leo per-launch capacity, but it is a fraction of what a mature constellation requires.
Amazon has requested a two-year extension from the FCC on its July deadline to have half its constellation deployed, citing delays across all three of its primary launch vehicles, according to SpaceNews. The company needs to reach 1,616 satellites in orbit to hit that milestone. It has 241.
Starlink, by contrast, has a working network. SpaceX equipped more than 1,400 commercial aircraft with Starlink in 2025, according to the company's own progress report. Independent analysis from Valour Consultancy puts the number of aircraft actively connected at end-2025 closer to 742, noting that Starlink's figure likely counts installation kits delivered or allocated rather than aircraft actually operating on the network. United Airlines confirmed 300 connected aircraft in February 2026. Qatar, Air France, Alaska, and WestJet are all in various stages of rollout. Starlink has contracts covering roughly 8,000 commercial aircraft globally per Valour's tracking. None of those contracts are theoretical.
The asymmetry in the performance specs deserves scrutiny. Amazon Leo's 1 Gbps down is a network-wide figure dependent on constellation density. Starlink Aviation's 310 Mbps is what a single terminal sees under current network load. As Amazon Leo adds satellites, its per-terminal speeds will improve. As Starlink adds customers, its per-terminal speeds will face the opposite pressure. The gap may narrow. The more immediate constraint is that Amazon Leo's network does not yet exist at the density needed to provide consistent service over routes like transoceanic corridors or polar regions, which the company acknowledges will require a denser constellation before Wi-Fi service can begin.
There is a land grab happening in in-flight connectivity, and Amazon knows it. Airlines are committing to multi-year, multi-aircraft IFC contracts that are difficult to unwind once installed. The companies that sign deals now with Starlink are not going to rip out working hardware to switch to Amazon Leo in 2028 or 2030. That is the window Amazon is trying to compete for, and it is narrower than the press release suggests.
Amazon says its service will be operational by mid-2026. The company needs to deploy roughly 1,375 more satellites before it can credibly offer continuous global aviation coverage. That requires launches. The launch vehicles Amazon depends on are behind schedule, grounded, or both. The antenna works. The constellation does not.
Andy Jassy wrote in his 2025 letter to shareholders that Leo's performance would be "six to eight times better on uplink, and two times better on downlink" than current alternatives. On uplink, that claim is plausible and the numbers support it. On downlink, the gap with Starlink is less pronounced and shrinking. The more relevant question is not what the network can theoretically do, but when it will actually exist at the density aviation routes demand.
The antenna Amazon unveiled today is a real piece of hardware. Whether it has a network to drive it is a different question, and one the company is still answering.
