Amazon has pushed its satellite internet service Leo toward the middle of 2026, a launch window that sounds tidy on paper and messy everywhere else. The company still wants to sell a commercial network of low-Earth-orbit broadband satellites, but it is arriving late, launching slowly, and leaning on other rockets to get the job done.
That matters because Amazon Leo is not entering a quiet market. SpaceX has already turned Starlink into the default name in satellite internet, with a sprawling fleet and a head start that Amazon cannot buy back. Amazon’s pitch is that it can still win on speed, pricing, and a tighter connection to AWS, which gives the service a more obvious enterprise angle than a simple consumer broadband play.
Jassy’s timetable suggests Amazon is finally preparing for a real launch rather than another demo. The company had previously pointed to a test rollout for corporate customers at the end of 2025, so the new target appears to be the point at which Leo becomes a service, not just a promise with a logo.
Amazon Leo still depends on other rockets
Unlike SpaceX, Amazon does not have its own regular launch machine for Leo. For now, it is using outside providers, including SpaceX itself, until Blue Origin’s New Glenn becomes fully operational. That dependence slows everything down and leaves Amazon hostage to someone else’s launch cadence, which is awkward for a company trying to challenge the market leader.
The scale gap is also hard to ignore. Amazon has approval from the US Federal Communications Commission to deploy 3,236 Leo satellites, but only 241 have been launched so far. The company must get half of the constellation up by July 2026, and it has already asked FCC chair Brendan Carr for more time.
What Amazon Leo is promising customers
Amazon says Leo will be faster and cheaper than existing alternatives once it is live. The bigger sell, though, is integration with AWS, which could make the service attractive to companies and governments that want satellite links feeding cloud storage, analytics, and AI workloads without stitching together separate vendors.
- Planned commercial launch: middle of 2026
- FCC authorization: 3,236 satellites
- Satellites launched so far: 241
- Launch partner dependency: outside rockets, including SpaceX
- Target customers: businesses, governments, and eventually broader users
The opening Amazon really wants is not a carbon copy of Starlink. It is the gaps Starlink has not filled cleanly enough: remote enterprise networks, government communications, and regions where customers want global coverage tied into a cloud stack they already use. If Leo hits even part of that brief on schedule, Amazon will have a niche. If it slips again, Starlink gets more time to look less like a competitor and more like the category itself.