Amazon is preparing to turn its satellite broadband promise into something closer to an actual service. After an Atlas V rocket launch on July 2 successfully propelled 29 satellites into low Earth orbit for the Amazon Leo network, the constellation now stands at 396 units. Amazon’s stated plan is to begin providing service sometime this year, which matters because broadband-from-space only becomes commercially real when customers can actually connect.

Amazon also gave a peek at how it intends to scale. It said it has “hundreds of flight-ready satellites standing by at the Cape,” and it has “a new, dedicated vertical integration facility” ready to support Leo Vulcan 1 and subsequent missions. The goal is clearer launch and deployment cadence so it can expand coverage after an initial service rollout later this year. Translation: it is trying to solve the boring but fatal problems, rocket capacity and integration throughput, that separate a prototype constellation from something that looks like a business.

Leo did not always have this name. Amazon’s satellite broadband plan was formerly known as Project Kuiper until November, but the core idea is unchanged: compete head-on with Starlink by using a large low Earth orbit network to deliver internet. Amazon says its download speeds will range from 100 Mbps to 1 Gbps depending on which of three antenna options customers choose. However, “planned service” is not the same thing as “tested service.” Starlink has years of operational runway, while Amazon is still building out the satellite layer that makes performance claims possible.

The company also has a dual track of capability building that helps explain the pace of satellite deployments. Amazon is not only launching its own hardware, it also acquired additional satellite networking capabilities earlier this year. In April, it agreed to pay more than $11.5 billion to acquire Globalstar and its constellation of 24 satellites. Globalstar provides the satellite network used for Apple’s satellite services, introduced with the iPhone 14 in 2022. For Amazon, that is a strategic shortcut into direct-to-device satellite communications, not just broadband to dish-like terminals.

Still, regulators and competition keep score in different ways. With 396 satellites in low Earth orbit, Amazon has far fewer than Starlink, which the source says has about 10,400 satellites in operation at the moment. Amazon’s eventual plan is to loft more than 7,700 of its own satellites. But the Federal Communications Commission license Amazon holds includes a specific deployment requirement: Amazon must have 50 percent of its planned constellation in orbit by July 30, 2026. The source says Amazon is “clearly going to miss” that deadline. Amazon sought an extension in January, and this was granted last month, but with strings attached.

Those “strings” are the part decision-makers should focus on, even if the market is watching the July 2 launch. A missed deadline can mean added scrutiny, tighter enforcement, or revised milestones that affect timelines and capital allocation. The source does not list the terms of the extension, but it does make one thing clear: Amazon is maneuvering around a regulatory gating item while still trying to promise an initial service rollout later this year. That is a delicate balancing act for boards because it puts operational execution and legal compliance on the same critical path.

On the competitive front, the gap between marketing and reality is also unavoidable. Amazon’s top-tier offering of 1 Gbps has a higher advertised maximum download speed than Starlink’s current residential tiers, according to the source. But real-world performance has yet to be established. Amazon has also not disclosed what it will charge. Starlink’s US residential plans cost up to $130 a month, with equipment charges varying by plan and location, and those price points are the benchmark the average customer will compare against when Amazon eventually launches in earnest.

Outside the US, the regulatory picture is similarly relevant to how quickly an operator can build demand. The source says UK telecom regulator Ofcom granted Amazon Leo a license to beam broadband down to UK consumers over a year ago. That means Amazon is not starting from zero in every major geography. But it also means it is walking into a market where consumers will compare service quality, terminal costs, and reliability, not satellite headcounts.

For executives at satellite operators, telecom incumbents, and infrastructure-heavy tech companies, this is a reminder that the satellite economy is both hardware and paperwork. Amazon’s July 2 launch moved its constellation to 396 units and strengthened its claim that it can begin an initial service rollout later this year, but the FCC deployment milestone by July 30, 2026 is already behind schedule. The strategic stakes are straightforward: the race is not just to launch satellites, it is to convert them into service that performs, is priced well enough to win customers, and stays inside whatever regulatory milestones define “progress” for the license you hold.