Amazon's Space Internet Bet: 396 Satellites In, Starlink Still King
Amazon says it's getting ready to turn on satellite broadband later this year. The announcement came alongside news that a July 2 Atlas V launch deposited 29 more Leo satellites into low Earth orbit, pushing the constellation past 396 units. That's a milestone worth noting in absolute terms — hundreds of machines humming up there, beaming data back to Earth. But step outside the press release for a second and the picture gets honest pretty quickly.
Starlink has roughly 10,400 satellites in operation right now. Amazon's entire fleet — every satellite it's launched since this thing started as Project Kuiper back in 2021 — amounts to less than four percent of what SpaceX has already deployed. The gap isn't closing at the speed investors probably hoped when Jeff Bezos first announced the program.
Here's what actually matters about where things stand, and why Amazon keeps talking about a 2026 launch window despite the math looking rough.
The Constellation Math Nobody's Talking About
Let's be clear about the FCC deadline situation. Amazon's license requires it to have 50 percent of its planned constellation in orbit by July 30, 2026. That means roughly 1,616 satellites — given its licensed authorization of 3,232. Right now it's at 396. The company is going to miss that deadline. It asked for an extension back in January, and the FCC granted it last month with conditions attached, but the underlying problem hasn't changed: Amazon needs to launch about 1,200 more satellites in the next three months to come anywhere close.
That's not impossible if you've got launch cadence, but Amazon doesn't have it yet. The company is still relying on Atlas V for its current deployments — a proven rocket, sure, but one that's aging and expensive. The next chapter involves Leo Vulcan 1, a new launch vehicle being developed in-house, and Amazon says it has a dedicated vertical integration facility at the Cape ready to support it. Melissa Wuerl, Amazon's director of launch systems, put it this way: "With hundreds of flight-ready satellites standing by at the Cape and a new, dedicated vertical integration facility ready to support Leo Vulcan 1 and subsequent missions, we have a clear path to increase launch and deployment cadence."
Translation: the infrastructure is being built, but it isn't operational yet. The company's long-term plan calls for more than 7,700 satellites total — a number that exceeds its current FCC authorization and will require additional regulatory approval at some point. That's a multi-year buildout, not a 2026 sprint.
What the Service Actually Looks Like
Amazon hasn't disclosed pricing. That's a deliberate choice, and probably the right one at this stage — you don't set consumer prices before you've proven the network works. What we do know comes from Amazon's own specifications: download speeds ranging from 100 Mbps to 1 Gbps, depending on which of three customer-selectable antenna configurations you buy into. The top-tier 1 Gbps figure is genuinely interesting because it exceeds Starlink's advertised residential speeds, which cap out around the $130-per-month tier in the US.
But advertised speeds and real-world performance are different animals. Starlink's actual throughput varies wildly based on network congestion, your location, and the time of day. Amazon has zero flight data to point to. The 1 Gbps claim is a lab number, and until someone actually plugs in an antenna over the Atlantic or the Midwest and runs a speed test, it's just a spec on a slide deck.
The three antenna options suggest Amazon is thinking about this the way it thinks about AWS — tier your hardware, tier your customers, let them self-select into whatever experience they can afford. That's a smart play if you're trying to capture both rural households and enterprise customers with the same constellation. It also means the entry price point could be lower than Starlink's, which matters in markets where $130 a month is a serious commitment.
The Globalstar Acquisition Changes the Board
The $11.5 billion purchase of Globalstar in April 2026 is the move that gets people most confused. Amazon Leo was always pitched as a broadband-from-space play, straight-up Starlink competition. So why drop eleven and a half billion on a company whose constellation numbers in the twenties?
The answer is direct-to-device communications. Globalstar provides the satellite network behind Apple's satellite services, which launched with the iPhone 14 in 2022. That means emergency SOS, text messaging from anywhere on Earth, and eventually a path to smartphone connectivity without any ground-based antenna at all. Amazon isn't just buying satellites — it's buying an existing customer base, operational expertise, and a relationship with Apple that didn't exist before.
This is the kind of acquisition that makes sense on a five-year horizon and looks bizarre on a quarterly earnings call. Amazon is signaling that satellite broadband is only the opening move. The real prize is becoming the orbital infrastructure layer for direct-to-device communications across multiple device categories — phones, tablets, IoT sensors, the works. SpaceX has talked about direct-to-phone capabilities for Starlink, but Amazon already owns a working network doing exactly that through Globalstar.
The financial scale of the deal also tells you something about Amazon's patience. Eleven and a half billion isn't a testing-the-waters number. It's a "we're building an empire up there" number.
Regulatory Position and Competitive Reality
Amazon has cleared the major regulatory hurdles in its key markets. The FCC license remains active with an extension granted through at least mid-2026, and Ofcom in the UK approved Amazon Leo's license to beam broadband to British consumers over a year ago. Regulatory risk is largely behind the company — the harder work is purely operational.
The competitive reality, though, deserves its own honest section. Starlink launched commercial service in 2020. Amazon is talking about an initial rollout later this year — six years behind. In the satellite broadband market, that's not a head start. That's a canyon. Starlink has accumulated years of operational data, customer feedback loops, ground station infrastructure, and brand recognition. People who want satellite internet and can afford it already have it.
Amazon's advantages are real but narrow: deeper pockets, a cloud computing ecosystem that could integrate satellite connectivity into AWS offerings in ways SpaceX can't easily replicate, and the Globalstar acquisition that gives it direct-to-device capabilities from day one. The 1 Gbps speed target is also a genuine differentiator if Amazon can deliver on it.
But here's the thing about markets like this: the first mover doesn't always win forever. Google's Project Loon proved that you can build a stratospheric network, even if it didn't scale commercially. OneWeb went from near-bankruptcy to being acquired by Eutelsat and rebuilding from scratch. The orbital broadband market has room for more than one player — but only if you can reach scale before the customer base stops looking.
Amazon's 396 satellites are a start. The question isn't whether they can reach 7,700 eventually — it's whether the service launch later this year will generate enough momentum to make the buildout financially sustainable. SpaceX's valuation recently briefly surpassed Amazon's own market cap, which says something about where the market thinks this race is headed. Investors are betting on execution, and Amazon's execution record in space has been mixed at best.