Product1 publisher3 min readPublished
Three FCC filings add 18 antennas across Florida, Minnesota and Texas, about one percent of the more than 300 ground stations Amazon says its 3,232-satellite constellation needs, and the pace of that queue is what a buyer is really buying.
The Product Desk · Product desk
Compiled by The Product DeskSomething wrong?How this is made
The address that matters most in these filings is 3652 170th Avenue in Twin Valley, Minnesota, which DatacenterDynamics describes as a vacant plot [4]. Put a gateway there and you have a construction job: a pad, power, backhaul, and six 2.4-meter dishes bolted down and aimed at Ka-band [6]. The Florida site is a different kind of asset, an operating teleport at 10161 Range Line Rd in Port St. Lucie run by United Teleports [3], where somebody has already solved the dull problems. The antenna counts on the two filings are identical, but the calendars behind them are not.
Three sites against a stated target of more than 300 leaves 297 or more still to come [4]. Divide the company's own two headline numbers and you get the load each site is implicitly carrying: 117Tbps spread across 300 stations averages roughly 390Gbps per station [3]. That figure is not an engineering spec; it is what the published numbers imply about the plumbing, and it explains why a single gateway filing bothers to specify six antennas with no more than four transmitting on the same channel at any moment [6].
The constellation number is the one that wins the slide: 3,232 satellites rated at 117Tbps fully deployed [7][8]. The number that actually decides whether the schools and hospitals named in the paperwork get a circuit they can plan around is different: whether the gateway landing their traffic has been licensed and built. Amazon's own counsel is direct about the dependency, asking the FCC for prompt grant to "accelerate the deployment of the Amazon Leo Systems" and its ability to expand broadband access across the United States [2].
The regulatory half is being engineered to be uneventful. The applications lean on the commission's own conclusion that its power-flux limits were "needlessly prescriptive, outdated, and overprotective of geostationary operations" [10], and the licensing overhaul that followed is pitched as an "assembly line process" [11] in which a filing is granted on a presumption of public interest unless it fails one of seven bright-line criteria [12]. DatacenterDynamics reads those criteria as highly unlikely to trip Amazon up [13]. If that holds, the filings queue is a better progress tracker than any launch count, because a satellite with no gateway beneath it has nowhere to hand traffic off.
The map worth keeping is a grid: your sites down one side, the gateway that would serve each of them across the top, and every gateway marked filed, granted, or standing. The commitment to invest more than $10 billion, restated in the applications [9], says Amazon intends to finish, not which quarter the concrete gets poured in Norman County. And the licensee signing the Florida application is still Kuiper Systems LLC [14], which is a fair reminder of how much faster a rebrand moves than a ground segment.
Ranked by verification strength, evidence, and original report placement.
Amazon Leo has applied to the FCC for fixed-gateway ground stations in St. Lucie County, Florida; Norman County, Minnesota; and Gaines County, Texas.
The 47-page Florida application, written by Kuiper Systems corporate counsel Negheen H. Sanjar, states that prompt grant "will accelerate the deployment of the Amazon Leo Systems and its ability to expand broadband access for consumers, schools, hospitals, businesses, and other organizations across the United States".
The Florida deployment is set for 10161 Range Line Rd in Port St. Lucie, a teleport operated by United Teleports.
The Minnesota deployment is planned at 3652 170th Avenue, described as a vacant plot in Twin Valley.
The Texas deployment is planned for 2271 FM 1757 in Seminole.
The Florida, Minnesota and Texas gateways will each consist of six 2.4-meter antennas in the 27.5-30 GHz, 17.8-18.6 GHz and 18.8-20.2 GHz bands, with no more than four antennas transmitting on the same channel at any one time.
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Primary documents, one reader
Page count, the name of the counsel who signed, three street addresses, dish diameter and band edges: these come off the FCC applications themselves, and DatacenterDynamics quotes them rather than paraphrasing. Two gaps keep this short of strong. No file or docket number is given, so verifying the record takes legwork, and so far only DatacenterDynamics appears to have read these three applications.
Three applications against a 300-station plan
What physically exists at these addresses is a slot at a working teleport in Port St. Lucie and two undeveloped parcels; nothing here says an antenna is on air. Against Amazon's own target of more than 300 stations, three filings is about one percent, and the story provides no count of stations already licensed or carrying traffic. Blue Origin's three Quartz test sites are the only operating ground segment activity mentioned, and they belong to a different company.
Filing language doing the forecasting
The selling in this story is Amazon's, not the outlet's: prompt grant, broadband for schools and hospitals, more than $10 billion committed, 117Tbps someday. DatacenterDynamics passes those through and adds a judgement of its own, that the FCC's seven bright-line categories are unlikely to catch an operator this size, without showing what the categories are. Strip both and the verified news is six antennas each in three counties.
Written to be granted
Amazon drafted the text it is quoted from, and an application asking for prompt grant is composed for the person deciding the grant; its capacity and capex numbers are promises rather than measurements. The FCC's unflattering assessment of its own EPFD limits also reaches this story through the applicant's choice to cite it, which is the selection an applicant would make. On the receiving side, a data centre trade title publishes buildout filings as routine news and has little reason to press on schedules.
Solid on the paperwork, thin past it
The filing facts should hold, because documents of this kind exist in the FCC record and can be checked line by line. Everything downstream is weaker: grant timing, construction, and the rate at which the remaining 297-plus stations get filed and built, none of which this reporting establishes. The per-station capacity arithmetic is ours rather than the outlet's, and it rests on an even split that goes beyond anything the filings actually claim.
product
SpaceX asks the FCC to keep ViaSat-3 F2 off the air weeks before service starts1 publisher
product
Blue Origin lights up three Quartz sites and bets the ground segment is the real bottleneck1 publisher
invest
A million satellites on paper: read the filings, not Musk's 2029 deadline1 publisher
science
Starwashing: treat space environmental claims as marketing until someone audits them1 publisher
Publishers with included, body-backed reporting in this cluster.
1 article · September 7, 2026