WASHINGTON, July 7: the challenger’s ledger, as the record has it through this spring. Amazon has put more than $10 billion behind Project Kuiper, its plan for a 3,236-satellite constellation between 590 and 630 kilometers up, and has renamed the effort Amazon Leo. The prototypes flew in October 2023; the rebrand and a public beta waitlist arrived in November 2025; by December, 212 production satellites were up. What there is not, by either comparison in the file, is commercial service: none had begun. The regulatory clock is the hard edge. The FCC mandate requires half the constellation, roughly 1,618 satellites, operational by July 2026, which is why the Kirkland, Washington plant is built to turn out five satellites a day and why the program is run by Rajeev Badyal, a former SpaceX vice president. The technical pitch is optical links between spacecraft at 100 Gbps, what one comparison calls a “network in the sky”, plus three terminal tiers: a 7-inch Nano at up to 100 Mbps, an 11-inch Pro at up to 400 and an Ultra at a gigabit.
The shadow is the point. Starlink, the incumbent since late 2020, counted more than 7,000 satellites on orbit by the February 2026 reckoning, serving 8 million or more customers in over 100 countries; a year earlier, the March 2025 comparison put it at 4 million subscribers across some 75 countries and revenue the March 2025 file estimated at $6.6 billion a year. Residential service runs $120 a month, business tiers $250 to $500 with priority bandwidth, and the aviation product is already flying with United and Hawaiian. The user terminal, a phased-array antenna known to its owners as “Dishy McFlatface”, started at $499 and later rose to $599 in some markets. The fleet flies at about 550 km, low enough for latency in the 20 to 40 ms range, and the second-generation satellites carry laser links that pass traffic between spacecraft, a mesh that matters most over oceans and poles. What is coming next is direct-to-cell: ordinary smartphones connecting to satellites with no special hardware.
OneWeb’s road was different and nearly ended. Founded in 2012, the company went through bankruptcy in 2020, emerged from it with the UK government and Bharti Global splitting the rescue, finished the first-generation fleet of 648 satellites in 2023, and that September merged with the French operator Eutelsat to form Europe’s largest satellite group. The fleet, 652 Ku-band spacecraft at 1,200 km built by Airbus at roughly 150 kg apiece across 12 orbital planes, trades altitude for count: fewer satellites for whole-earth coverage, at the price of latency around 70 to 100 ms. The first generation has no inter-satellite links at all, so each satellite requires a ground station somewhere in range, a real constraint over open ocean and the poles, though the inclination was chosen precisely to cover high latitudes, the Arctic included, and a 2023 test satellite points toward lasers for the next generation. The customer list explains the model: telecom operators reselling under their own brands, mobile networks backhauling remote towers, defense ministries and EU agencies, ships and airlines. A household cannot buy it, and is often not meant to know it is there.
The prize is the roughly three billion people the comparisons count as still lacking reliable access, and the analysts sizing it put the market above $30 billion a year by 2030, with some projections near $50 billion once enterprise and government demand is included. As one satellite telecom market analysis puts it, “The LEO satellite internet market is projected to surpass $30 billion by 2030. The three-way competition between Starlink, OneWeb, and Amazon Leo will drive dramatically lower prices and faster connectivity for billions of people in underserved regions worldwide.” By use case, the file’s honest verdict is that no single “outright winner” exists; the right answer depends on the job. For a household today that answer is Starlink by default, since it is the only one of the three a consumer can order and install; the second half of 2026 is the earliest Leo could matter there, if the price is sharp. For enterprise and government the field levels: OneWeb holds the European relationships, the defense contracts and the telco backhaul; Starlink sells priority business access; and Leo arrives with the AWS ecosystem attached, satellite-to-cloud through a dozen ground stations. The physics, in the end, sorts the shelf: altitude sets latency, and 550, 590-to-630 and 1,200 km are three different products.
On the timeline, the story compresses to a few lines. Late 2020: Starlink opens to the public; OneWeb files for bankruptcy. 2023: OneWeb finishes its fleet and merges into Eutelsat; Amazon flies its prototypes. November 2025: the rebrand and the waitlist. December: 212 machines up. July 2026: the FCC’s halfway deadline, with service promised somewhere inside the year. The record shows one operator at a time turning constellation into revenue, and a challenger whose hardest numbers, the 1,618-satellite halfway mark and the service date, are still ahead of it. What the sky holds by the end of 2026 is the question the next comparison will have to answer.
Sources
- OneWeb vs Starlink vs Amazon Kuiper — Who Wins | GReverse
- Starlink vs Kuiper vs OneWeb: Satellite Internet Comparison 2025
