"Amazon Leo vs Starlink: The Satellite Internet Duopoly Begins in 2026"

For most of the last five years, satellite internet has had exactly one name that mattered to consumers: Starlink. SpaceX now operates more than 11,000 satellites and serves over 12 million subscribers worldwide, and no rival has come close to matching that scale. That monopoly is about to end. Amazon says its Leo network — the project formerly known as Kuiper — will begin commercial service before the end of 2026, and the numbers behind that claim have finally stopped looking like a press release and started looking like a factory output report.

The factory is the story. Amazon's plant in Kirkland, Washington recently turned out its 1,000th Leo satellite. The first 27 took six months to build; the plant now makes about 27 a week, up to five a day, shipping finished spacecraft to Florida every two or three days. What the factory produces and what customers can use are still different things — only 396 satellites are in orbit after 14 launches, against a first-generation constellation of 3,232 — but Amazon's own executives say the current fleet is enough for an initial service. "We're still doing work to raise them into final position, and there's a lot of work on the ground to do. But the plan is to launch initial service in 2026," Chris Weber, Leo's vice president of business, told Ars Technica. The company already missed its original mid-2026 target, so that sentence deserves to be read as a plan rather than a promise. Still, the direction is unambiguous, and the Philippines and Uzbekistan both expect Leo service before the year is out.

That last row is the whole strategic problem in one line. SpaceX launches its own satellites on its own rockets; Amazon rents rides from ULA, Arianespace, Blue Origin and even SpaceX, at several times the cost per satellite. Amazon cannot win a price war fought satellite-for-satellite, and its executives do not seem to plan to. Their weapon is distribution. Weber has said openly that Leo will be bundled with the rest of Amazon: Prime memberships, Prime Video, eero routers, Ring doorbells, Fire TV. Starlink sells you a dish; Amazon can sell you a dish that arrives with everything else you already buy from Amazon, possibly at a price that looks like a rounding error inside a Prime subscription. That is a genuinely different way to compete, and it is the reason I find this story more interesting than a second constellation in the sky.

The enterprise side is already moving before consumers get their turn. Delta Air Lines has chosen Leo over Starlink for its planes, with service expected in 2028; AT&T will use Leo for business broadband; JetBlue plans Leo in-flight Wi-Fi from 2027; Vodafone and Vodacom have been beta customers across Europe and Africa since April. Enterprises do not care about the romance of space. They care about a second supplier, about AES-256 encryption and private routing into AWS data centers without touching the public internet, and about an uplink Amazon claims is ten times Starlink's. If Leo works for them first, the consumer launch becomes a formality rather than a gamble.

Here is what I am watching, and where I will put my own number on it. The question is not whether Leo eventually exists — with 1,000 satellites built and a July 2029 regulatory deadline for the full constellation, it will. The question is whether the company that missed mid-2026 makes the end of 2026, with no launch date and no pricing announced yet. I put that at 70 percent: the fleet is sufficient, the ground work is the stated bottleneck rather than the rockets, and two governments are publicly expecting service this year, which is a hard thing to walk back. The honest risk is the one Amazon itself names — satellites still climbing to final position, ground systems still being finished — plus the launch manifest, which depends on other companies' rockets. A slip into early 2027 would not kill Leo, but it would be the second missed date, and second misses change how seriously the market takes the third one.

If it lands, the satellite internet market stops being a monopoly and becomes a duopoly with two completely different theories of winning: one company that owns the rockets and sells the sky, another that owns the storefront and sells the bundle. I know which kind of competition usually produces better prices. It is the kind where the customer has somewhere else to go.