Project Kuiper is Amazon’s attempt to pretend it is not already a decade late to the satellite broadband party. After years of glossy renderings and PowerPoint schedules, real satellites finally went up in 2025. More than a hundred are in orbit, which sounds impressive until you remember Starlink throws up that many in a month.

Amazon now faces an FCC deadline that requires 1,600 satellites by mid-2026, a number that would be difficult even if its launch plan did not rely on Vulcan, Ariane 6, and New Glenn. Betting your rollout on rockets famous for being late is less strategy than optimism.

The service is not active yet, but the marketing is. Australia’s NBN has already agreed to use Kuiper to replace its aging GEO system, signing up hundreds of thousands of rural homes that may or may not actually see service by the promised date. That is the Kuiper story in miniature: governments and telcos willing to buy the promise because it has Amazon’s name on it, even if the satellites and ground network remain unfinished.

The real play is not consumer broadband. Amazon does not want to fight Musk for rural households one by one. Kuiper is about tying connectivity into AWS, selling backhaul to operators, and making sure anyone buying data in the future is locked into Amazon infrastructure. Starlink is a retail ISP with a sideline in defense. Kuiper is a corporate utility in waiting, designed to expand Amazon’s cloud grip into orbit.

For now Kuiper is a constellation of press releases and a trickle of satellites. It is real enough to keep regulators patient and investors calm, but nowhere near the scale needed to be competitive. Bezos can fund it indefinitely, which means it will not collapse, but funding is not cadence and cash does not accelerate rockets. Starlink already owns the market and Kuiper is still racing the calendar.