Amazon's Project Kuiper is a planned 3,232 satellite Starlink rival. It has placed about 400 in 15 months; SpaceX launched 1,589 Starlinks in the first half of 2026.
SpaceX launched 1,589 Starlink satellites in the first half of 2026, roughly 100 more than it had placed by the same point in 2025 and enough to put the company on track to beat last year's full-year record of 3,180 satellites. The H1 2026 number is a deployment-rate figure, not a usage one, and it comes from independent tracker Jonathan McDowell, not from SpaceX's own filings.
McDowell's catalog at planet4589.org is the de facto public ledger for who has put how many objects into orbit. He has logged 12,400-plus Starlink satellites launched since the program began, of which about 11,000 remain functional, according to his running log. The Verge's synthesis of those numbers is what carried the "on track" framing into the news cycle.
The 1,589 figure is a measure of launch cadence. SpaceX's deployment engine is the partially reusable Falcon 9, which can fly the same booster back to a landing zone, refurbish it, and re-fly it within weeks. That cycle lets a small fleet of first stages support more than twenty Starlink missions in a six-month window from SpaceX's two regular Starlink pads, Vandenberg in California and Cape Canaveral in Florida. Spaceflight Now's Vandenberg launch coverage captures the kind of routine mission that pads the count: a Sunday Starlink flight, one of several each month from that pad alone.
Compared with Amazon's Project Kuiper, the H1 2026 number reads differently. Kuiper, Amazon's planned 3,232-satellite low-Earth-orbit network built to rival Starlink, has placed about 400 satellites in the 15 months since its first operational launches, per the McDowell tracker. At SpaceX's current H1 2026 cadence, a single year of Starlink flights would deploy more satellites than Amazon's entire planned constellation.
Amazon's constraint is launch capacity, not satellite engineering. Kuiper's spacecraft are real and they fly; the bottleneck is the combined tempo of the rockets Amazon has booked across multiple providers. Space.com's running tracker is a useful mirror here: it documents how few non-SpaceX missions carry Starlink-class payloads at Starlink-class cadence, and that gap is the structural reason Kuiper will not match Starlink's deployment numbers in 2026 or 2027 on the current launch portfolio.
A caveat: "on track" rests on H2 2026 holding. The H1 number is already in the log. The second half depends on Falcon 9 continuing to fly at its current tempo, which the Federal Aviation Administration can pause for any anomaly investigation. A multi-week stand-down over a booster issue would still leave 2026 ahead of 2025 by H1's margin, but it would clip the headline number. The base rate of Falcon 9 anomalies is low, but Starlink's H1 tempo leaves less margin than a slower program would.
Astronomers have flagged Starlink streaks in survey images for years, and a higher deployment rate compresses the time any brightness-mitigation work has to scale to the new fleet size.
The next concrete marker is the H2 cadence itself. SpaceX typically lists its next Starlink launch within days of each flight, and the McDowell log updates within hours of confirmed deployments. If the second half of 2026 keeps H1's tempo, the annual number will be the new benchmark. If a stand-down interrupts the cadence, the gap to Kuiper's catch-up path widens, not because Kuiper slowed, but because SpaceX would have to spend the second half catching up to its own first half.