09 / 10

What comes next?

Operators have plans for thousands more satellites. This view follows four programmes with published deployment dates.

On schedule, or two years later?

Shading compares dated plans with a 2-year delay. Selected programmes only; other future launches are outside this chart.

Plans, dates and assumptions

Solid: all recorded payload launches. Upper dashed path: selected plans met on time. Lower path: every remaining deployment delayed two years. Both use the same final satellite count. This is a timing scenario, with no probability attached.

Starlink Gen2

FCC deployment deadline · source 2026-01-09

8,210 already launched in this cohort by 30 Sep 2026. 7,500 by 2028-12-01; 15,000 by 2031-12-01.

Gen2 only. Gen1 and speculative Gen3 filings excluded. Launched counts include returns; additional replacement launches are excluded.

Read primary source ↗

Amazon Leo Gen1

FCC final deadline · source 2026-06-05

396 already launched in this cohort by 30 Sep 2026. 3,232 by 2029-07-30.

Uses the revised 3,232 authorisation. The interim milestone was waived. Two KuiperSat prototypes are excluded from this deployment cohort.

Read primary source ↗

Lightspeed initial phase

Operator deployment target · source 2026-01

0 already launched in this cohort by 30 Sep 2026. 2 by 2027-01-01; 156 by 2028-01-01.

Two pathfinders in late 2026, then 156 by end-2027. The later 225-satellite expansion has no verified full deployment date here and is excluded.

Read primary source ↗

OneWeb renewal orders

Operator replenishment plan · source 2026-09-10

0 already launched in this cohort by 30 Sep 2026. 669 by 2035-01-01.

669 new satellites (440 plus 229), renewing and expanding the fleet through 2034. Existing OneWeb satellites are not subtracted from this new order. Timing within this window is interpolated.

Read primary source ↗

Remaining additions equal each target minus unique already-launched cohort records, floored at zero. Counts use the same source snapshot as the historical total. We interpolate between milestones; these are not published annual launch manifests. Licence deadlines specify deployment and operation, whereas this chart counts launches. Replacement needs and launch failures are excluded except for the explicit OneWeb renewal order.

McDowell’s undated filings are not assigned launch dates. This subset is not a global forecast, and its shaded range is not a confidence interval. A delayed regulatory scenario does not imply a deadline waiver.

McDowell: filed constellation plans ↗

Data & methods

A Crowded Space

GCAT / Jonathan McDowell and CelesTrak SATCAT · snapshot 30 September 2026.

The source table contains 0 objects. The cumulative launch chart uses 0 Earth-payload records. The shared scene contains 0 payload records with valid plotting data; the current selection contains 0. Debris and rocket bodies are outside that payload selection. Fengyun uses a separate debris family.

One JCAT record, one point. The globe uses representative altitude and inclination, with illustrative orientation and phase. Orbital heights use a compressed radial scale.

Orbital Lives follows the original selection: ongoing Earth-payload O/AO phases, blank phase end, no CelesTrak decay, valid plotting values. Its 0 records include inactive satellites and potentially stale statuses.

Atlas angle encodes launch date across 320 degrees; radius encodes logarithmic representative height. The histogram includes all plotted historical payload records by launch year, including later re-entries. Starlink is separated from other payloads by default.

Historical presence uses launch and documented removal dates. Incomplete phase histories and unresolved endings limit the reconstruction. The opening sky uses the original Highlands observer, a 78° field of view and deterministic Kepler orbits. The sky and trail phase use real-time modelled orbital speed: one second on screen equals one orbital-model second. Trails build gradually and retain up to the latest five minutes. These are modelled historical positions; brightness and current tracking are not measured.

Collisions can create fragments that cause further collisions, a self-sustaining process known as Kessler syndrome. The Fengyun test illustrates fragmentation and long-lived debris; these snapshots do not show a collision cascade.

Fengyun-1C: the parent marker is schematic. Transitions connect snapshots, not physical fragment trajectories. The one-day view is reconstructed from later January elements. The 12 October 2026 view uses elements at most seven days old. Fragment identities persist between views; absent records fade out. These samples have different coverage and do not form a survival curve.

Cumulative launches include unique Earth-payload records with a known launch date through 30 September 2026, without requiring valid orbital plotting values. They include later re-entries. The future band compares selected dated deployment plans with a two-year delay, subtracting launches already recorded in each cohort. Linear interpolation between milestones is assumed. Other future launches and replacement needs are excluded, except for explicit OneWeb renewal orders. This is a partial timing scenario, not a global forecast or probability interval. Starlink uses the original name-or-programme match. Mass uses the original four launch-mass bands; zero or missing values use a cross.

Filed constellation plans
Jonathan McDowell · 13 Oct 2026
One million (paper) satellites
Falle et al. · ITU filings, 2017–2022
Space Environment Report
ESA · end-2025 observations

Asset licences and credits

Records in this view by purpose
Purpose	Records
Communications	0
Earth observation	0
Navigation	0
Science	0
Technology	0
Other	0

Total launched · selected plans: on time / two-year delay