Orbital data centers in your night sky

Satellites visible across the whole sky, through the night

▬ no mitigation ▬ optimistic mitigation ┄ stars

When

Night beginning

Where

Latitude ° N

Negative for south. Longitude does not matter: this constellation is fixed to the Sun, so only latitude and sun time do.

Brightness model

Light pollution

	whole sky	in view
No mitigation
Optimistic mitigation
Stars

More settings

Satellites in orbit

Orbital plane spread

Field of view

Altitude circles

The plane spread is the one number here that is not from SpaceX's filing. It barely moves the no-mitigation count and moves the optimistic one by up to an order of magnitude. See the notes below.

What this is

Orbits. The sun-synchronous groups in Table 1 of SpaceX's 29 May 2026 supplement to the FCC (ICFS File No. SAT-LOA-20260108-00016): 500,000 satellites between 565 and 1,002 km in dawn-dusk orbits, which keep them sunlit while the ground below is already dark. The roughly half of the filing near 30 degrees inclination is not included, so these counts are a floor for the full filing.

No mitigation is the idealized reflector from Boley, Lawler & Rein 2026 (arXiv:2608.02757), their equation 2. It is what these do if nobody tries to darken them.

Optimistic mitigation is the measured on-orbit brightness of a Starlink Gen2 Mini in its darkening mode (Mallama et al. 2023, arXiv:2306.06657), scaled up for the larger spacecraft. It assumes that mitigation transfers to a satellite roughly six times bigger, which is an assumption, not a measurement.

Neither is a prediction. The gap between them is the point. SpaceX has not published an expected brightness for these spacecraft, and one published magnitude and phase function would collapse the range.

Sky brightness follows the Sun through twilight, calibrated to measurement at the zenith and modeled in shape across the rest of the sky, with artificial skyglow added from the light pollution setting. A satellite or star is drawn only if it is bright enough to see against the sky at its own position, so the picture and the in-view counts agree. The whole-sky counts use the limit at the zenith, the convention the published tables use. The Moon is not modeled.

Stars are real: 8,912 stars brighter than magnitude 6.5 from the HYG catalog, in their correct positions for the date.

Season matters more than you would guess, especially for the optimistic model. Near a solstice the geometry lights the satellites almost face-on, where the measured phase function is several magnitudes brighter than at the side-lit angles of an equinox. How large that swing is depends on the orbital plane spread: it is large at the ±10 degrees used by default and much smaller at the ±30 degrees SpaceX filed.

Precision. The satellites' positions within each orbital plane are one random realization. The no-mitigation counts are steady to about 2% between realizations; the optimistic count near an equinox moves by about 30%, so read it to one significant figure.

Motion. "Satellites moving" advances the orbits and holds the clock. "The night passing" runs the clock and holds the orbits, because at that speed each satellite would lap the sky every few seconds and all you would see is flicker; any one moment of it is statistically the same sky.

Sun time, not clock time. Times here are local solar time: noon is when the Sun is highest. Your clock can differ by an hour or more depending on where you sit in your time zone, and another hour in summer.

This page runs the same physics as the Python in the repository, and it has been checked against it: every count here matches the Python to the satellite for the same inputs (validate_web.py). Code, assumptions and every correction: github.com/sdross0/orbital-datacenter-brightness.