Most satellite communications make use of big satellites which orbit up at 36,000km. Perched at such a height a satellite seems to sit still in the sky, and that vantage allows it to serve users spread across very large areas. But even if such a satellite is big, the amount of bandwidth it can allocate to each user is often quite limited. The orbits used by Starlink’s much smaller satellites are far lower: around 550km. This means that the time between a given satellite rising above the horizon and setting again is just minutes. To make sure coverage is continuous thus requires a great many satellites, which is a hassle. But because each satellite is serving only a small area the bandwidth per user can be high. And the system’s latency—the time taken for signals to get up to a satellite and back down to Earth—is much lower than for high-flying satellites. High latencies can prevent software from working as it should, says Iain Muirhead, a space researcher at the University of Manchester. With software, rather than just voice links, increasingly used for tasks like controlling artillery fire, avoiding glitches caused by high latency is a big advantage.