Imagine a piece of space debris the size of a hockey puck slams into a Starlink satellite at about 10 kilometres per second. The kinetic energy is equivalent to two kilograms of TNT, or a fully-loaded semi-truck travelling at 100 kilometres an hour.
The Starlink satellite sprays out dozens of new debris pieces into an expanding cloud. Other satellites will pass by the new debris within minutes — some will need to manoeuvre to avoid yet another collision.
Read more:
As corporations race for the stars, we need international collaboration on space governance
As corporations around the world continue to fill low-Earth orbit with megaconstellations, such collisions are increasingly likely. We have developed something called the CRASH Clock to measure this. It asks a simple question: if all satellites in orbit suddenly lost the ability to manoeuvre and control their orientation, how long would it take for two to come close enough to crash?
In 2018, before megaconstellations began launching, the CRASH Clock value was 164 days. It has been steadily dropping since then. Our new research finds that as of May 2026, it is at 2.5 days.
Lethal pieces of debris
Satellites fragment for a variety of reasons. Sometimes they explode internally, as happened to Starlink 34343 in March 2026. Sometimes they collide with debris or a meteoroid. Sometimes it’s even on purpose, for example when Russia tested an anti-satellite weapon in 2021.
When a satellite collides with debris, ground-based radar stations gather information and send alerts to satellite companies and government agencies, who scramble to assess how dangerous low-Earth orbit has become and protect their satellites. Typically it takes about 100 days to catalogue half of the debris from a collision event like this.
(ESA/ID&Sense/ONiRiXEL)
Today, there are more than 10,000 SpaceX Starlink satellites and 5,000 other satellites orbiting above our heads. There are tens of thousands of large pieces of debris with well-measured orbits, which often need to be avoided by satellites with onboard propulsion systems.
There are also more than one million pieces of potentially lethal debris that are too small to be tracked, some just like the hockey puck in the hypothetical scenario above.
Frequent close calls
We haven’t had a satellite-on-satellite collision since 2009, when the Iridium 33 and the defunct Cosmos 2251 collided at an altitude of 770 kilometres, even though there are nearly 20 times more satellites in orbit today.
This is due to careful satellite constellation design, station-keeping manoeuvres, collision avoidance manoeuvres and to some extent, luck.
However, close approaches happen very frequently. Approximately every two minutes, a satellite in the Starlink megaconstellation makes a manoeuvre to avoid another satellite or debris. Currently, they manoeuvre whenever the calculated probability of collision rises above one in 30 million — a successfully conservative approach. That made for around 300,000 manoeuvres in 2025.
The collision-avoidance manoeuvre rate is increasing over time as more and more satellites are launched. The likelihood of a collision with an untracked piece of debris is also increasing over time.
If other satellites are hit by collision debris, more debris clouds will form, possibly causing even more collisions. With clouds of debris come clouds of uncertainty. These clouds also quickly shear into thick rings.
(ESA)
A fragile house of cards
The CRASH Clock highlights how reliant we are on flawless operations to avoid collapsing the fragile house of cards we’ve built in low-Earth orbit.
The CRASH Clock value can be calculated from the publicly available orbits of all satellites and tracked debris. While the calculation we use is based on a worst-case scenario — that all the satellites in orbit suddenly lose the ability to manoeuvre and control their orientation — this situation is not impossible.
An exceptionally strong solar storm, a bad software update or a cybersecurity event are sobering possibilities that could trigger widespread satellite control outages.
CC BY-NC
The CRASH Clock ticks
To reiterate, at the beginning of 2018, before megaconstellations began launching, the CRASH Clock value was 164 days. By May 2026, it had dropped to a mere 2.5 days.
Our direct simulations agree with our probabilistic calculations. They also highlight how averaging techniques can smooth out events — a collision may take days or weeks to occur, or it could occur a few hours after losing control.
Starlink has by far the most satellites in orbit, densely packed within a narrow altitude range, mostly near 550 kilometres from Earth. A recent analysis by renowned space debris researchers Hugh Lewis and Donald Kessler shows that this dense bit of orbit is the only altitude below 800 kilometres that is above the collisional runaway threshold.
In other words, if there is a collision at 550 kilometres altitude, the debris would collide with other satellites, making more debris and causing more collisions. This is called the Kessler Syndrome.
The CRASH Clock shows how dependent we are on collision avoidance systems continuing to work perfectly, every moment of every day, indefinitely.
After the crash
Our CRASH Clock only explores the typical time before collisions could occur. It’s not a countdown to the Kessler Syndrome, nor does it signal an end to our use of satellites in low-Earth orbit. But any collision in orbit makes future collisions more likely.
The CRASH Clock is a measure of how little wiggle room we have to recover from anything that disrupts satellite control.
We are now launching about 100 satellites per week and the CRASH clock is ticking down. This means we have less time to recover from mistakes, while the consequences of any one mistake grow.
