Supercomputers simulated the orbits of 1 million satellites between Earth and the moon — and less than 10% survived

If 1 million satellites were positioned at different points between Earth and the moon, less than 10% would survive long enough to be worth the hassle of sending them up in the first place, new supercomputer simulations suggest. This is not as disastrous as it initially sounds, but it does highlight the complex challenges of expanding humankind's orbital capabilities, the study shows.

Over the past few years, the number of active spacecraft orbiting our planet has skyrocketed — largely thanks to the emergence of private satellite "megaconstellations," like SpaceX's infamous Starlink network and China's growing Thousand Sails project — and the trend is just beginning.

Some experts predict that low Earth orbit (LEO) could safely support up to 100,000 satellites before spacecraft would start uncontrollably crashing into one another, potentially leading to a worst-case domino-effect scenario known as the "Kessler syndrome." Such a scenario would make it impossible to launch anything else into space.

Once LEO is fully saturated with spacecraft, the next logical step would be to start putting satellites in cislunar space — the region between Earth and the moon, according to Live Science's sister site Space.com. Doing so would not only benefit our planet's infrastructure but also provide internet and other services to future human colonies on the moon.

However, it is much harder to predict the orbits of spacecraft in cislunar space because they get caught in a gravitational tug-of-war between Earth, the moon and the sun (which has a greater influence on objects farther from our planet). Without Earth's protective magnetic shield, the radiation streaming out of our home star can also destabilize orbital trajectories in this region of space.

To remedy this issue, researchers at Lawrence Livermore National Laboratory (LLNL) in California used two of their supercomputers — Quartz and Ruby — to simulate the trajectories of approximately 1 million cislunar objects. The simulations required roughly 1.6 million CPU hours and would have taken a single computer around 182 years to complete, according to an LLNL statement. The supercomputers finished the task in just three days.

Of these simulated orbits, roughly 54% remained stable for at least one year, but only 9.7% remained stable throughout the simulations' six-year period. The orbital data were published in August 2025 in the journal Research Notes of the AAS, and the team's analysis was uploaded to the preprint server arXiv in December. (The second paper has not been peer-reviewed yet.)

The team designed the simulated trajectories to be as broad as possible, to account for a wide variety of potential issues, including some the team could not predict.

"The point of it was to not assume anything about what types of orbits we want," study lead author Travis Yeager, a research scientist at LLNL, said in the statement. "We tried to go into it pretending we knew nothing about this space."

Uncertain orbits

Unlike simulating LEO trajectories, which are more stable and repetitive, there is much more uncertainty with cislunar orbits. This meant the team's calculations had to "step forward in time in discrete chunks," making them much more computationally intensive, the researchers wrote.

"If you want to know where a [cislunar] satellite is in a week, there's no equation that can actually tell you where it's going to be," Yeager said. "You have to step forward a little bit at a time."

One of the most surprising factors affecting these orbits was Earth's gravitational influence, which subtly shifts as our planet spins, the researchers noted. "The Earth is not a point source," Yaeger said. "It is actually blobby." For example, there is lower gravity over Canada than there is over the Atlantic Ocean, he added.

While a low percentage of the simulated satellites survived, the results still translate to around 97,000 stable orbits in cislunar space, opening up numerous possibilities for future exploration of the region. Learning which orbits didn't work is just as valuable as knowing which ones did, the team noted.

"From a data-science point of view, this is an interesting data set," Yaeger said. "When you have a million orbits, you can get a really rich analysis."

The researchers have shared the orbital trajectories on an open-source platform to allow anybody to freely access the data in future studies surrounding cislunar satellites.