SpaceX’s dream for a million data centers in space is veering closer to reality than science fiction.
On May 29, the company submitted a filing to the Federal Communications Commission that answered questions about its plan. Ten days later, Elon Musk, the chief executive of SpaceX, sat down for a video interview posted on his social media platform X to offer even more details.
“Space is really big. It’s not like space is going to get crowded,” Musk said. “These satellites are so tiny you can’t even see them. They’re very, very tiny compared to Earth.”
Putting data centers in space would solve two of the biggest problems facing AI companies face in the US today: 7 in 10 Americans don’t want them built where they live, and they use a staggering amount of electricity and water.
But while Musk likes to remind people there’s plenty of room in space, scientists have already raised alarms about the dangers of overcrowding in our current satellite ecosystem: Ozone depletion, an unsustainably high number of maneuvers and even debris from satellites falling back to Earth. What happens if we go from our current count of 15,000 satellites to one million?
Since SpaceX announced its plan in January, the space industry experts I’ve spoken with have generally scoffed at the ridiculousness of that number. “I find it very hard to see, but then I’m pretty surprised they’ve gotten to 10,000 now,” Jonathan McDowell, an astrophysicist who tracks satellite launches, told me. “Historically, betting against SpaceX changing the way space is done has not gone so well.”
New details have emerged recently about the reality of this actually happening. In my recent interviews, I’ve noticed an obvious shift from disbelief to genuine worry. “That orbital altitude is not in a great state, and it wasn’t in a great state two decades ago,” says Hugh Lewis, a professor of astronautics at the University of Birmingham, on the planned half a million satellites between 900 and 1,000 kilometes. “It’s not going to end well.”
It’s probably not a coincidence that SpaceX is stoking the data center in space conversation right now, either. The company’s initial public offering on June 12 is poised to be the largest IPO in history and make Musk the world’s first trillionaire. AI-related stocks currently account for nearly half the S&P 500’s market value.
The timing of that IPO helped explain the slapped-together feel of SpaceX’s FCC filing, said Hanno Rein, an astrophysicist at the University of Toronto.
“A lot of the numbers are a bit random and wishy-washy,” Rein told me. “It’s hard to disentangle how much of this we should take seriously and how much of it is just an advertisement for the IPO.”
I reached out to a SpaceX representative for comment and have not received a response.
Bigger satellites, and a lot more of them
Whether you take Musk at his word or with a heavy dose of salt, the sheer size of SpaceX’s data center ambitions is hard to overstate.
The data center satellites — dubbed AI1 in the Musk interview — will be 70 meters long and 20 meters high, for a total area of 1,400 square meters. That’s roughly the size of an NHL hockey rink and significantly above the 800 square meters that SpaceX described in its FCC filing just 10 days earlier.
The AI1 satellites will be roughly 12 times larger than the most common Starlink satellites in the sky today. One million of them would take up an amount of space that’s difficult to conceive.
SpaceX’s data center satellites will be 12 times larger than Starlink’s V2 mini satellites.
“It’s like 1% of the area of New Jersey,” says Rein. “If someone were to propose that on Earth, there would be all kinds of logistical challenges. In space, everything is much more complicated.”
Those massive solar arrays will occupy the same space in low Earth orbit as the more than 15,600 active satellites in that area today. SpaceX says that more than 500,000 of them will live between 946 and 1,002 kilometers — an altitude that’s been stressing scientists out for decades.
That’s because objects at that altitude have a tendency to run into each other. In 2006, NASA determined that space debris would continue to increase for the next 200 years, even without any new launches. This was “primarily driven by the high collision activities” between 900 and 1,000 kilometers — right where SpaceX plans to put as many as half of its orbital data centers.
That altitude is a death trap for satellites because atmospheric drag is so low. That means debris won’t naturally get pulled back into Earth’s atmosphere to get burned up.
“It’s kind of the worst altitude for a lot of things,” says McDowell. “If you have a satellite failure, it’s not going to re-enter for probably hundreds of years.”
Why would SpaceX position so many satellites in this notoriously dangerous zone? It’s essentially the same reason that debris sticks around for centuries: low atmospheric drag. Because the AI1’s solar panels are so big, they’ll have a huge amount of drag at lower altitudes. Essentially, they need to get above the outer atmosphere to overcome that.
It also puts them at far greater risk for contributing to a doomsday event called the Kessler syndrome, a feedback loop in which one collision creates thousands of pieces of debris that would lead to even more collisions. In a worst case scenario, this would make space inoperable for satellites and space travel alike.
Satellite graveyards in space
The other major detail unveiled in SpaceX’s FCC filing is its new methods for disposing of its AI1 satellites. In an effort to mitigate some of these overcrowding issues, the FCC began mandating in 2022 that satellites be de-orbited after five years in the sky — a significant reduction from the previous 25-year rule.
But what you do with those satellites after their five years are up has presented more problems. SpaceX currently lowers the altitude of its satellites until they burn up in the atmosphere. Only about a thousand Starlink satellites have been de-orbited so far, but each one adds significant quantities of aluminum and lithium aerosols to the atmosphere, which could eat away at the ozone layer and potentially accelerate climate change.
One study funded by NASA and published in Geophysical Research Letters in mid-2024 found that a 550-pound satellite releases about 66 pounds of aluminum oxide nanoparticles when it’s deorbited. The presence of these nanoparticles grew eightfold from 2016 to 2022, when there were still relatively few satellites being de-orbited. SpaceX says its new data center satellites will weigh about 6,600 pounds, compared to 1,760 pounds for the V2 satellites.
SpaceX plans to drastically increase the payloads it’s sending into space over the next six years.
This is the ripple effect that seems to weigh most heavily on the scientists I’ve spoken with, so I was interested to learn that SpaceX is looking to move away from atmospheric re-entry for its AI1 satellites above 600 km. At these altitudes, the company is asking that the FCC allow them to retire satellites into “Earth disposal orbits or heliocentric orbits.”
“Earth disposal orbits” essentially means anything outside of low Earth orbit, which stops at approximately 2,000 km. It’s an open question where exactly SpaceX would send these defunct data centers, but just because they’re out of low Earth orbit doesn’t mean they stop posing a risk.
“If they were just proposing a constellation of 100 satellites to dispose into that area, I’d go, ‘Yeah, OK, that’s probably fine,’” McDowell says. “But a million, that’s going to be a very serious ring around the earth of dead stuff that is going to be a hazard to traffic.”
Both McDowell and Lewis brought up the 2008 animated Disney movie WALL-E when I asked them about these graveyard orbits, in which the Earth has become surrounded by a dense field of space flotsam.
Scientists I spoke with mentioned the 2008 movie Wall-E, in which ships have to navigate a dense field of abandoned satellites.
“If you think about a million satellites, each with a lifetime of five years, that’s 200,000 a year that are going to be retiring on average,” says Lewis. “They’re going to collide, and the fragments will get ejected back into low Earth orbit.”
“It’ll be a cloud around the Earth made up of what will be after five years a million satellites, and then another five years later is another million you’ve added. And you just keep doing that for as long as you operate the constellation.”
SpaceX also mentioned “heliocentric orbits” as another method of disposal, which would send old satellites past Earth’s gravitational pull and into orbit around the sun. While this would take care of the space debris and Ozone concerns, it would also be a lot more technically challenging.
These orbital data centers will live between 550 and 1,000 kilometers. To escape Earth’s gravitational pull, they would have to travel beyond geostationary satellites at 36,000 kilometers. That would require either massive amounts of fuel, or years of time — years in which SpaceX would be responsible for avoiding collisions, says Lewis.
“You’re basically just launching fuel, or it’ll take years, and in that time you’re passing through every altitude to get it out of the way,” says Lewis. “I can’t see how this is a viable prospect.”
Potential casualties on the ground
One of the more grim details from SpaceX’s FCC filing is when it notes the casualty risk to people on the ground is “less than 0.0001,” which it says is “considered zero.” This would only happen if pieces of a satellite survived re-entry and struck someone — an extremely rare outcome, but not “zero,” according to Rein.
“It does not make any sense,” Rein says. “They say they have a probability of 10 to the minus four casualty risk, but that’s per satellite, so if you multiply 10 to the minus four times a million, then it’s 100 people.”
Could we accept a world in which 100 people are killed by falling pieces of satellites each year? It might not be as far-fetched as it sounds.
Last year, farmers near Saskatoon found pieces of a Starlink satellites the size of large laptops in their fields. Debris from rocket launches are much more common: Last year alone, rocket parts fell on a mine in Australia, on a farm in Argentina, in the Algerian desert, near a school in Argentina and at a warehouse in Poland.
Bekah Hinojosa, a community organizer with the South Texas Environmental Justice Network, lives in Brownsville, TX, near a SpaceX launch site.
“Every time they launch a rocket, my apartment starts vibrating. And I live 20 miles away from the launch pad,” Hinojosa says.
What happens if we reach a million satellites?
Satellite megaconstellations are a relatively recent phenomenon. When SpaceX launched its first batch of Starlinks in May 2019, there were only around 2,000 active satellites in the entire sky.
Since then, scientists have been ringing alarm bells about what the unintended consequences could be of sending so many objects into space. 15,000 satellites is already a concerning number; one million (much larger) satellites could exacerbate these issues in profound ways.
“With a million satellites, it kind of doesn’t matter where you put them. They’re going to be a pain wherever they are,” McDowell says. “On the one hand, this seems crazy. On the other hand, this is not the first time that we felt that and they’ve gone ahead and made it work.”
