This Wednesday SpaceX will launch its first batch of Starlink satellites—a “mega constellation” of thousands of spacecraft to provide high-speed Internet access to billions of people at any location on the planet. Starlink is only the first of many such projects; there are at least eight more mega constellations in the works from other companies. Although they promise to revolutionize global telecommunications, these efforts are not free of peril: as the number of satellites inexorably grows, so, too, does the risk of creating dangerous debris that could threaten the continued safe use of Earth orbit. “This is something we need to pay attention to,” says Glenn Peterson, a senior engineering specialist at the Aerospace Corporation, headquartered in El Segundo, Calif. “We have to be proactive.”

Today Earth orbit is a busy place. Almost 2,000 active satellites whiz around our planet, along with nearly 3,000 dead satellites and 34,000 pieces of “space junk” larger than 10 centimeters in size. Whenever debris or a defunct spacecraft gets too close for comfort to an active satellite—typically when a collision risk rises to one part in several thousand—the satellite’s operator must perform a collision-avoidance maneuver. The International Space Station, for example, is moved when the chance of a collision is greater than one in 10,000.

These close encounters already occur thousands of times each year, but the sheer vastness of mega constellations such as Starlink will change the game, resulting in an estimated 67,000 annual collision-avoidance maneuvers if all of them are launched. As Earth orbit becomes jam-packed with satellites, the risk increases. A worst-case scenario would be the Kessler syndrome, a positive feedback loop in which debris-generating collisions create more and more collisions, which in turn create more and more debris, rendering parts of Earth orbit essentially unusable.

Nine companies total—including SpaceX, Amazon, Telesat and LeoSat—have been licensed by the U.S. Federal Communications Commission to launch such constellations. SpaceX alone plans to launch nearly 12,000 satellites by the mid-2020s, which will operate either at an altitude about 500 kilometers in low-Earth orbit (LEO) or a higher altitude of roughly 1,200 kilometers in nongeostationary orbit (NGSO). It is the first company of the nine to launch any fully functional satellites of its constellation. OneWeb, the next front-runner, has plans for a 650-strong constellation in NGSO. Six of its test satellites were launched this past February, and its first proper launch of three dozen or so satellites are planned for later this year. Monthly launches of 30 to 36 satellites will follow, with the service coming online in 2021. Every other company has similar plans for incrementally launching hundreds to thousands of satellites of its own.

Risk versus Reward

The benefits of mega constellations would be manifold. Blanketing the entire planet with high-bandwidth, low-latency, always-on Internet access means ships out at sea, high-flying planes and people in remote, undeveloped areas (even Antarctica!) will suddenly be connected as never before. “Connectivity is just not [currently] available to everybody,” says Mike Lindsay, a space mission designer at OneWeb. “Half the world lacks an affordable access point to broadband Internet.”

Questions remain, however, on how to safely operate so many satellites in orbit. If the satellites fail, they could easily add to the growing problem of space junk. At altitudes of 500 kilometers, failed satellites will not be a huge problem: within several years, atmospheric drag will naturally pull them back toward Earth to burn up on reentry. Indeed, to combat space junk, SpaceX recently modified its license with the FCC to lower the planned altitudes of more than 1,500 of its satellites by half. But at an altitude of 1,200 kilometers, where satellites remain aloft for longer, the dilemma becomes clear: “It’ll be thousands of years at those altitudes,” says Hugh Lewis, a professor of engineering and physical sciences at the University of Southampton in England, who developed a model called DAMAGE to track and monitor space debris.

There are no binding rules currently in place for how long a satellite can safely linger in orbit. The United Nations recommends that satellites be deorbited no more than 25 years after the end of their missions, but these guidelines lack strict penalties for noncompliance. “They are voluntary guidelines,” says Brian Weeden, director of program planning at the Secure World Foundation. The longer a satellite is in orbit, the greater the odds of it colliding with another one are. And such collisions are not unprecedented—in 2009 the American Iridium 33 satellite slammed into the defunct Russian Kosmos 2251 satellite, producing thousands of new pieces of debris.

Some companies are being proactive in how to approach this problem. OneWeb, for example, will attach a handle to each of its satellites, offering an easy way for future orbital scrappers to haul them back down for disposal. No company has yet proved such technology, but progress is being made by entities such as the Japan-based Astroscale. “It is expected that a very small percentage of satellites will fail in such a way that the satellite operator is unable to deorbit them,” says Harriet Brettle, a business analyst at Astroscale. But “Astroscale and other emerging companies are looking to provide a backup service that will remove such failed satellites and maintain a sustainable space environment.”

Other companies licensed by the FCC, however, plan to solely use the onboard propulsion of each of their satellites to ensure a safe deorbit. In principle, doing so seems fine, but in practice, satellite failure rates are not negligible. Even a 99 percent reliability rate for mega constellations would still result in hundreds of dead satellites adrift in orbit. As these numbers stack up, the chance of catastrophe would only grow.

“The real problem is that we don’t have a great track record of getting [satellites] back out of orbit,” says Stijn Lemmens, a space debris analyst at the European Space Agency. “Long-term environment simulations indicate that we would need to reduce the orbital lifetime of about 90 percent of all objects that are launched into orbit. And in reality, we see this is happening successfully for about 5 to 15 percent. So we’re way off the target goal.”

Noisy Skies

Another issue is the radio-based communications of the satellites themselves. Each satellite constellation will be awarded a chunk of the electromagnetic spectrum in which to communicate, but picking out a satellite amidst the noise of so many can be tricky. With thousands more satellites set to enter orbit, actually communicating with a single one among those flying overhead could be difficult. “The radio frequency interference is a big thing that is overshadowed by the potential collision risk,” Weeden says.

These mega constellations could cause problems for astronomy, too. Already, astronomers using optical telescopes have to contend with satellites occasionally crossing their view. Such interference could increase by a factor of several times with the emergence of mega constellations, says Mark Hammergren, a planetary scientist at the Adler Planetarium in Chicago. And for radio astronomers, things could become even more vexing. “Any time a satellite would pass through the observing beam of a radio telescope, there’s a chance that its transmission might be received and interpreted as a celestial signal,” Hammergren says.

Wednesday’s Starlink launch will be rightly lauded as a means to bring the Internet to the masses, but the greater plan of more than doubling the number of active satellites in orbit unavoidably comes with huge complications and seemingly scant room for contingencies. Even if launches and operations unfold smoothly for every mega constellation operator, just one of them experiencing financial difficulties could make the risk of space junk suddenly skyrocket. “The worst case is: you launch all your satellites, you go bankrupt, and they all stay there,” Lemmens says. “Then you have thousands of new satellites without a plan of getting them out of there. And you would have a Kessler-type of syndrome.”