SpaceX’s million-satellite AI plan turns disposal into an orbital waste problem

SpaceX asked the Federal Communications Commission on January 30, 2026, for authority to operate up to one million AI satellites between 500 and 2,000 kilometers above Earth, a plan that would create an orbital waste problem as roughly 200,000 satellites would need disposal each year under a five-year replacement cycle. The FCC accepted the application for review, while the Secure World Foundation, the American Astronomical Society, Amazon, and Viasat called for more detailed scrutiny or denial, citing the environmental risks of disposing of computing hardware via atmospheric reentry or distant disposal orbits.

SpaceX’s million-satellite AI plan turns disposal into an orbital waste problem - SpaceX asked the FCC on January 30, 2026, for authority to operate up to one million satellites between 500 and 2,000 kilometers above Earth. 1 https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf - A five-year replacement cycle would imply about 200,000 satellites leaving service each year, although that is an analytical assumption rather than a confirmed company operating plan. 2 https://arstechnica.com/science/2026/08/spacexs-orbital-data-centers-would-create-a-new-category-of-e-waste/ - SpaceX’s proposed disposal paths include atmospheric reentry, higher Earth orbits and heliocentric orbits, creating different forms of material loss and environmental risk. 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf - The FCC accepted the application for review, while the Secure World Foundation, the American Astronomical Society, Amazon and Viasat called for more detailed scrutiny or denial. 4 https://www.swfound.org/publications-and-reports/swf-filing-comments-on-spacexs-orbital-data-center-application 5 https://aas.org/sites/default/files/2026-03/American%20Astronomical%20Society%20-%20SpaceX%20Orbital%20Data%20Centers%20Petition%20to%20Deny.pdf SpaceX’s plan to put as many as one million AI satellites into orbit would create an unusual waste stream: computing hardware sent beyond practical recovery, burned into the upper atmosphere or parked in distant disposal orbits. 1 https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf The proposal is still a regulatory application, not an approved deployment schedule. But its scale makes satellite retirement a design issue rather than an end-of-mission footnote. If the fleet eventually operated with a five-year average life, maintaining one million satellites would require replacing roughly 200,000 each year. 2 https://arstechnica.com/science/2026/08/spacexs-orbital-data-centers-would-create-a-new-category-of-e-waste/ A disposal plan with several escape routes SpaceX filed the Orbital Data Center application with the Federal Communications Commission on January 30, 2026. The company requested authority for a non-geostationary constellation operating across orbital shells from 500 to 2,000 kilometers, using optical links between satellites and connections to existing and future Starlink systems. 1 https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf The FCC’s February 4 public notice says the application covers up to one million satellites and includes requests to waive standard deployment milestones and surety-bond requirements. The agency accepted the application for filing and opened a comment process; acceptance does not mean the system has been approved. 1 https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf The application says satellites would undergo complete disposal through atmospheric reentry or Earth- and heliocentric-disposal orbits. It does not specify how SpaceX would select among those options for each orbital shell. 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf That distinction matters because conventional electronics recycling depends on bringing equipment back to a facility where metals, circuit boards and other components can be separated. SpaceX’s proposed end-of-life options generally remove those materials from that loop. The result is closer to permanent disposal than to recycling, even when a spacecraft is successfully removed from an operational orbit. 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf The replacement math is more important than the headline number SpaceX has publicly described an AI1 spacecraft capable of about 120 kilowatts of average computing power and up to 150 kilowatts at peak, with radiative cooling and large solar arrays. The company has also discussed a target of roughly 100 kilowatts of compute per ton of satellite mass. Those are proposed design and performance figures, not results from an operational fleet. 6 https://content.spacex.com/cms-assets/FINAL Documents%20and%20Updates/SpaceX%20-%20EU%20Prospectus%20%28Approved%20by%20Bafin%29%20-%20June%205%2C%202026.pdf The five-year replacement estimate comes from SpaceX’s stated or described operating-life assumption for the proposed satellites, combined with the standard arithmetic of dividing one million units by five years. It is not a forecast that SpaceX will retire exactly 200,000 spacecraft annually. Failures, launch delays, hardware upgrades and changes in demand could cluster retirements or spread them across time. 2 https://arstechnica.com/science/2026/08/spacexs-orbital-data-centers-would-create-a-new-category-of-e-waste/ 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf The Federal Aviation Administration has noted that regulators do not generally require large-constellation filings to provide projected annual disposal counts or cumulative reentry hazards for a fixed period. Its analysis of large constellations therefore uses average disposal rates derived from constellation size and mission lifetime, while warning that actual rates may vary. 7 https://www.faa.gov/sites/faa.gov/files/Report to Congress Reentry Disposal of Satellites.pdf The deployment burden would also be large. Independent estimates cited by Ars Technica range from about 17,500 to 77,000 Starship launches to place one million satellites in orbit, depending on spacecraft mass and launch capacity. That translates to roughly 3,500 to 15,300 launches a year over a five-year buildout. For comparison, the world recorded 329 orbital launch attempts in 2025, according to astrophysicist Jonathan McDowell’s tally cited by Ars Technica. 8 https://arstechnica.com/space/2026/07/how-hard-is-it-to-build-orbital-data-centers-actually/ These are scenario estimates, not SpaceX commitments. They show why satellite lifetime, launch cadence and mass are inseparable from the environmental case for orbital computing. A short-lived fleet needs a continuous industrial pipeline, and that pipeline produces both launch emissions and retired spacecraft. Reentry removes debris from orbit, but moves the material elsewhere Atmospheric disposal is often presented as the cleanest way to end a satellite’s mission because it prevents an intact spacecraft from remaining in orbit. SpaceX has used targeted reentry for Starlink satellites, providing an existing operational example of the company’s preferred approach for some spacecraft. 9 https://www.tomshardware.com/tech-industry/space/spacex-vaporizes-260-starlink-satellites-in-six-months-using-earths-atmosphere-new-environmental-concerns-emerge-over-burning-2-700-pound-orbital-data-centers-fcc-seeks-to-exempt-satellites-from-regulations But a satellite that burns up has not simply vanished. Aluminum and other spacecraft materials ablate into the upper atmosphere, where their chemical and climate effects remain the subject of active research. Atmospheric scientists cited by Space.com said rocket emissions and satellite reentries could alter upper-atmosphere chemistry, while emphasizing that the effects and possible thresholds remain uncertain. 10 https://www.space.com/space-exploration/launches-spacecraft/spacex-wants-to-put-1-million-ai-satellites-in-orbit-what-would-that-do-to-earths-climate That uncertainty is important. There is no reliable basis for treating every proposed satellite as an equivalent atmospheric pollutant, because the outcome depends on spacecraft composition, mass, reentry profile and the fraction of material that becomes vapor or particulate matter. SpaceX has not published a complete design and bill of materials for the one-million-satellite system. 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf 5 https://aas.org/sites/default/files/2026-03/American%20Astronomical%20Society%20-%20SpaceX%20Orbital%20Data%20Centers%20Petition%20to%20Deny.pdf A distant disposal orbit creates a different problem. It avoids immediate atmospheric deposition but leaves hardware in the space environment. If future servicing or recycling infrastructure never arrives, those satellites become a long-lived stock of inactive mass. Amazon’s FCC petition said SpaceX’s application contemplated disposal orbits with passive decay times exceeding 200 years at some altitudes and did not explain how the company would choose among the proposed methods. 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf The choice therefore involves trade-offs rather than a zero-waste outcome: atmospheric reentry reduces orbital persistence while dispersing material; disposal orbits preserve the material in principle while increasing the amount of inactive hardware in space. Regulators are being asked to evaluate the whole system The regulatory response has focused on the gap between SpaceX’s headline capacity and the information needed to assess a constellation of that size. The Secure World Foundation said the application raises technical, environmental and governance questions beyond routine licensing and urged the FCC to require system-level analysis of cumulative collision risk, post-mission disposal performance and aggregate interference. 4 https://www.swfound.org/publications-and-reports/swf-filing-comments-on-spacexs-orbital-data-center-application The foundation also recommended a phased, demonstration-based authorization with reporting requirements and performance benchmarks before expansion. That approach would test whether the company can operate a smaller group of compute satellites, manage conjunctions, execute disposal maneuvers and document reentry performance before regulators consider a million-satellite ceiling. 4 https://www.swfound.org/publications-and-reports/swf-filing-comments-on-spacexs-orbital-data-center-application The American Astronomical Society formally petitioned the FCC to deny the application, citing concerns about optical and radio interference, atmospheric effects and orbital safety. It said the application did not provide the satellite size or disposal frequency needed to evaluate those impacts. 5 https://aas.org/sites/default/files/2026-03/American%20Astronomical%20Society%20-%20SpaceX%20Orbital%20Data%20Centers%20Petition%20to%20Deny.pdf Amazon’s Kuiper subsidiary and Viasat separately argued that the application lacked the technical specificity needed for meaningful review. Amazon highlighted the absence of detailed collision-risk modeling, fuel-reserve assumptions and a clear disposal strategy; Viasat described the proposed system as several orders of magnitude larger and more complex than SpaceX’s prior proposals. Those are filings by commercial competitors with interests in the proceeding, not independent validation of the risks, but they identify specific gaps for the FCC to examine. 3 https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf 11 https://aas.org/sites/default/files/2026-03/PUBLIC%20-%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application%20%283-6-2026%29.pdf Those requests do not establish that orbital data centers are impossible. They identify what a credible deployment case would need to show: the spacecraft’s actual mass and propulsion reserve, failure rates, collision-avoidance workload, disposal reliability, annual reentry volume, atmospheric effects and the launch capacity required to sustain the fleet. SpaceX’s broader materials describe orbital AI compute as an early-stage effort dependent on new spacecraft, manufacturing capacity, launch economics and access to large numbers of AI chips. 6 The waste question follows directly from those dependencies. If hardware improves quickly, the economic pressure may favor replacing satellites before their structures fail. If launch capacity or chip supply falls short, the fleet may remain far smaller than the authorization request. Further sources 1 FCC Public Notice DA 26-113 says SpaceX filed on January 30, 2026, for an NGSO … ↗ https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf 2 Ars Technica’s analysis uses a roughly five-year expected lifetime to estimate … ↗ https://arstechnica.com/science/2026/08/spacexs-orbital-data-centers-would-create-a-new-category-of-e-waste/ 3 Amazon’s March 6, 2026, petition quotes SpaceX’s application on disposal option… ↗ https://aas.org/sites/default/files/2026-03/Amazon%20Petition%20to%20Deny%20SpaceX%20Data%20Center%20Application.pdf 4 The Secure World Foundation recommends system-level analysis, phased authorizat… ↗ https://www.swfound.org/publications-and-reports/swf-filing-comments-on-spacexs-orbital-data-center-application 5 The American Astronomical Society’s petition to deny cites concerns about optic… ↗ https://aas.org/sites/default/files/2026-03/American%20Astronomical%20Society%20-%20SpaceX%20Orbital%20Data%20Centers%20Petition%20to%20Deny.pdf 6 SpaceX’s 2026 prospectus describes proposed AI compute satellites, a target of … ↗ https://content.spacex.com/cms-assets/FINAL Documents%20and%20Updates/SpaceX%20-%20EU%20Prospectus%20%28Approved%20by%20Bafin%29%20-%20June%205%2C%202026.pdf +5 more The stories that matter, in one email. Free — unsubscribe anytime.