• Satellite systems are designed for remote regions with low population density and no pre-existing infrastructure. In these "greenfield" areas, deploying fiber from scratch would have a noticeable carbon impact, making satellites more competitive at the cost of lower throughput and higher latency, particularly with GEO satellites.
• This study helps refine an environmentally friendly procurement and system design policy. For mixed portfolios, strategies that combine LEO and GEO connectivity services can enable to minimize greenhouse gas emissions for communication services in remote rural areas.
Abstract
Low Earth Orbit ( ) and Geostationary Orbit ( ) satellite systems provide complementary capabilities but follow fundamentally different design philosophies that shape their embodied carbon footprints. This article presents an analysis of the carbon footprint of a large LEO constellation, comparing it to GEO satellites. The methodology quantifies both embodied emissions (from satellite manufacturing, launch, and ground infrastructure) and operational emissions, (from ground stations and user terminals). Drawing on published lifecycle assessments for GEO systems and representative bill of materials (BOM)–based analyses for contemporary LEO constellations, we find that LEO system’s carbon footprint is dominated by satellite manufacturing (40%) and subscriber terminals (38%), challenging thus prevailing assumptions and common focus on launch emissions as they represent only 17% in this study. Embodied and operational emissions from ground stations contribute a smaller share at 5%.
GEO, by contrast, concentrates embodied emissions into a small number of large, long-lived spacecraft, often reported at roughly 56 ktCO2e over a 15 years’ lifetime per satellite in the literature. Because architectures and utilization patterns differ, fair comparison requires normalization to delivered service (e.g., per subscriber, per Gigabyte). We discuss methodological choices, uncertainties, and implications for climate aware procurement and system design.
Introduction
Satellite connectivity is increasingly viewed as a complement to terrestrial networks, especially for rural and remote coverage (low density population area aways from already installed connectivity), where fiber cannot be deployed because of high cost of civil engineering and where fixed mobile access is not available, in addition to mobility, and resilient backhaul. Yet the carbon implications of choosing one orbital architecture over another remain difficult to compare. Public data on spacecraft manufacturing is sparse, launch emissions vary by vehicle and propellant, and ground segment configurations differ across deployments. LEO systems use many smaller satellites at low altitude, launched frequently and replenished on relatively short cycles, whereas GEO systems use a handful of large spacecrafts at geostationary altitude that operate for more than a decade and a half. These structural differences carry consequences for embodied carbon, upgrade cadence, and the allocation of impacts per unit of delivered service.
Scopes of the comparison and methodology
The comparison presented here covers embodied emissions from three key phases: spacecraft manufacturing (including structures, solar arrays, batteries, electronics and payload subsystems), launch activities, and ground-segment manufacturing (gateway sites and user equipment). The scope also includes operational emissions generated by ground gateways and user terminals during the exploitation phase. However, the analysis excludes in-orbit operations, end-of-life handling, and non-CO2 high-altitude effects that go beyond standard emission factors. Furthermore, these impacts may differ significantly between LEO and GEO systems due to variations in terminal types, duty cycles, and link budgets.
Two complementary approaches dominate current assessment practice. For GEO, published lifecycle assessments provide whole‑spacecraft figures that combine manufacturing and launch, typically amortized over 15 years. For LEO, where public spacecraft Bills of Materials (BOM) and detailed manufacturing data are limited, analysts often combine a BOM‑based approach with launch accounting. In a representative LEO analysis reflecting early‑2020s constellations, spacecraft manufacturing is estimated using material and subsystem factors (for example, aluminum on the order of 18kgCO2e per kilogram; solar modules on the order of 1 tCO2e per square meter; battery impacts varying by chemistry and supplier), while launch emissions are attributed using vehicle‑specific factors, with kerosene‑based medium‑lift missions typically on the order of two kilotonnes CO2e per launch. Ground‑segment manufacturing impacts are estimated from typical gateway counts and user equipment bills of materials. To enable comparison, results are annualized and, where possible, normalized to the number of subscribers for the considered constellations.
Although fiber appears to emit less CO2 than satellites, the actual carbon footprint for rural fiber deployment could be significantly higher due to extensive civil engineering works, especially for greenfield installations.
Results
The study shows that ground segment (ground stations and user terminals) is the largest contributor, accounting for 43% of this total, followed closely by the space segment at 40%, and the launch segment at 17%. A detailed breakdown of the ground segment reveals that emission from user satellite kits is the dominant source, representing 38% of the total constellation emissions, while the ground stations contribute the remaining 5%.
Notably, these results challenge the common perception that launch activities are the primary source of emissions; in this analysis, the space segment’s footprint is more than double that of the launch segment. This profile of emission reflects the LEO system architecture. Furthermore, segment-specific analysis indicates that operational energy consumption is the primary driver for the ground segment, accounting for 87% of its emissions. For the launch segment, propellant combustion is the most significant factor, responsible for 85% of its total. Order‑of‑magnitude totals for large LEO constellation fall in a more than five hundred of ktCO2e per year when aggregated across embodied and operational emissions across all three segments, with the exact values depending on the constellation size, replenishment rate, vehicle class, material choices, and, critically, the total number of user terminals.
For GEO, published assessments commonly report around 56 ktCO2e per spacecraft over a 15‑year lifetime, implying approximately 3.7 ktCO2e per year when annualized. Because a small number of GEO satellites can cover broad regions with stable capacity, embodied emissions per unit of geographic coverage can be low; however, the long refresh cycle slows the adoption of lower‑impact materials or higher‑efficiency payloads. Ground station manufacturing in GEO‑centric networks varies by application but is often smaller than the spacecraft footprint when amortized over long service lives.
A per‑customer perspective illustrates the importance of allocation choices. For large LEO constellation, we reported a roughly 130kgCO2e per subscriber per year but it is matter to illustrate that it can range from single‑digit to greater than hundreds kilograms CO2e per customer‑year, depending on the number of subscriber served by the constellation, on how launch emissions are allocated, the assumed utilization, and the treatment of user equipment. This footprint is primarily attributed to the constellation itself with approximately 60% and the subscriber equipment with almost 40%. Conversely, in GEO constellation, emissions are more supported by user equipment with 70% from the total emission estimated at roughly 39 kgCO2e per subscriber per year.
Annual per subscriber carbon footprint (kgCO2e) of LEO, GEO and fiber connectivity.
Compared to terrestrial broadband, the LEO constellation’s annual per-subscriber emission is substantially higher than the French national average for a fiber[1]. However, this direct comparison is misleading for two key reasons. First, the satellite system’s total emissions are allocated across a much smaller subscriber base. Second, and more critically, national fiber averages often include low impact “brownfield” deployments. For France, the fiber rollout had indeed a controlled environmental impact by massively reusing the existing copper network’s infrastructure—including many thousand kilometers of underground ducts and millions utility poles. This strategy avoided the heavy carbon-intensive civil engineering required for a new build.
Satellite systems, by contrast, are designed for remote regions with low population density and no pre-existing infrastructure. In these “greenfield” areas, deploying fiber from scratch would have a noticeable carbon impact, making satellite more competitive. However, this advantage comes with significant trade-offs for the end-user, namely lower throughput and higher latency, particularly with GEO satellites. A fair comparison between total satellite emissions, including terrestrial equipment, and fiber should indeed account only for remote rural connections and deployed fiber infrastructures in that area.
Besides, the environmental comparison between satellite and fiber yields starkly different results depending on the metric. On a per-Gigabyte basis, fiber is far more efficient due to its superior bit rate. Conversely, when emissions are normalized by covered area (per km²), satellite becomes more environmentally sound choice, but only on the condition that this area is populated and requires service. The advantage lies in efficiently connecting people, not just covering territory.
The absence of harmonized normalization standards presents then a significant challenge in assessing the environmental impact of satellite constellations. Depending on the assumed number of subscribers, calculated emissions per customer can fluctuate by more than an order of magnitude. Consequently, metrics such as kgCO2e per subscriber, per Gigabyte delivered, or per square kilometer covered yield inconsistent and often incomparable results. For methodologies’ harmonization issues, European commission spearheaded, in December 2025, an open consultation for a Product Environmental Footprint Category Rules (PEFCR) initiative, with the aim of establishing a uniform and systematic methodology to calculate carbon emissions from spatial activities. Orange participated in this consultation. In parallel, a Space Act is in discussion at the European Union level.
This analysis clearly confirms that the LEO architecture results in a substantially higher emission profile than both GEO systems and terrestrial fiber networks. This is, of course, an inherent consequence of the LEO model, which relies on a large number of launched satellites with short operational lifespans requiring frequent replenishment. This issue is poised to become a more significant environmental concern as LEO constellations will be deployed for next decades in lower orbits (e.g., 350 km), where increased atmospheric drag will shorten satellite lifespans and further accelerate the manufacturing and launch cycle. Although the end-of-life phase is beyond the scope of this study, it is important to note that recent LEO satellites raise concerns regarding space debris pollution. So, moving to very low orbits may mitigate long-term space debris accumulation, but potentially at the cost of increased carbon emissions on earth.
Limitations of the study and recommendations for selecting Orbital technologies
It is important to acknowledge that the precision of these results is subject to uncertainties. These stem primarily from two sources: (i) the scarcity of public data and the inherent variability in emission factors, and (ii) the simplifying assumptions required to model the complex LEO system components. While certain emissions were necessarily excluded from the scope of this analysis, these limitations highlight the critical need for greater industry transparency and standardized data to refine future assessments and guide sustainable development in the space sector.
From telecommunication operator point of view, connectivity from LEO or GEO satellites is viewed not as a replacement but as a vital complement to terrestrial networks. Selecting an orbital satellite actor through a climate lens should start with a clear statement of service requirements—latency, mobility, coverage, and capacity—and proceed to a normalized, scenario‑based comparison of embodied and operational impacts.
[1] ADEME document, “Evaluation de l’empreinte environnementale de la fourniture d’accès à internet en France”. Action collective de mise en œuvre du RCP Fourniture d’Accès à Internet, 2024.
The Impact of the LEO Satellite Lifecycle on Current Orbital Congestion :
Orbital congestion at around 550 km is caused by obsolete (abandoned) satellites and/or their debris, which continue to drift randomly in the thermosphere before being completely vaporized in the mesosphere after about ten years.
This is prompting players in the space sector to move closer to Earth, which reduces satellite lifespans to about 2 to 4 years, instead of more than 5 years. This phenomenon leads to an increased need to replace constellations, thereby increasing their carbon footprint, in addition to the potential impact on the ozone layer due to the vaporization of satellites near that layer.
