Orbital Urbanism and the Infrastructures of Urban AI Synthesis

I saw two shooting stars last night
I wished on them but they were only satellites
Is it wrong to wish on space hardware
I wish, I wish, I wish you’d care

Song Text by Billy Bragg, 1983

‘How do you make a city happy with the help of satellites?’¹ESA Space Solutions, “Urban Future 2026 in Ljubljana, Slovenia | 24 – 27 March | ESA Space Solutions,” March  2026, 37 sec., https://www.youtube.com/watch?v=PpidA7G3Ts8. was a question posed at the Annual 2026 Urban Futures Conference. It was asked by a member of the European Space Agency’s (ESA) business arm. ESA attended the conference to engage with start-ups and city decision-makers about how space technologies are game-changers for cities. That sentiment is summarised in a quote on ESA’s website:

“From improved traffic flow and autonomous vehicles to digital twins for enhanced urban planning, water source mapping and air quality monitoring, space is driving change in our cities, and the potential of these technologies is huge.”²“The Quest for Happy Cities – Does the Answer Lie in Space?,” The European Space Agency, accessed March 30, 2026, https://business.esa.int/news/quest-for-happy-cities-does-answer-lie-space.

What a sales pitch! Not only that, the increasing interlinking of the urban and the orbit merits closer attention in making sense of the infrastructures of urban AI synthesis. As the editors of this special section of Mediapolis note, the synthetic city is not only one that increasingly relies on data and synthetic artefacts, but it is also shaped by a synthesis between digital infrastructures and the urban lived environment. As I contend below, making sense of urban AI synthesis requires accounting for the nexus between urban processes and Earth-orbiting satellites.

The times when songwriters like Billy Bragg were compelled to sing about “space hardware” after the rare sighting of two paralleling satellites³Dave Simpson, “How We Made: Billy Bragg’s a New England,” The Guardian, January 24, 2022, https://www.theguardian.com/culture/2022/jan/24/how-we-made-billy-braggs-a-new-england-john-peel-biryani. are long gone. Depending on the level of light pollution, that sight is now pretty normal. There are many visualisations showing how earth is surrounded by what looks like a beehive of satellites and space debris.⁴Frederick O’Brien, Ashley Kirk, and Oliver Holmes, “‘This Feels Fragile’: How a Satellite-smashing Chain Reaction Could Spiral Out of Control,” The Guardian, May 28, 2026, https://www.theguardian.com/science/ng-interactive/2026/mar/31/this-feels-fragile-how-a-satellite-smashing-chain-reaction-could-spiral-out-of-control. The sustainable use of low Earth orbit (LEO) is a hot topic in the space industry. LEO, the space around Earth where most ‘space hardware’ entangled with urban processes are located, is crowded. LEO is also where Earth observation data is captured, and it provides the backbone for navigation and communication technologies on Earth.

Earth observation (EO) applications are routine in emergency response and have long been used by activists in uncovering ground-level atrocities.⁵“The Architecture of Genocidal Starvation,” Forensic Architecture and World Peace Foundation, August 22, 2025, accessed May 6, 2026, https://forensic-architecture.org/investigation/aid-in-gaza. Maybe less known are more everyday applications. Looking through ESA’s archives for use case studies focused on urban applications shows a breadth of more mundane examples. To highlight just two: in 2021, AgeSpot was a feasibility study combining census and EO data to produce spatially explicit mappings of socio-economic parameters.⁶“AgeSpot” All Themes, ESA Space Solutions, April 21, 2021, accessed May 22, 2026, https://business.esa.int/projects/agespot. A more recent project, AERIUS, explores monitoring urban ecosystems to track changes in blue and green infrastructure.⁷“AERIUS” All Themes, ESA Space Solutions, May 8, 2026, accessed May 22, 2026, https://business.esa.int/projects/aerius. Both can conjure images of positive monitoring as well as frightening surveillance – the latter, not least, because of EO’s origins in the military complex.⁸Mia M. Bennett et al., “The Politics of Pixels: A Review and Agenda for Critical Remote Sensing,” Progress in Human Geography 46, no. 3 (January 27, 2022): 729–752, https://doi.org/10.1177/03091325221074691. Increasingly, applications are moving from case studies to applied practice, with cities experimenting with EO data for planning tasks and asset management, such as assessing whether quay walls need maintenance.⁹Arjan A. M. Venmans et al., “Reliability of InSAR Satellite Monitoring of Buildings Near Inner City Quay Walls,” Proceedings of the International Association of Hydrological Sciences 382 (April 22, 2020): 195–199, https://doi.org/10.5194/piahs-382-195-2020. As the ESA quote above highlights, these LEO infrastructures are also increasingly used to ensure the supply of vital resources such as water.

The urban/orbit nexus is deeply entangled with artificial intelligence (AI). Taking EO as my key example, it can be noted that many steps of synthesis via algorithmic systems are at play between data recording and its use. Applications range from simple automated queries to more advanced, often deep learning-based applications that extract semantic detail from large spatiotemporal image databases, often via so-called geo-embeddings that turn complex spatiotemporal data into vectors that can be processed by machine learning models. Before EO data reaches earth, the AI synthesis chain starts with deployment and space-side AI. Not for nothing is the name of ESA’s 2026 academic conference “SPAICE.” Yes, Space but with AI. That conference will focus on interdisciplinary discussions at the interface between AI and space.¹⁰“SPAICE 2026: AI in and for Space,” European Space Agency, 2025, accessed May 22, 2026, https://spaice.esa.int/2026/. While not limited to urban or even earth-bound applications, the conference will highlight new ways in which data generated by space infrastructures will be mediated by algorithms – something that has been experimented with for a few years. For example, surveillance giants like Palantir, as well as the broader research community, have been advancing Edge AI for satellite data.¹¹“MetaConstellation: Harness the power of satellite constellations to empower decision-makers on Earth,” Palantir, accessed May 5, 2026, https://www.palantir.com/offerings/metaconstellation/. Where Edge AI refers to the prefiguration of data at its source (in this case, the satellite). For EO Edge AI is supposed to facilitate smoother and more effective data downloads.¹²Ning Ruan et al., “Edge AI for Earth Observation,” IEEE Internet Computing 29, no. 3 (May 1, 2025): 31–40, https://doi.org/10.1109/mic.2025.3587325. Despite enthusiasm about the potential of earth observation for urban applications, there are still pesky things like clouds and trees that block the view of EO satellites. More efficient data downloads would include only useful data, likely with technical rather than social parameters determining what constitutes useful.

How satellites see is a matter of multiple steps of computational limiting, as well as the specificity of the EO instruments and types of satellites used. There are now many different types of EO satellite missions, those that orbit Earth but also geo-stationary ones, all carrying different observation tools that are more or less useful for observing different earth-bound phenomena. This diversity makes for versatile data and analytical processing. Once EO data is available, advanced machine learning approaches are used for faster pattern recognition and near-real-time probabilistic predictions. In other words, such AI applications are now key to making EO data actionable. Analytical approaches travel as institutions within a single location, using specific types of EO data, and learn from others elsewhere. One example is research on developing deep learning-enhanced computer vision methods using geostationary satellites, in which European weather warnings build on US advances in the field.¹³“A geostationary satellite AI application for severe weather warnings (MSO project),” KNMI, accessed May 5, 2026, https://www.knmi.nl/research/satellite-observations/projects/a-geostationary-satellite-ai-application-for-severe-weather-warnings-mso-project. As Lappe, Calovi and Winther highlight, earth observation and remote sensing practices are always influenced by technical choices and approaches to image interpretation.¹⁴Ronja Lappe, Martina Calovi, and Hannah Winther, “The View From Somewhere: What Remote Sensing Still Has to Learn About the Satellite Gaze,” Progress in Physical Geography Earth and Environment 50, no. 2 (December 24, 2025): 266–280, https://doi.org/10.1177/03091333251409690. Notably, technical and analytical choices are also always already prefigured by available resources, technological developments, but also by politics. For example, geopolitical tensions play a key role in Taiwanese decisions about the types of sensors built into their space programme.¹⁵Yi-Ting Chang, “Engineering the State: Taiwan’s Satellite Production Amid Geopolitical Tensions,” Territory Politics Governance, April 30, 2026, 1–16, https://doi.org/10.1080/21622671.2026.2651322.

Entanglements in the urban/orbit nexus are deeply seeped in a view from somewhere. That somewhere is not easy to pin down within complex EO data assemblages. This insight is even more acute if we explicitly acknowledge that routine space infrastructure development is driven by public-private partnerships, rather than the remit of public institutions working exclusively on public-interest technologies. In light of the new space age, where space technology development is propelled by private equity, it has become much more feasible to deploy smaller satellites with ever-evolving onboard technologies. Predictive successes and, increasingly, political priorities shape which observation approaches and technologies prevail, and ultimately what data is available for urban applications and how that data is analysed.

In extension, what can be sensed is more likely to become the object of interventions, sidelining approaches that look at processes that cannot be remotely sensed.¹⁶Fran Meissner and Karin Pfeffer, “From Remote Monitoring to Local Action,” Nature Cities 2, no. 11 (August 8, 2025): 1020–1021, https://doi.org/10.1038/s44284-025-00304-z. Such shifts matter as the chain of AI mediations continues once data is integrated into local systems. Urban monitoring and planning support tools, including those described as Urban Digital Twins, increasingly rely on advanced satellite image analysis methods, which, as noted, are made possible with AI. As Borbach and colleagues note, tools like digital twins are a byproduct of making cities and how they are governed accountable via data rather than political responsibility.¹⁷Christoph Borbach, Wendy Hui Kyong Chun, and Tristan Thielmann, “Making Everything Ac-count-able: The Digital Twinning Paradigm,” New Media & Society 27, no. 8 (August 1, 2025): 4369–4384, https://doi.org/10.1177/14614448251338289.

The algorithms at play in such systems are not only those relevant for detecting patterns in EO data, but also a wider range of what has become known as GeoAI – advanced machine learning approaches applied to a variety of spatial data. The types of approaches and computational methods used are manifold and include automated image segmentation for geo-fencing applications, object detection, and learning-algorithm-based urban layout generation. In addition, research now frequently calls for the integration of large language model-based generative AI into planning support tools (e.g., ChatGPT, Claude). Arguably, the latter is what most of the public thinks of when they think of AI. It is also the AI that they may interact with via chatbots. At the same time, other modes of advanced machine learning that link the urban with the orbit blend into the background and are less known or even acknowledged. As Simon Marvin, discussing the relevance of weather data for urban geography, eloquently argues, we are witnessing a shift in ‘the distribution of visibility, care, and anticipatory capacity’ in the ways EO data comes to matter in cities.¹⁸Simon Marvin, “Smart Weather: Why Does It Matter for Urban Studies?,” Urban Geography 46, no. 7 (July 9, 2025): 1630–1641, https://doi.org/10.1080/02723638.2025.2527131.

If proponents of AI applications in urban EO are to be believed, by “leveraging AI, a range of alternative approaches to urban planning and design processes can be introduced, resulting in more effective and targeted solutions to the challenges faced in current urban development.”¹⁹Qihao Weng et al., “How Will Ai Transform Urban Observing, Sensing, Imaging, and Mapping?,” Npj Urban Sustainability 4, no. 1 (November 28, 2024), https://doi.org/10.1038/s42949-024-00188-3. Such statements tend to focus on integrating different types of remote sensing data (e.g., drone imagery) with other data sources through an extensive stack of technologies and multiple algorithmic translations. Within this chain, the link to LEO-based infrastructures plays a key role in a reconfiguration of urban space as machine (and satellite) readable. As the various examples pointed out earlier attest, EO applications are increasingly becoming part of urban infrastructure, supporting the governance of cities and their resources. Urbanism – simply defined as how individuals interact with the city’s infrastructure – increasingly has an orbital dimension. It does not seem far-fetched to claim that we are witnessing a rise of orbital urbanism.

To clarify, orbital urbanism, as I use it, is not about cities in space. Previous uses of the term may have focused on space colonisation and dreamt up self-sufficient cities floating through space. Cities like those housed in the Bernal Sphere, as the cover-image artist rendering alludes to. Even as those visions are seeing a revival, with various space powers racing to establish a permanent base on the moon, I contend that a more immediate urgency lies in conceptualising orbital urbanism as concerned with the cities we already inhabit. Doing so starts with the proactive acknowledgement of urban/orbit entanglements, charting how those entanglements are changing and what the governance implications of those changes are. The idea of orbital urbanism, as used here, is also not solely about understanding how satellites look down on cities. It is also about looking up and acknowledging that the geography of LEO matters for the geographies of cities. There is limited space in Earth’s orbit, and the promise of happier cities globally may hinge on decisions about how that space is used.

There are geographic differences in the prevalence and configurations of orbital urbanism. Data-sharing agreements and the ability to influence the design and deployment of satellite infrastructures systematically differ by global regions. Less developed countries increasingly have space ambitions but remain marginal actors, vulnerable to uneven availability of EO data, especially at higher resolutions.²⁰Florian Rabitz, Inga Popovaitė, and Vidas Vilčinskas, “Satellite Infrastructures, Earth-space Sustainability, and the Global South,” Earth System Governance 28 (March 14, 2026): 100323, https://doi.org/10.1016/j.esg.2026.100323. The geopolitical power of cities consequently often determines if and how EO applications are being used. Considering more advanced dependencies within the urban/orbit nexus has to go hand in hand with a better understanding of the infrastructural power of orbital technologies. As Keller Easterling argues, it is necessary to systematically trace if and how processes of extra-statecraft are at play by making visible the processes that order the influence of orbital urbanism on build and social spaces in our cities.²¹Keller Easterling, Extrastatecraft: The Power of Infrastructure Space (Verso Books, 2016). Doing so would be a first step in raising awareness of, and lobbying for, a more critical engagement with how cities are being recast through and with orbital technologies, including through the layering of AI synthesis. What such an increased engagement would look like is subject to speculation, but with a number of EO projects pushing for making data and analysis accessible to local disadvantaged populations, we may see first shifts in that relationship.²²“Imago – SDR,” Imago – SDR, n.d., https://imago.ac.uk/.

If the number of urban EO research articles and start-ups is anything to go by, it is safe to assume that the number of urban applications taking advantage of space technologies has not yet reached its peak, nor has the number of satellites orbiting Earth. Such developments are only rarely apparent to urban dwellers unfamiliar with the many infrastructure/AI translations sketched out above.  Earth observation and other satellites influencing urban processes and governance are so far removed from our experiences of the quotidian city that they are not so much sunk in but silently hovering components of larger urban information infrastructures. Yet the way they are developed, the data they produce, and the orbits they circulate on increasingly shape the everyday experiences of urban inhabitants.

The entanglements highlighted in this short paper are an attempt at starting to trace the links between the urban, specifically urban planning, and the orbit. The aim is to offer an exploratory analysis that highlights avenues for recognising and delineating the infrastructural power of urban earth observation. Returning to the song lyrics that I used to open this paper, we are left with the question of whether it is wrong to wish on space hardware. I am not sure this paper can answer that question. I do caution that we may want to start wishing on space hardware if, as the ESA employee suggested, urban happiness depends on it. As developments stand, it will be particularly important that urban happiness should not be the foray of some cities well placed in the global hierarchy of cities. Happiness surely should be more equitably distributed. Further visibilising the orbit/urban nexus and theorising orbital urbanism, in my eyes, is a needed step towards achieving that.

Fran Meissner

Fran Meissner is an Assistant Professor of Critical Geodata Studies and Geodata Ethics at the University of Twente in the department of Planning, Land and Urban Systems, in the Netherlands. Her research focuses on contemporary urban social configurations and how these are transformed through datafication and urban technologies – technologies in cities and those in the earth’s orbit. She has previously been an EUI Max Weber Fellow and a Marie Skłodowska Curie Fellow, and has worked on various international research projects.