Innovation is a key determinant of the future trajectory of the space economy, shaping the development of new technologies, applications and capabilities. Measuring space innovation, however, requires a multi-dimensional approach, as no single indicator captures its full scope. This chapter combines complementary evidence from scientific publications, patent data and selected technology developments to explore the changing geography, intensity and direction of innovation in the space economy. It provides granular insights into research capabilities, collaboration networks and emerging scientific fields, as well as inventive activity and technological development. It further seeks to identify early signals of emerging technologies advances that may not yet be visible in bibliometric or patent statistics.
The Space Economy at a Glance 2026
Abstract
Scientific publications provide a valuable window into the research base underpinning space innovation. They offer insights into the scale and direction of scientific activity, the development of research capabilities, patterns of international collaboration, and the emergence of new scientific fields and technological opportunities. While publications do not directly measure innovation outcomes, they can provide early signals of evolving knowledge frontiers and help identify areas where future technological advances may emerge. For policy makers, they provide an early view of emerging fields that may shape future technological development and competitiveness in the space economy.
Space-related publication has become an increasingly dynamic field of scientific production. Since 2009, space-related scientific output of publications in journals and conference proceedings has more than doubled, growing more quickly than broader related scientific fields such as engineering and “earth and planetary science” and total scientific publication activity, pointing to growing interest and capabilities in this area. This expansion reflects the growing strategic and economic relevance of space, the wider availability of satellite data and signals, and the development of new missions and applications across earth observation, communications, navigation, space science, exploration and downstream data services (Figure 3.1).
Indexed growth in fractional counts for space literature and selected ASJC codes, 2010=100
Note: the number of fractional counts will be affected by its associated number of international author affiliations and Scopus All Science Journal Classification (ASJC) codes. A paper with only one country affiliation and ASJC code equals 1, whereas multiple ASJC codes and international affiliations reduces a paper’s relative weight in the total count.
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
Space remains a specialised but growing field within the broader scientific landscape. In 2024, space-related publications accounted for 1.3% of total fractional publication counts in the OECD area. This modest share, alongside sustained growth, points to a niche yet expanding area of scientific activity. Growth is driven by the expansion of national space programmes, wider use of space-based data, and a broader set of actors involved in missions and downstream services. It also reflects the cross-disciplinary nature of space research, spanning astrophysics, aerospace engineering, Earth and planetary sciences, remote sensing, data science, materials, robotics and communications.
Global space-related scientific output has increased over the past 15 years, alongside a marked shift in the geography of research (Figure 3.2). Emerging economies now play a larger role, with China and India contributing significantly to growth in space-related publications and to a more multipolar innovation landscape.
Annual counts of journal papers and conference proceedings fractionalised by ASJC category and country affiliation
Notes: BRIICS refers to Brazil, Russia, India, Indonesia, China and South Africa; Fractional counts depend on the number of international author affiliations and Scopus All Science Journal Classification (ASJC) codes associated with a publication. A paper with one country affiliation and one ASJC code receives a weight of 1, while papers with multiple country affiliations or ASJC codes receive a lower relative weight in the total count.
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
In 2024, China accounted for the largest share of space-related scientific literature, with 40.8% of global fractional counts. Space has been a recurring priority in China’s medium-term planning, with successive space white papers outlining objectives spanning launch capabilities, satellites, human spaceflight, lunar and deep-space exploration, earth observation, navigation, communications and space science. These ambitions have created sustained demand for scientific and engineering expertise across a wide range of disciplines, including aerospace engineering, planetary science, remote sensing, telecommunications, robotics and materials science. The rapid growth of China’s space-related scientific literature therefore reflects more than an increase in publication volumes. It also points to the expansion and diversification of national capabilities associated with major space missions and the development of supporting technologies.
The OECD area continues to represent a substantial share of global space-related publications, but its relative contribution has declined. Between 2009 and 2024, the OECD area’s share of global space-related fractional counts decreased from 73.3% to 46.9%.
Citation-based indicators provide a complementary perspective on scientific impact and research excellence. The OECD area remains a leading contributor to highly influential space research, accounting on average for 12% of the world’s top-cited space-related publications (Figure 3.3). Several OECD countries also demonstrate particularly research excellence relative to their overall scientific output, producing results with high international visibility and impact. In 2024, Denmark recorded the highest share of space publications among the world’s top 10% most-cited publications (22.9%), followed by the Netherlands (17.9%), United Kingdom (16.3%) and Switzerland (15.5%).
These indicators point to two complementary dimensions of scientific performance. Large producers such as China and the United States account for a high share of the world’s most-cited space literature in absolute terms. Smaller and highly internationalised research systems, many of them in the OECD area, such as the Netherlands and Switzerland, often show stronger performance when excellence is measured relative to their own publication output.
Scientific excellence measured as the share of documents that are among the top-10 most highly cited
Note: Panel A shows the publication-weighted average share of each regional group’s space publications (fractional counts) that rank among the world’s top 10% most-cited papers. Countries with larger publication outputs carry greater weight in the average. Higher values indicate a larger share of highly cited space research. In both panels, the analysis covers economies with at least 100 fractional-count publications.
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
International collaboration appears to play an important role in scientific excellence (Figure 3.4). Countries with higher levels of international co-authorship tend to perform well in citation-based indicators, reflecting the wider visibility, dissemination and potential impact of collaborative research. Citation indicators should nevertheless be interpreted with care, as citation patterns vary across fields, publication types and over time, and ratio-based indicators can favour countries with smaller publication volumes.1
Share of an economy’s papers in top-10 percentile for citations (x-axis) and internationally co-authored (y-axis)
Note: The analysis includes economies that have produced 100 or more fractional document counts.
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
International co-authorship indicates scientific collaboration and is often linked to higher research quality and visibility. In space-related literature, the number of participating countries rose from 71 in 1996 to 119 in 2024, while active bilateral links more than tripled, from 566 to 1 958. Over the past 28 years, authors from countries engaged in space research worked, on average, with three times as many international partners (Figure 3.5).
At country-level, the United States consistently recorded the largest number of international partners in space research, rising from 52 unique partners in 1996 to 101 in 2024. In an aggregate measure of research intensity between bilateral country partners, the United States maintains the highest ranking in every year across the time period. This suggests the United States is engaged in broader and more intense international collaboration in space research than all other countries. Other notable countries in this regard include France, Germany, and the United Kingdom which all appear in the top-5 countries for total number of partners and total research intensity in all years between 1996 and 2024.
Lower quartile, median, and upper quartile values from count of distinct bilateral partners per country per year
Notes: The chart shows, for each year, the distribution across countries of the number of distinct international partner countries in space-related publications, using the lower quartile, median and upper quartile. The dataset covers eligible peer-reviewed articles, reviews and conference papers from a predefined set of “pure space” journals. Publications are assigned to country pairs based on author affiliation countries recorded in citations. Only country pairs with at least 10 co-authored publications over the period are included, and affiliation countries are counted only for authors appearing in at least two publications in the dataset.
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
The importance of international co-authorships in a country’s total publications of space-related literature however varies widely across countries. In 2024, among countries producing more than 100 space-related fractional counts of papers, the share of publications involving international co-authorship ranged from 82.8% in Chile to 14.8% in China (Figure 3.6). These differences reflect several factors, including the size of national research communities, countries’ areas of specialisation and their integration into international research networks. Smaller or more specialised research systems may rely more heavily on international collaboration, particularly in fields that require access to specific infrastructure, data or locations. Chile is a case in point: its strong specialisation in astronomical observations is supported by favourable geographic and climatic conditions, as well as by the presence of major international observatories.
Share of fractional counts with international co-authorship, 2024
Notes: International co-authorships are based on the affiliation of the authors. The analysis includes economies that have produced 100 or more fractional document counts. Multiple All Science Journal Classification (AJSC) codes can be assigned to each journal.
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
These collaboration patterns are important for policy makers because space research relies on shared infrastructure, international missions, data access, distributed expertise and long-term institutional partnerships. A key policy challenge will be to preserve openness while managing security risks, resilience needs and strategic dependencies in space research and innovation.
Bibliometric indicators based on academic journal articles and conference proceedings provide useful insights into countries’ scientific capabilities, collaboration patterns and emerging research fields. The analysis draws on 254 space-related journals and conference proceedings indexed in Elsevier’s Scopus database, selected using journal names, Scopus All Science Journal Classification codes and the SCImago Journal Rank. The dataset covers 491 496 fractional counts of scientific papers published between 2009 and 2024, referred to here as “space literature”. Fractional counts are used to attribute documents across fields and country affiliations. This approach improves comparability, but results should be interpreted with care, particularly for multidisciplinary journals and internationally co-authored publications. Publication volumes, co-authorship, citations and research themes can help identify international collaboration networks, areas of research excellence and emerging topics of interest to policy makers. However, publication behaviour, citation practices, journal coverage and the growing use of generative artificial intelligence may affect cross-country and comparisons over time.
Scientific publications can provide early signals of emerging innovation fields. Multiple factors can influence thematic trends in scientific literature, including technological developments, access to new data, government funding strategies and broader societal trends.
When looking at trending topics in space literature since 2015, remote sensing change detection and remote sensing semantic segmentation2 stand out in terms of novelty and growth in the number of associated papers (Figure 3.7). This coincides with the increasing use of digital technologies to process and analyse satellite imagery. Other significant topics include atmospheric pollution and aerosols, lightning and terrestrial gamma flashes, and coastal and marine environment mapping, corresponding with improved space-based sensors and the broader deployment of earth observation satellites.
These trends point to the growing importance of data-intensive space innovation. Advances in sensing technology, satellite revisit times, cloud computing, artificial intelligence and access to large open datasets are expanding the analytical uses of satellite imagery and signals. This creates opportunities for environmental monitoring, climate science, disaster management, agriculture, security and commercial data services.
Indexed growth in papers, 2015=100
Note: Topics identified using a BERTopic influenced topic model (cluster size 750).
Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 1.2026, https://www.elsevier.com/products/scopus, April 2026.
Patent data provide a useful, albeit partial, proxy for measuring innovation in the space economy. They capture codified technological advances and make it possible to track trends in inventions across countries, organisations and technology domains. However, not all forms of space innovation are covered by patents, particularly where knowledge is kept confidential or where advances are embedded in system integration, software, operational processes or mission design.
Space-related patenting has increased steadily since 2000, pointing to sustained growth in technological development across the sector (Figure 3.8). Patenting activity has risen across six major technological fields: propulsion, spacecraft electrical power, space system control, life support, space debris and quantum technologies for space applications. A slight decline has been observed since 2019 in most fields.
Number of patents, by patent offices and priority date
Note: Partial information for patent applications from priority year 2022. IP5 patent families correspond to patent families filed in at least two offices worldwide, including at least one of the Five IP largest offices (IP5, i.e. the European Patent Office, EPO; the Japan Patent Office, JPO; the Korean Intellectual Property Office, KIPO; The China National Intellectual Property Administration, CNIPA; and the US Patent and Trademark Office, USPTO). Patents related to space-related technologies have been identified by the EPO using the EPO-PATSTAT 2025b data (https://www.epo.org/patstat).
Source: OECD (2026[1]), STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, June 2026.
The growth in space-related patenting has been driven primarily by advances in spacecraft electrical power and space system control technologies (Figure 3.9). These two fields account for a large share of patent applications in the sample, reflecting the importance of energy generation, storage and management, as well as guidance, navigation and control capabilities, for increasingly complex space missions. In 2023, spacecraft electrical power alone represented 46% of space-related patent applications, underscoring its central role in contemporary space innovation. The period also saw the emergence of newer technology areas, including quantum technologies for space applications and space debris-related technologies. Although still smaller in volume than more established fields, these areas point to evolving innovation priorities, linked respectively to next-generation sensing, communications and timing applications, and to the growing need to monitor, mitigate and manage orbital debris.
Note: Partial information for patent applications from priority year 2022. IP5 patent families correspond to patent families filed in at least two offices worldwide, including at least one of the Five IP largest offices (IP5, i.e. the European Patent Office, EPO; the Japan Patent Office, JPO; the Korean Intellectual Property Office, KIPO; The China National Intellectual Property Administration, CNIPA; and the US Patent and Trademark Office, USPTO). Patents related to space-related technologies have been identified by the EPO using the EPO-PATSTAT 2025b data (https://www.epo.org/patstat).
Source: OECD (2026[1]), STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, June 2026.
The United States remained the leading source of space-related patenting activity in 2020-23, accounting for 34.2% of applications. It was followed by the EU27, Japan and China (Figure 3.10).
IP5 patent families, by priority date and applicant’s location, using fractional counts
Note: Partial information for patent applications from priority year 2022. IP5 patent families correspond to patent families filed in at least two offices worldwide, including at least one of the Five IP largest offices (IP5, i.e. the European Patent Office, EPO; the Japan Patent Office, JPO; the Korean Intellectual Property Office, KIPO; The China National Intellectual Property Administration, CNIPA; and the US Patent and Trademark Office, USPTO). Patents related to space-related technologies have been identified by the EPO using the EPO-PATSTAT 2025b data (https://www.epo.org/patstat).
Source: OECD (2026[1]), STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, June 2026.
This distribution points to the continued concentration of space-related inventive activities in economies with long-standing space programmes, large public research and defence capabilities and well-established industrial ecosystems.
At the same time, the geography of space-related innovation is gradually broadening. Several economies considerably increased their share of total space-related patenting activity over the period, notably Australia, China, Finland, India, Poland, Spain and the United Kingdom. These developments reflect a combination of national space strategies, participation in international programmes, specialised firms and research centres, and rising demand for space-based applications in connectivity, earth observation, navigation and security. Space-related patenting is also concentrated in a limited number of leading innovation regions, although their importance within national patenting activity has increased over time. California recorded the highest number of space-related patents over 2020-23, reflecting the region’s strong concentration of space firms, research organisations and technology capabilities (Figure 3.11). It was followed by Southern Kanto, Occitanie, Chungcheong and Colorado, which combine established aerospace industrial bases, specialised research ecosystems and proximity to major public space or defence institutions.
Patent applications filed under the Patent Co-operation Treaty (PCT) by inventor's location and priority date
Note: Data refers to patent applications filed under the Patent Cooperation Treaty (PCT), by the inventor's region at Territory Level 2 (TL2) and priority date. Patents related to space-related technologies have been identified by the EPO using the EPO-PATSTAT 2025b data (https://www.epo.org/patstat ).
Source: OECD (2026[1]), STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, June 2026.
Patent data for 2020-23 suggest distinct areas of space innovation specialisation across major economies (Figure 3.12). European organisations lead in propulsion and quantum-related patents. The United States leads in life support systems, spacecraft electrical power and space systems control. Japan stands out in space debris innovation, illustrated by mission activity related to active debris removal and inspection of non-co-operative objects. These patterns should be interpreted with caution, as relevant innovations may also be taking place in other economies or remain unpublished, unpatented or protected through secrecy for commercial, strategic or security reasons (Olivari, Jolly and Undseth, 2021[2]).
IP5 patent families, by priority date and applicant's location, using fractional counts
Note: IP5 patent families correspond to patent families filed in at least two offices worldwide, including at least one of the Five IP largest offices (IP5, i.e. the European Patent Office, EPO; the Japan Patent Office, JPO; the Korean Intellectual Property Office, KIPO; The China National Intellectual Property Administration, CNIPA; and the US Patent and Trademark Office, USPTO). Patents related to space-related technologies have been identified by the EPO using the EPO-PATSTAT 2025b data (https://www.epo.org/patstat).
Source: OECD (2026[1]), STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, June 2026.
The convergence of technologies can be an important driver of innovation and new applications, as discussed in the OECD Science, Technology and Innovation Outlook 2025 (OECD, 2025[3]). In the space sector, convergence with digital technologies has been particularly impactful over the last decade.
Artificial intelligence (AI) has revolutionised data analysis (see for example OECD (2019[4]; 2023[5]),and it has become increasingly important across the space data value chain. Initially associated mainly with downstream data analysis, AI is now being integrated at earlier stages of data production, including onboard processing, data triage and autonomous prioritisation of downlinks. This is particularly relevant for earth observation missions, where satellite downlinks face bandwidth constraints and growing data volumes. The growing number and sophistication of earth observation sensors, discussed in Chapter 3, is increasing the volume and diversity of satellite data available for AI-enabled processing and analysis.
An emerging application of AI is onboard data processing and triage, allowing spacecraft to autonomously prioritise which data to downlink. Examples include segmenting and masking cloudy pixels, rapidly detecting natural disasters, classifying objects and compressing imagery to reduce downlink volume (Goodwill, Wilson and MacKinnon, 2023[6]). In 2024, the European Space Agency launched the PhiSat-2 technology demonstration satellite, equipped with a multispectral camera and an AI computer able to turn imagery into maps, classify objects, detect wildfires and compress information before transmission.
Machine learning can reduce the need for manual analysis, provide near-real-time access to information and identify patterns for predictive insights. However, many applications still require geospatial expertise, adequate ICT infrastructure and access to high-quality training datasets. These requirements can create barriers to broader uptake, particularly for smaller firms, public administrations and users in countries with limited digital and geospatial capacity.
The emergence of geospatial foundation models could represent a breakthrough for future satellite data analysis. These models may lower barriers to access while strengthening analytical capability. The NASA and IBM-led Prithvi models for earth observation, weather and climate illustrate this trend. The Prithvi-EO 2.0 model is pre-trained on millions of data points from harmonised Landsat and Sentinel-2 datasets and supports applications such as carbon flux estimation, landslide detection, burn intensity estimation, crop pattern identification and flood mapping. Figure 3.13 illustrates the model’s performance across three disaster-response applications, using F1 scores ranging from 0 to 100% to capture the balance between recall, the ability to detect all relevant targets, and precision, the ability to avoid false positives. Prithvi-EO-2.0 performs nearly perfectly in wildfire-scar and flood mapping. Its performance is lower for landslide detection, where it is more effective at identifying landslide areas than at excluding false positives (Szwarcman et al., 2026[7]).
However, there are still challenges associated with automated interpretation of aerial and satellite imagery Figure 3.14 presents the results of a benchmark study assessing the accuracy of 13 vision-language models, which differ fundamentally from geospatial foundation models, across tasks including classification, counting and spatial reasoning (Danish et al., 2025[8]). For the multiple-choice tasks shown, each question had five answer options: one correct answer, one semantically similar alternative and three other plausible alternatives. Random guessing would therefore yield an expected accuracy of 20%. The results show considerable dispersion in model performance, which may partly reflect differences between general-purpose and geospatially specialised models, as well as differences in architecture, training data and task difficulty. Although some models performed well on scene and land-use classification, performance remained close to random guessing for several granular counting and interpretation tasks, such as tree health assessment, general vehicle counting and crop classification.
F1 scores from separate application-specific evaluations
Notes: This represents the best reported results from separate application-specific evaluations; datasets, sample sizes and performance metrics differ across tasks. The F1 score balances false positives and missed detections (precision and recall). A score of 100% indicates that the model detected all relevant targets without producing false positives.
Source: Based on Szwarcman et al. (2026[7]), “Prithvi-EO-2.0: A Versatile Multitemporal Foundation Model for Earth Observation Applications”, Tables 4-13, https://doi.org/10.1109/TGRS.2025.3642610.
Highest, lowest and median shares of correct answers to multiple-choice questions.
Notes: The benchmark study assessed 13 vision-language models, featuring over 10 000 manually verified instructions and spanning diverse visual conditions, object types, and scales. Each question had five response options and one correct answer, yielding a random response baseline of 20%.
Source: Danish et al. (2025[8]), “GEOBench-VLM: Benchmarking Vision-Language Models for Geospatial Tasks”, https://doi.org/10.1109/ICCV51701.2025.00670.
AI-enabled processing therefore links space innovation to broader digital transformation. It can increase the value of satellite data, support new commercial and public-service applications, and help users manage growing volumes of observations. At the same time, it raises questions about data quality, model transparency, validation, access to training datasets and the governance of automated geospatial analysis.
Space exploration is increasingly aimed at enabling sustained robotic and human operations in orbit and on planetary surfaces. This requires advances across a broad set of technologies. The International Space Exploration Coordination Group (ISECG), including some 27 organisations, helps co-ordinate global exploration planning by publishing regularly updated roadmaps of shared objectives, mission scenarios and technology needs (ISECG, 2013[9]; 2018[10]; 2024[11]). Over time, there has been a gradual shift in strategic objectives, both in the destinations and the types of missions envisaged, which also affects technology needs. The Moon is no longer viewed mainly as a stepping stone to Mars, but as a destination for long-term presence. There is also greater emphasis on partnerships with industry and academia, including the use of commercial services such as lunar transport.
The list of critical technologies to mature or develop has been expanded from transportation and life support systems to technologies for long-duration habitation, power generation and in-space communication and navigation (ISECG, 2024[11]). The 2024 Global Exploration Roadmap lists 48 critical technologies distributed across eight exploration areas: propulsion, landing and return; autonomous systems; life support and habitability; crew health and performance; communication and positioning, navigation and timing; power; transversal technologies; and extra-vehicular activity mobility and robotics, as shown in Table 3.1. They vary considerably in terms of technological maturity (for some, adequate testing facilities do not yet exist), reliance on third-party developments (autonomous systems and communications) and regulatory barriers (international trade and exchanges in nuclear fission material and technology are strictly controlled).
Source: Adapted from ISECG (2024[11]), The Global Exploration Roadmap: 2024, https://www.globalspaceexploration.org/wp-content/isecg/GER2024.pdf.
Future space activities may require more resilient and increasingly autonomous power systems, including for operations in orbit, on planetary surfaces and in deep space. Space nuclear power could become an important enabler of such activities. Nuclear technologies have been used in space for more than six decades, mainly through radioisotope power systems providing electricity or heat for missions where solar power is unavailable, insufficient or unreliable (Figure 3.15).
Notes: The payload of a space launch is normally its most valuable part, responsible for achieving the primary objectives of a mission (e.g. satellite, lunar lander, rover, etc.). The data includes nine failed missions: 4 in 1961-70, 2 in 1971-80, 1 in 1981-90, 1 in 1991-2000 and 1 in 2021-23.
Source: Adapted from Krebs (2025[12]). “Nuclear Powered Payloads”, Gunter's Space Page, https://space.skyrocket.de/doc_sat/nuclear.htm, data retrieved 13 October.
Four economies have launched nuclear power systems into space (Figure 3.16). The USSR and later the Russian Federation, and the United States, account for most historical missions, while China and India have more recently used nuclear power systems in lunar exploration missions. Most space nuclear applications have involved radioisotope heater units or radioisotope thermoelectric generators, which use natural heat radiation from selected radioisotopes to generate heat or electricity.
Number of missions with nuclear power sources
Note: Reactor: Nuclear fission reactor; RHU: Radioisotope heater unit; RTG: Radioisotope thermoelectric generator. The data includes nine failed missions involving re-entries: 4 in1961-70, 2 in 1971-80, 1 in 1981-90, 1 in 1991-2000 and 1 in 2021-23.
Source: Adapted from Krebs (2025[12]). “Nuclear Powered Payloads”, Gunter's Space Page, https://space.skyrocket.de/doc_sat/nuclear.htm, data retrieved 13 October.
Radioisotope power systems, fission surface power and nuclear propulsion offer high-density, reliable and long-duration energy solutions. These systems could support deep-space exploration, life support, in-situ resource utilisation, surface operations and transport. They may be particularly relevant for missions beyond Mars, lunar night operations, permanently shadowed regions of the Moon, high-power surface systems and future in-space propulsion applications. Nuclear fission reactors are a different technology. They rely on a controlled chain reaction that releases heat, which is subsequently converted into energy. Compared with radioisotope power systems, fission systems can generate higher power levels and may become relevant for more energy-intensive activities, including sustained surface operations and propulsion concepts. Opportunities also exist to exploit synergies with terrestrial small modular reactor development, including knowledge transfer on safety, materials, power conversion and systems engineering.
Key challenges vary by technology and include technology readiness, testing infrastructure, mission design, fuel form, power conversion, system reliability and integration with spacecraft or surface architectures. The wider safety, environmental, supply and governance implications of nuclear power in space are important but are addressed in the chapter on vulnerabilities and resilience.
Space innovation is expanding rapidly and becoming more geographically diverse. China now accounts for much of the recent growth in scientific output, while OECD countries continue to perform strongly in research excellence and international collaboration. Patent indicators show sustained technological development, particularly in spacecraft power systems, control systems and emerging fields such as quantum technologies.
For policy makers, the key challenge will be to maintain strong research and industrial capabilities while supporting collaboration, skills development and access to specialised infrastructure and data. Emerging technologies such as AI-enabled data processing and nuclear power systems could open new mission possibilities but will require forward-looking policy frameworks. Issues related to safety, supply constraints and resilience are addressed in later chapters.
References
[8] Danish, M. et al. (2025), “GEOBench-VLM: Benchmarking Vision-Language Models for Geospatial Tasks”, 2025 IEEE/CVF International Conference on Computer Vision (ICCV), pp. 7132-7142, https://doi.org/10.1109/iccv51701.2025.00670.
[6] Goodwill, J., C. Wilson and J. MacKinnon (2023), “Current AI Technology in Space”, in Krittanawong, C. (ed.), Precision Medicine for Long and Safe Permanence of Humans in Space, Elsevier, https://ntrs.nasa.gov/api/citations/20240001139/downloads/Current%20Technology%20in%20Space%20v4%20Briefing.pdf.
[11] ISECG (2024), The Global Exploration Roadmap: 2024, International Space Exploration Co-ordination Group, https://www.globalspaceexploration.org/wp-content/isecg/GER2024.pdf.
[10] ISECG (2018), The Global Exploration Roadmaps: 2018, International Space Exploration Co-ordination Group, https://www.globalspaceexploration.org/wp-content/isecg/GER_2018_small_mobile.pdf.
[9] ISECG (2013), The Global Exploration Roadmap: 2013, International Space Exploration Co-ordination Group, https://www.nasa.gov/wp-content/uploads/2015/01/GER-2013_Small.pdf.
[12] Krebs, G. (2025), Nuclear Powered Payloads, https://space.skyrocket.de/doc_sat/nuclear.htm.
[13] OECD (2026), “Science and bibliometric indicators”, https://www.oecd.org/en/data/datasets/science-and-bibliometric-indicators.html.
[1] OECD (2026), STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats (accessed on June 2026).
[3] OECD (2025), OECD Science, Technology and Innovation Outlook 2025: Driving Change in a Shifting Landscape, OECD Publishing, Paris, https://doi.org/10.1787/5fe57b90-en.
[5] OECD (2023), The Space Economy in Figures: Responding to Global Challenges, OECD Publishing, Paris, https://doi.org/10.1787/fa5494aa-en.
[4] OECD (2019), The Space Economy in Figures: How Space Contributes to the Global Economy, OECD Publishing, Paris, https://doi.org/10.1787/c5996201-en.
[2] Olivari, M., C. Jolly and M. Undseth (2021), “Space technology transfers and their commercialisation”, OECD Science, Technology and Industry Policy Papers, No. 116, OECD Publishing, Paris, https://doi.org/10.1787/0e78ff9f-en.
[7] Szwarcman, D. et al. (2026), “Prithvi-EO-2.0: A Versatile Multitemporal Foundation Model for Earth Observation Applications”, IEEE Transactions on Geoscience and Remote Sensing, Vol. 64, pp. 1-20, https://doi.org/10.1109/tgrs.2025.3642610.
[14] US National Geospatial Advisory Committee (2020), Landsat data: Community standard for data calibration, a report of the Landsat Advisory Group, https://www.fgdc.gov/ngac/meetings/october-2020/ngac-paper-landsat-data-community-standard-for.pdf.
← 1. The top 10% most cited documents is an indicator of “excellence”. This rate indicates the amount (in percentages) of a country’s scientific output that is included into the group of the 10% of the most cited papers in their respective scientific fields. The indicator of scientific excellence is calculated at the document level using fractional counts. Documents – organised by document type, ASJC field and year – are sorted in descending order based on the number of citations received. A threshold of 10% most cited documents is calculated for each category. Only documents with a fixed number of citations above the threshold are included. Documents with the same number of citations as the threshold are sorted according to the Scimago Scientific Journal Rankings (SJR) value of the journal in which they were published; those with the highest scores are selected. It is likely that this step of the procedure – used to select the citation tiebreak documents – favours countries publishing in higher prestige journals versus a more random approach. The resulting documents are included until 10% is attained. Only after this selection has been made can the identification by country and field be made to determine their share in total output. The citation window is based on the whole period, so top 10% most cited documents for the reference year uses whole period citations. No citation window is imposed as citation-based indicators are calculated on the basis of comparisons with documents published in the same year. The world average is 10% for the period (OECD, 2026[13]).
← 2. Change detection identifies changes in specific land features by sensing them at different times. Semantic segmentation means classifying every pixel within images in specific categories, such as vegetation, urban infrastructure, bare soil, etc. Satellite image calibration is essential for reliable change detection and semantic segmentation because it ensures that variations in pixel values reflect genuine changes on the ground rather than differences in sensors, atmospheric conditions, illumination, or acquisition settings. US Landsat imagery serves as a reference benchmark for other satellite sensors, due to their long and consistent time series, pre-launch testing, continuous comparison with ground reference sites and onboard calibration (US National Geospatial Advisory Committee, 2020[14]).