Tracking how top scientists move and collaborate reveals how global knowledge networks are evolving – and why policy must shift from attracting talent to enabling its circulation.
Science today is more interconnected than ever. Many of the most consequential scientific challenges are global in nature, and progress on issues such as climate change and public health, as well as in fast-moving fields such as artificial intelligence, increasingly depends on knowledge-sharing and collaboration across countries, institutions and disciplines. Scientists continue to work through international teams, joint research projects and global research networks, making cross-border mobility an important part of how knowledge is exchanged and advanced.
The OECD has underscored the importance of sustaining open international research networks as global challenges become more complex. At the same time, a 2025 Nature survey found that many US-based researchers were considering moving abroad – another indication that the cross-border movement of scientific talent continues to shape the global research landscape.
This makes scientist mobility an increasingly important signal of how the global innovation system is changing. But it is easy to read that signal too narrowly. A simple “brain gain versus brain drain” frame treats scientists as assets moving from one national container to another. In reality, leading researchers often hold multiple affiliations, collaborate across borders, and move through networks of universities, hospitals and companies. Mobility is therefore not only about who leaves and who arrives. It is also about how knowledge circulates.
The Global Mobility and Collaboration of Top Scientists Report 2026, produced by the research team from the Beijing Institute of Talent Development Strategy (BITS), offers a way to look at that circulation. Using Elsevier’s Scopus database, the report analysed 11,431 valid records covering changes in publication affiliations between 2024 and 2025. “Top scientists” were identified using research impact and citation criteria. Cross-border changes in publication affiliation serve as a proxy for both physical mobility and participation in international knowledge networks.
Key trends in global scientist mobility
Three findings stand out. First, intercontinental mobility is now a defining feature of top-level science. The analysis records an intercontinental mobility volume of 6,984, compared with 4,447 within continents. Asia, Europe and North America account for 73.7% of intercontinental mobility among the six continents.
What is striking is not just the scale, but the balance. Asia recorded an inflow volume of 2,158 and an outflow of 2,018; Europe, 2,057 and 2,056; North America, 1,901 and 1,908. These are highly connected systems in which talent moves in both directions.
That matters because a mature innovation hub is not a sealed container. It attracts people, but it also sends scientists outwards to build collaborations, access facilities, form teams and connect to other centres of expertise. High inflows combined with high outflows can therefore be a sign of connectivity and circulation, rather than weakness. Balance is an intrinsic sign of a maturing science and innovation ecosystem.
Second, national concentration coexists with a much more distributed urban network. At national and regional levels, mobility remains concentrated: the United States recorded a total mobility volume of 3,421, the Chinese mainland 2,579, the United Kingdom 1,891 and Germany 1,132. The major countries and regions in the dataset account for more than 90% of international mobility, and the top five represent 47.9% of the total among the top 50.
Cities tell a different story. The top 50 cities account for only 35.1% of all international intercity mobility, suggesting a wider network of active nodes. London recorded a mobility volume of 573, followed by Beijing at 450, Singapore at 439 and Boston at 261. Ten Chinese cities were among the top 50, while established hubs across Europe, North America, Asia and Oceania also featured prominently.
This suggests that the geography of science cannot be understood through national rankings alone. Researchers experience innovation ecosystems locally: through universities, laboratories, hospitals, firms, funding communities and the quality of connections between them. Cities are becoming increasingly important gateways into global science.
Third, there is no single global market for scientific talent. Mobility patterns vary significantly by discipline. In basic sciences, exchanges between Beijing and Singapore were particularly active. In computer science, Mountain View in Silicon Valley and London formed one of the most visible corridors. In biomedical sciences, London, Boston and New York were among the most active cities, with strong links to other global centres.
Career stage matters too. Some 82.8% of international mobility in the sample occurred within the first 30 years of a research career, and 47.5% was concentrated in career years 11 to 20. The most globally mobile researchers are therefore often not early-career entrants, but established scientists in highly productive phases of their careers.
What scientific mobility means for innovation policy
For policy-makers, these patterns point to a particular approach to talent strategy. The first priority is to build strong research platforms, not simply recruitment packages. Top scientists are more likely to contribute over time when they can assemble teams, access major facilities, secure competitive funding and connect with outstanding collaborators.
The second is to design for circulation. Joint appointments, visiting programmes, cross-border research infrastructure and predictable rules for international collaboration can allow knowledge to move even when permanent relocation is neither possible nor desirable. Openness and research security should not be treated as a binary choice; the challenge is to manage risks without breaking the networks on which science depends.
The third is to tailor policy to disciplines, cities and career stages. A biomedical hub needs different infrastructure from a computer science cluster, and a mid-career scientist may value team-building capacity and institutional autonomy more than a short-term financial incentive.
The key question, then, is not whether every top scientist stays. It is whether an innovation system is attractive enough to bring talent in, connected enough to send people out and bring knowledge back, and capable of turning circulation into durable collaboration. In a more multipolar scientific world, the places that manage these flows well may be the ones that remain closest to the frontier.
