Draft, not yet released. In stewardship review. Not indexed until the first approved release.

Science is where the frontier is made

Part I describes systems and their couplings. It ranks no country or operator and assesses no authority.

7 min read2 figures2 tables17 notesNR-SECTOR-SCIENCE-GLOBAL-2026-BASELINE

Contents
  1. ·Edition home
  2. ·Executive summary
  3. 1Science is where the frontier is made
    1. 1.1 The questions this part answers
    2. 1.2 The argument
    3. 1.3 Twenty couplings between five systems
    4. 1.4 Reinforcing loops
  4. 2Key findings
  5. 3Systems and sectors at baseline
  6. 4Risks and capability
  7. 5Loss of trust in the research record during a fast-moving emergency
  8. 6Readings from the other desks
  9. 7The 2027 agenda
  10. 8Systems in depth
  11. AMethodology
  12. BData appendix
  13. CReferences and record
  14. ·Complete edition, for print

1.1The questions this part answers

Key questions

  • Where is science and engineering risk concentrated?
  • Which couplings carry science and engineering shocks into other systems?
  • Which science and engineering capabilities hold under compound stress?
  • What must 2027 evidence measure first?

In this part

  • Narrative
  • Headline finding
  • Lead figure
  • Five coupling chapters
  • Twelve-domain reading
  • Sector screens
  • Dependency matrix
  • Risk objects
  • Capability ladder
  • Scenario
  • Three lenses
  • Portfolio

1.2The argument

Science is usually reported as a series of results. The 2026 evidence reads it as a system with funders, people, instruments, journals and data that other systems depend on. Medicines, forecasts, chips and models all start in it111794.

The system is large and uneven. Worldwide spending reached USD 3.48 trillion in 2024, with China and the United States together performing more than half of it1. Researcher density runs from 88 per million in Sub-Saharan Africa to 5,006 in Northern America2.

The record is growing faster than it can be checked. Articles, preprints and AI-assisted outputs rise each year159, and retraction notices peaked at 15,582 in 20236. Case studies show that coordinated fraud is large, resilient and growing14, and a replication project in psychology found 36% of replications significant against 97% of originals16.

The frontier is becoming more computational. Five exascale systems now sit on the TOP500 list4, training compute has doubled about every six months since the early 2010s13, and AI systems are entering hypothesis generation27. Governance is moving in parallel through UNESCO recommendations and a WHO framework on dual-use research252618.

Figure 1.1How unevenly is research capacity spread? Researchers per million inhabitants (FTE), 2023

The densest region has about fifty-seven times as many researchers per million people as the thinnest one. Spending shares follow the same order, so capacity and funding compound each other2.

ItemBar, scaled to the largest valueValue
Northern America5006
Europe4025
Eastern Asia2632
Latin America and the Caribbean668
Southern Asia347
Sub-Saharan Africa88

Source: as cited in the note to this figure. Year 2023.

Figure ID FIG-SECTOR-SCIENCE-GLOBAL-2026-BASELINE-01-003

88per million

researchers in Sub-Saharan Africa2

5,006per million

researchers in Northern America2

0.38%

of GDP spent on R&D in Sub-Saharan Africa2

70per million

researchers in Least Developed Countries2

1.3Twenty couplings between five systems

The science and engineering-centred matrix types the twenty couplings between the five core systems, read for their science and engineering content. Strengths are Nexus estimates (ESTIMATED) from the cited evidence and analogue events; they set the graph that 2027 evidence will test.

Figure 1.2Couplings between the five systems

Each line is one coupling from one system to another, drawn by the strength the edition assigns. Strengths are Nexus estimates from the cited evidence and analogue events.

  • Strong
  • Moderate
  • Weak

Source: this edition’s coupling matrix (Table 1.1); status Estimated.

Table 1.1Coupling matrix: from the row system to the column system
From, toWATENEFOOHLTBIO
WAT Water·WEAK
hydropower research		MODERATE
irrigation science and crop trials		WEAK
exposure studies		MODERATE
river and lake monitoring
ENE EnergyWEAK
instrument and desalination research load		·WEAK
processing research		MODERATE
power for laboratories and clinics		WEAK
field-station power
FOO FoodWEAK
farm monitoring data		WEAK
bio-based fuels research		·MODERATE
nutrition science		MODERATE
crop genetics and land-use science
HLT HealthWEAK
public health surveillance		WEAK
workforce		MODERATE
zoonosis and food-safety science		·STRONG
genomics and structural biology tools
BIO BiodiversityWEAK
catchment monitoring		WEAK
biomass research		MODERATE
breeding resources		STRONG
sequences, structures and medicines		·
Strength: strong, moderate or weak; a Nexus estimate (status Estimated) from cited evidence and analogue events. The scale is defined in Annex A.

Transmission paths

  1. 01

    BIO → HLT

    Sequence, structure and trial data feed medicines and gene therapy.

    Reported AlphaFold database and the 2023 authorisation1217

  2. 02

    AIC → DAT → GOV

    Mass-produced or AI-assisted papers dilute the record and strain peer review.

    Reported paper-mill case studies141

  3. 03

    FIN → DAT

    Funding and talent concentrate capacity in a few regions.

    Reported R&D shares and researcher density12

  4. 04

    INF → AIC

    Exascale and training compute decide which groups can work at the frontier.

    Reported TOP500 list and compute trend413

  5. 05

    TRD → DAT

    Co-authorship and mobility carry knowledge across borders.

    Reported 22% of articles; AI researcher moves18

1.4Reinforcing loops

A shock entering either system of a loop can return amplified through the other.

Table 1.2Pairs of systems where both directions are rated strong
LoopFirst directionSecond direction
HLT and BIOHealth to biodiversity: genomics and structural biology toolsBiodiversity to health: sequences, structures and medicines
A causal-loop reading of the matrix (Sterman, Business Dynamics, 2000): structural, from estimated strengths, not a measured feedback.

Notes

Numbers are the edition’s reference numbers, so a note reads the same in every chapter. The full list, with every source and its licence, is in Annex C. Annex C, references

  1. 11Abramson, J. et al., Accurate structure prediction of biomolecular interactions with AlphaFold 3, Nature 630, 493 to 500, 2024 (abstract and open full text read through Europe PMC). Springer Nature, London. Observation year 2024. Accessed 2026-11-03.back
  2. 17US Food and Drug Administration (ed.), FDA Approves First Gene Therapies to Treat Patients with Sickle Cell Disease, press announcement of 8 December 2023. FDA, Silver Spring. Page read. Observation year 2023. Accessed 2026-11-03.back
  3. 9Stanford HAI (ed.), The 2026 AI Index Report, Chapter 5 Science, chapter page with its highlights. Stanford Institute for Human-Centered AI, Stanford. Page read; the chapter PDF was not opened. Observation year 2025. Accessed 2026-11-03.back
  4. 4TOP500 project (ed.), TOP500 list, 67th edition, June 2026 (release text read in full for the leading systems). TOP500, Prometeus GmbH. Observation year 2026. Accessed 2026-11-03.back
  5. 1NSB (ed.), Science and Engineering Indicators 2026: The State of U.S. Science and Engineering (NSB-SEP-2026-1), May 2026, National Science Board and National Center for Science and Engineering Statistics, US National Science Foundation, Alexandria. Full PDF opened; global R&D, publications, collaboration, venture investment, semiconductor and doctoral-degree figures read from the text. Observation year 2024. Accessed 2026-11-03.back
  6. 2UNESCO Institute for Statistics (ed.), UIS data API, data version 20260507 (February 2026 data release): indicators EXPGDP.TOT, RESDEN.INHAB.TFTE and FRESP.THC, world and regional aggregates. UNESCO, Montreal. Data pulled and read. Observation year 2023. Accessed 2026-11-03.back
  7. 5POST-CLOSE, context only: arXiv (ed.), Monthly submissions statistics (monthly CSV, 1991 to November 2026; annual totals summed by the editors). Cornell University, Ithaca. Observation year 2025. Accessed 2026-11-03.back
  8. 6Crossref (ed.), Crossref REST API, work counts by type and publication year and counts of retraction notices (update-type filter), queried live on the access date. Crossref, Oxford. Counts are subject to later registration. Observation year 2025 (series 2015 to 2025). Accessed 2026-11-03.back
  9. 14Richardson, R.A.K. et al., The entities enabling scientific fraud at scale are large, resilient, and growing rapidly, Proceedings of the National Academy of Sciences 122(32), 2025 (abstract and open full text record read through Europe PMC). National Academy of Sciences, Washington. Observation year 2025. Accessed 2026-11-03.back
  10. 16Open Science Collaboration, Estimating the reproducibility of psychological science, Science 349(6251), aac4716, 2015 (abstract read through Europe PMC; full text not opened). American Association for the Advancement of Science, Washington. Observation year 2015. Accessed 2026-11-03.back
  11. 13Sevilla, J. et al., Compute Trends Across Three Eras of Machine Learning, arXiv 2202.05924, 2022 (abstract page read). Cornell University, Ithaca. Observation year 2022. Accessed 2026-11-03.back
  12. 27Gottweis, J. et al., Accelerating scientific discovery with Co-Scientist, arXiv 2502.18864 (version 2, 2026; abstract read through the arXiv API). Cornell University, Ithaca. Observation year 2025. Accessed 2026-11-03.back
  13. 25UNESCO (ed.), UNESCO Recommendation on Open Science, 2021, programme page (193 Member States). UNESCO, Paris. Page read; the Recommendation text itself returned an access error. Observation year 2021. Accessed 2026-11-03.back
  14. 26UNESCO (ed.), Recommendation on the Ethics of Artificial Intelligence, November 2021, programme page (193 Member States). UNESCO, Paris. Page read; the text itself was not opened. Observation year 2021. Accessed 2026-11-03.back
  15. 18WHO (ed.), Global guidance framework for the responsible use of the life sciences: mitigating biorisks and governing dual-use research, 174 pages, 2022. World Health Organization, Geneva. Publication page and overview read; the full text was not opened. Observation year 2022. Accessed 2026-11-03.back
  16. 12Varadi, M. et al., AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences, Nucleic Acids Research 52 (database issue), 368 to 375, 2024 (abstract read through Europe PMC). Oxford University Press, Oxford. Observation year 2024. Accessed 2026-11-03.back
  17. 8Stanford HAI (ed.), The 2026 AI Index Report, overview page with the ten top takeaways. Stanford Institute for Human-Centered AI, Stanford. Page read; the report PDF was not opened. Observation year 2025. Accessed 2026-11-03.back

Cite this chapter

Valadbaygi, S. (Ed.). (2026). Chapter 1: Science is where the frontier is made. In Nexus Global Science and Engineering Frontier Risk and Resilience Report 2026: Baseline Edition, The State of the World’s Science and Engineering Frontier. (Baseline Edition 2026, Report No. NR-SECTOR-SCIENCE-GLOBAL-2026-BASELINE) [Draft under review]. The Global Centre for Risk and Innovation (GCRI).

Every citation form of the whole report, BibTeX and RIS included, is on the edition home. Cite this report

ReportsContribute

Send evidence or a correction on Global Science and Engineering Frontier.

Reports is the Consortium’s quarterly publication, written by six desks. This page sits with the Global Centre for Risk and Innovation (GCRI) Systems Evidence desk. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work. Sent to the wrong desk, it is forwarded, not refused.

GCRI Systems Evidence deskGlobal Sector Risk and Resilience

Never send personal data, sensitive infrastructure detail or anything you have no right to share. A contribution is reviewed before use; sending it confers no authorship or editorial rights, and sponsorship never buys a finding.

CC BY 4.0 Free to read. Text and figures: CC BY 4.0. Data: CC0 1.0. Authors keep copyright. Published by GCRI Canada.

The Global Centre for Risk and Innovation

Science · evidence · measurement

Not for profit

Our articles, maps and twins are free to read and free to republish. Readers and partners keep it that way.

Partner with us

Who funds us

GCRI is a not-for-profit institution. To ask who funds a piece of work, use the funding questions form.

Independence

Editors and the expert reviewers named on each article decide what we publish. Funders and partners do not review or approve articles before publication.

Corrections policy

Licence

Articles: CC BY 4.0. Figures, maps, twins and data: CC BY 4.0. Other uses: ask through the permissions form.

Privacy choices