Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems

The first annual edition of the AI and agentic systems report. It fixes the 2026 reference state of AI as physical infrastructure, and the model, voice and positions, against which the 2027 edition and every national readiness chapter will be read.

Read the executive summaryStart with chapter 1Cite this report
Editor
Saeed Valadbaygi
Publisher of record
GCRI for the Nexus Consortium
Evidence close
30 September 2026
Status
Draft in stewardship review; not yet released
ReportsXT-01

Exponential Technology

AI and Agentic Systems 2026

Baseline Edition 2026NR-XTECH-AI-GLOBAL-2026-BASELINEGCRI for the Nexus Consortium
Findings12each with its status and source
Sources40numbered, editor first
Chapters and annexes11plus the executive summary
Indicators15in the data appendix, with CSV and JSON

01 · Key messages

What the edition finds.

Five of the questions the edition answers, each with its status. The executive summary has all of them, with their sources.

  1. Data centres used about 415 TWh in 2024, about 1.5% of the world total, rising to about 945 TWh by 2030 in the IEA base case.

    How much power does AI use?Reported

  2. Frontier training compute has grown four- to five-fold a year.

    How fast is capability scaling?Reported

  3. The cost of GPT-3.5-level performance fell more than 280-fold in two years.

    Is AI getting cheaper?Reported

  4. About 92% of leading-edge chip capacity sat in one region in 2019.

    How concentrated is supply?Reported

  5. 78% of organisations used AI in 2024.

    How widely is AI used?Reported

All key messages and the headline finding

02 · Contents

Eight chapters and three annexes.

  1. ·

    Executive summary

    The edition on a few pages: the abstract, the key messages, the headline finding and what each steward reads in it.

  2. 1

    Chapter 1. AI is infrastructure, not only software

    Part I describes systems and their couplings.

  3. 2

    Chapter 2. Key findings

    Twelve findings carry the AI baseline. They are grouped by coupling, never ranked.

  4. 3

    Chapter 3. Systems and sectors at baseline

  5. 4

    Chapter 4. Risks and capability

    Four AI risk objects open the portfolio. Order carries no rank.

  6. 5

    Chapter 5. Chip supply disruption under grid constraint

    A conditional statement under stated assumptions, never a prediction.

  7. 6

    Chapter 6. Lenses and readings

    The AI baseline is read through each steward’s lens, then once as a whole.

  8. 7

    Chapter 7. The 2027 agenda

    Eight bounded questions open the 2027 AI agenda.

  9. 8

    Chapter 8. Systems in depth

    One section per system: its state at evidence close, its exposures and what the operational series will watch.

  10. A

    Annex A. Methodology

    How the edition was made: methods, the model card, the evidence cut and its age, the acceptance record and the glossary.

  11. B

    Annex B. Data appendix

    The indicator set with its sources, the source registers, the licences of every kind of material and the formats of the edition.

  12. C

    Annex C. References and record

    Every source the edition rests on, the publication record, the corrections log and the open record.

List of figures (4)
  1. S.1Data-centre electricity use (TWh)
  2. 1.1How fast is AI being adopted, and at what scale? Selected indicators, 2024
  3. 1.2Couplings between the five systems
  4. 2.1Data-centre electricity use (TWh)
List of tables (43)
  1. 1.1Coupling matrix: from the row system to the column system
  2. 1.2Pairs of systems where both directions are rated strong
  3. 3.1Each domain read in words, not scored
  4. 3.2Sectors: baseline evidence and the exposure to watch
  5. 4.1Capabilities on the ladder of evidence: listed, exists, demonstrated, tested
  6. 6.12027 AI readiness pack
  7. 6.2Standards and regulatory watch
  8. 6.3Peer-reviewed science and assessments
  9. 6.4Institutional flagships
  10. 6.5Statistics, labour and access
  11. 6.6Supply chains, incidents and electricity
  12. 7.1Objects proposed for the first operational quarter
  13. 7.2Universe 2027, evidence pack for the AI candidate mission
  14. 7.3Handoffs and routing
  15. 7.4Routing rules by publication class
  16. 7.5Feeds watched from 2027
  17. 7.6Signals at baseline
  18. 8.1State at baseline
  19. 8.2State at baseline
  20. 8.3State at baseline
  21. 8.4State at baseline
  22. 8.5State at baseline
  23. A.1Methods
  24. A.2Coupling strength scale (matrix and model card)
  25. A.3Status grammar
  26. A.4From the baseline cut to the federation cut
  27. A.5Evidence age and coverage
  28. A.6Coverage by system at baseline
  29. A.7Quality dimensions at baseline
  30. A.8Publication acceptance record
  31. A.9The status every value carries
  32. A.10Series terms
  33. A.11Abbreviations used in this edition
  34. A.12Technical annex
  35. B.1Global baseline indicator set
  36. B.2Global baseline indicator set
  37. B.3Global source register
  38. B.4Regional reading, AI across ten clusters
  39. B.5How countries appear in the AI edition
  40. B.6National Baseline Editions in the 2026 programme
  41. B.7Licences of each kind of material
  42. B.8This edition in every format
  43. C.1Release path for this version

03 · Global Risks Alliance reading

How Global Risks Alliance reads this edition.

Readiness implications only. Investment is large and concentrated, supply risk sits in a few regions, and liability and insurance for AI are nascent. These show where finance-readiness evidence will matter. Nothing here is advice, a rating or an approval.

ADO-006 · MIN-005

Chapter 6, lenses and readings

04 · Publication record

Version, dates and contributors.

The identity of this edition as its title pages record it. A review copy has no DOI; the DOI is issued at deposit.

Publication ID
NR-XTECH-AI-GLOBAL-2026-BASELINE-v1.0.0
Version
1.0.0, review copy
Edition date
Evidence close
Status
Draft under stewardship review; not yet released
Editor
Saeed Valadbaygi
Publisher of record
The Global Centre for Risk and Innovation (GCRI)
Licence of the text
CC BY 4.0
DOI
Issued at deposit, on release
Previous edition
none, first edition of the series
Next edition
NR-XTECH-AI-GLOBAL-2027-v1.0.0

Contributors

RoleContributionBy
ConceptualisationSeries architecture and the technology system modelS. Valadbaygi (ed.)
MethodologyBaseline method, dependency scoring and status grammarReports editorial team, GCRI
Investigation · data curationSource review of AI energy, compute, safety, labour and governance evidenceReports editorial team, GCRI
Writing, original draftAll chapters; AI-assisted drafting, disclosed belowEditor · editorial team
Writing, review and editingFive review passes before releaseSteward reviewers (scheduled)
VisualisationPages, figures and tables in the Nexus design standardReports editorial team, GCRI
SupervisionEdition governance, registry record and depositS. Valadbaygi (ed.)

Review before release

PassScopeByState
1Scientific and technicalGCRI · key TScheduled
2Reproducibility of analytical valuesGCRI · key TScheduled
3Public meaning and claimsThe Global Risks Forum (GRF) · key MScheduled
4Finance-readiness boundaryGRA · key FScheduled
5Regional context, safeguards and dual useRegional consortia · key NScheduled

AI assistance was used for drafting, consistency checks and page composition. Every number, citation and status label was checked against its source by the editorial team before freeze. No AI system is recorded as an author or a reviewer, and no AI output is printed as fact without a source.

05 · Downloads

Read it in print, take the data.

FormatWhat it isStateGet it
Complete edition, print or PDFA4 or Letter, chosen in the print dialog: cover, contents, every chapter and annex, endnotes and running heads Review copyOpen for print
Indicators, CSV15 indicators with unit, year, status and source On this pageDownload
Indicators, JSONThe same indicators with the edition’s identity and licence On this pageDownload
Findings, CSV12 findings with status, evidence-quality level and references On this pageDownload
Citation, BibTeXFor LaTeX and reference managers On this pageDownload
Citation, RISFor Zotero, EndNote and Mendeley On this pageDownload
Signed PDF with DOIThe archival PDF/A with its DOI, signed with the web edition and the data bundle At releaseNot yet issued

The print and the files are produced from the same edition pack as the web pages, so they always agree. Until release they are review copies.

06 · Cite

Cite this report.

This edition is a draft in stewardship review. Cite it as a draft; the DOI and the UNI are issued at deposit, and the citation then names them.
Catalogue keyNR-XTECH-AI-GLOBAL-2026-BASELINE

Standards applied to NR-XTECH-AI-GLOBAL-2026-BASELINE

Cite as

S. Valadbaygi (ed.), Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems. Reports, Baseline Edition 2026. The Global Centre for Risk and Innovation (GCRI) for the Nexus Consortium, Toronto, 2026. NR-XTECH-AI-GLOBAL-2026-BASELINE. DOI at deposit.

Applied

NS-1 catalogue identity (a UNI is allocated at deposit and resolves to this record), NS-2 citation and attribution, NS-3b dataset metadata, NS-5 figures and maps, NS-12 software and AI disclosure.

Monitoring

Corrections are new versions with a published notice; the earlier version stays resolvable.

An attestation is evidence, not a certificate. Nexus Standards parts are preparatory working drafts (2026).

APA 7
Valadbaygi, S. (Ed.). (2026). Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems (Baseline Edition 2026, Report No. NR-XTECH-AI-GLOBAL-2026-BASELINE) [Draft under review]. The Global Centre for Risk and Innovation (GCRI). https://therisk.global/nexus-reports/editions/xt-01/
Chicago
Valadbaygi, Saeed, ed. Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems. Baseline Edition 2026. Nexus Reports NR-XTECH-AI-GLOBAL-2026-BASELINE. Draft under review. Toronto: The Global Centre for Risk and Innovation (GCRI), 2026. https://therisk.global/nexus-reports/editions/xt-01/.
BibTeX
@techreport{NR-XTECH-AI-GLOBAL-2026-BASELINE,
  editor      = {Valadbaygi, Saeed},
  title       = {{Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems}},
  institution = {The Global Centre for Risk and Innovation (GCRI)},
  address     = {Toronto},
  year        = {2026},
  type        = {Nexus Reports, Baseline Edition 2026},
  number      = {NR-XTECH-AI-GLOBAL-2026-BASELINE},
  note        = {Draft under stewardship review; not yet released. DOI assigned at deposit},
  url         = {https://therisk.global/nexus-reports/editions/xt-01/}
}
RIS
TY  - RPRT
ED  - Valadbaygi, Saeed
TI  - Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems
T3  - Nexus Reports
ET  - Baseline Edition 2026
PY  - 2026
PB  - The Global Centre for Risk and Innovation (GCRI)
CY  - Toronto
SN  - NR-XTECH-AI-GLOBAL-2026-BASELINE
UR  - https://therisk.global/nexus-reports/editions/xt-01/
N1  - Draft under stewardship review; not yet released. DOI assigned at deposit.
LA  - en
ER  -
NR-CIT-1 (house style)
S. Valadbaygi (ed.), Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems. Nexus Reports, Baseline Edition 2026. The Global Centre for Risk and Innovation (GCRI) for the Nexus Consortium, Toronto, 2026. NR-XTECH-AI-GLOBAL-2026-BASELINE. DOI at deposit.
ReportsContribute

Send evidence or a correction on Artificial Intelligence and Agentic Systems.

Reports is the Consortium’s quarterly publication. 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.

Global Risks AllianceExponential Technology

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.

Reports · Global Risks Alliance

Bring evidence to the next edition.

Every quarter closes on a sealed evidence cut. A correction, a source or a method received before the cut is reviewed for that edition.

ContributeCorrections log
  • NR-XTECH-AI-GLOBAL-2026-BASELINE
  • CC BY 4.0
  • Evidence close 30 September 2026
The Global Risks Alliance

Finance · trust · precision

Not for profit

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

Support our work

Who funds us

GRA 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.