Reports
Baseline Edition 2026
Complete edition

Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems

Choose A4 or Letter in the print dialog; running heads and page numbers are added in print. This is a review copy: the signed PDF with its DOI is issued at release.

ReportsXT-01

Exponential Technology

AI and Agentic Systems 2026

Baseline Edition 2026NR-XTECH-AI-GLOBAL-2026-BASELINEGCRI for the Nexus Consortium

Imprint

Editor: Saeed Valadbaygi, Chief Steward and Managing Editor, the Global Centre for Risk and Innovation (GCRI)
Publisher of record: The Global Centre for Risk and Innovation (GCRI) for the Nexus Consortium, Toronto
Report number: NR-XTECH-AI-GLOBAL-2026-BASELINE
Evidence close: 30 September 2026, cut DC-XTECH-AI-GLOBAL-2026-B0
Status: Draft under stewardship review; not yet released
DOI and UNI: Issued at deposit
Licence: CC BY 4.0 for Nexus-authored text, tables and figures; third-party material keeps its own licence
Cite as: 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/

Contents

·
Executive summary
1
Chapter 1. AI is infrastructure, not only software
2
Chapter 2. Key findings
3
Chapter 3. Systems and sectors at baseline
4
Chapter 4. Risks and capability
5
Chapter 5. Chip supply disruption under grid constraint
6
Chapter 6. Lenses and readings
7
Chapter 7. The 2027 agenda
8
Chapter 8. Systems in depth
A
Annex A. Methodology
B
Annex B. Data appendix
C
Annex C. References and record

AI and Agentic Systems 2026
Executive summary

Executive summary

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

4 min read1 figure0 tables10 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

S.1Abstract

Artificial intelligence enters 2027 as physical infrastructure. Data centres used about 415 TWh of electricity in 2024, about 1.5% of the world total, and the IEA projects roughly 945 TWh by 2030, driven by accelerated servers for AI¹. The training compute of frontier models has grown by four to five times a year¹⁴, while the cost of using a model at a fixed level of performance fell more than 280-fold in two years². Adoption has spread fast: 78% of organisations reported using AI in 2024². Supply is concentrated: in 2019 about 92% of the world’s most advanced logic-chip capacity was in Taiwan¹⁷, and the refining of key minerals is concentrated in a few countries¹⁸. Reported AI incidents rose by more than half in 2024², and governance is moving from principles to law, led by the phased application of the EU AI Act⁴. This Baseline Edition fixes the 2026 reference state for AI as a system of compute, energy, water, chips, labour and institutions, and sets the questions the 2027 edition will answer. It ranks no model, company or country.

S.2Key messages

Six readings that frame the AI baseline, and the eight questions this edition answers, each with its status and source.

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.¹
How much power does AI use?Reported
Frontier training compute has grown four- to five-fold a year.¹⁴
How fast is capability scaling?Reported
The cost of GPT-3.5-level performance fell more than 280-fold in two years.²
Is AI getting cheaper?Reported
About 92% of leading-edge chip capacity sat in one region in 2019.¹⁷
How concentrated is supply?Reported
78% of organisations used AI in 2024.²
How widely is AI used?Reported
A quarter to two-fifths of jobs are exposed; exposure is not job loss.¹⁶³²
What does it mean for work?Reported
Reported incidents reached a record 233 in 2024.²
Are harms rising?Reported
A chip supply disruption coinciding with grid limits at data-centre clusters. SCN-XT01-CHG-01
What will 2027 test first?Expert input

S.3Headline finding

ReportedEQL 4NRF-XTECH-AI-GLOBAL-2026-BASELINE-ENE-001

Data centres used about 415 TWh of electricity in 2024, about 1.5% of the world total, and the IEA projects about 945 TWh by 2030, driven mainly by accelerated servers for AI.

Figure S.1Data-centre electricity use (TWh)

Item	Bar, scaled to the largest value	Value
2024 · observed estimate		415
2030 · IEA base case		945

Source: IEA, Energy and AI¹.

Because demand is clustered, the first effects fall on particular grids, water basins and communities: connection queues, local price pressure and competition for clean power. The global share stays small while local shares become large¹¹⁵.

S.4At a glance

415TWh

data-centre electricity, 2024¹

78%

organisations using AI²

252US$ bn

corporate AI investment, 2024²

92%

leading-edge chips in one region¹⁷

233incidents

reported in 2024²

2.6billion

people offline, 2024³⁴

S.5What each steward reads in it

GCRI · KEY T

Technical Truth

The evidence supports reading AI as physical infrastructure with fast-rising, clustered demand for power and water, and a highly concentrated supply chain.

GRF · KEY M

Public Meaning

For most people AI arrives as a tool at work, a change in a job or a new data centre nearby.

GRA · KEY F

Capital Meaning

Readiness implications only. Investment is large and concentrated, supply risk sits in a few regions, and liability and insurance for AI are nascent.

NEXUS

Synthesis

AI is scaling faster than the infrastructure and institutions around it.

S.6What this edition is, and is not

What this edition is

A baseline: the state of AI compute, energy, water, chips, adoption, labour, incidents and governance in 2026, and the couplings through which AI carries risk into power systems, supply chains, work and institutions. It treats AI as infrastructure, not only software.

What it is not

Not a capability ranking of models or companies, a forecast of artificial general intelligence, a safety certification or a policy recommendation. Benchmarks are cited with their dates and methods.

What comes next

National readiness chapters and federation evidence feed the 2027 edition. Each edition reports what changed against this baseline, new measurements, source revisions, method changes, scenario updates and corrections.

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

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024.back

AI and Agentic Systems 2026
Chapter 1

AI is infrastructure, not only software

Part I describes systems and their couplings. It ranks no model, company or country and certifies no system.

7 min read2 figures2 tables17 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

1.1The questions this part answers

Key questions

Where does AI demand meet physical limits first, power, water or chips?
Which couplings carry AI-driven shocks into other systems?
Which AI capabilities and safeguards have been tested?
What must the federation measure first in 2027?

In this part

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

1.2The argument

AI is usually described by what models can do. This baseline describes what AI needs: electricity, water, chips, minerals, networks, skilled people and institutions. Read that way, AI is one of the fastest-growing loads on the world’s physical systems, and one of the most concentrated supply chains in the global economy.

Demand is rising fast. Data centres used about 415 TWh of electricity in 2024, about 1.5% of the world total, and the IEA projects about 945 TWh by 2030 in its base case¹. In the United States, data centres used about 4.4% of electricity in 2023, with 6.7–12% projected for 2028¹⁵. Demand is clustered in a few regions, so its effects on grids and water are local before they are global.

Capability is scaling at a pace few infrastructures have seen. Frontier training compute has grown by four to five times a year¹⁴, while the cost of reaching a fixed level of performance has collapsed: GPT-3.5-level use became more than 280 times cheaper between late 2022 and late 2024². Cheaper use spreads adoption, 78% of organisations reported using AI in 2024², and spreads the energy demand with it.

Supply is concentrated. In 2019 about 92% of leading-edge logic-chip capacity was in Taiwan¹⁷. The refining of minerals used in chips and batteries is concentrated in a few countries¹⁸, and export controls on chips, tools and minerals have tightened since 2022¹⁹²⁹. A single regional disruption could constrain AI capacity worldwide.

Harms are rising with use. Reported AI incidents reached a record 233 in 2024², and the International AI Safety Report documents risks from malicious use, malfunction and systemic effects on labour and information³. Agentic systems that act on their own add new failure modes, while incident data on them remain thin.

Governance is moving from principles to law. The EU AI Act applies in phases from 2025⁴; the Council of Europe opened a binding convention in 2024¹⁰; and in 2025 the UN General Assembly created an independent scientific panel and a global dialogue on AI¹². About 2.6 billion people remain offline³⁴, so access is itself a governance question.

This edition therefore reads AI as part of the coupled system. The matrix, risk objects, capability ladder and scenario that follow ask the same two questions: where does an AI-driven shock travel next, and who holds the capability to contain it?

Figure 1.1How fast is AI being adopted, and at what scale? Selected indicators, 2024

Adoption rose by 23 points in a year, faster than most general-purpose technologies. Exposure estimates for jobs vary by definition, so the labour question is how exposure turns into augmentation or substitution²¹⁶³².

Item	Bar, scaled to the largest value	Value
Organisations using AI (%)		78
Same, 2023 (%)		55
Share of jobs exposed, IMF (%)		40
Share of jobs exposed, ILO (%)		25

Source: as cited in the note to this figure. Year 2024.

Figure ID FIG-XTECH-AI-GLOBAL-2026-BASELINE-01-003

252US$ bn

corporate AI investment, 2024²

170million

jobs expected to be created by 2030²⁰

92million

jobs expected to be displaced by 2030²⁰

2.6billion

people offline, 2024³⁴

1.3Twenty couplings between five systems

The technology-centred matrix types the twenty couplings between the five core systems, read for the demands and effects of AI. Strengths are Nexus estimates from the cited evidence; they set the graph that measurement will test in 2027.

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, to	WAT	ENE	FOO	HLT	BIO
WAT Water	·	STRONG hydro · thermal cooling	STRONG irrigation · rain-fed yield	STRONG safe water · sanitation	STRONG flows · wetlands
ENE Energy	MODERATE pumping · treatment · desalination	·	STRONG fuel · fertiliser · cold chain	STRONG air pollution · cooling	MODERATE land use · mining
FOO Food	STRONG ≈70% of withdrawals	WEAK bioenergy feedstock	·	STRONG nutrition · food safety	STRONG land conversion
HLT Health	WEAK workforce	WEAK workforce	MODERATE labour · disease	·	WEAK residues
BIO Biodiversity	STRONG regulation · filtration	MODERATE biomass · hydrology	STRONG pollination · soils	MODERATE zoonoses · 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

01
AIC → ENE → cities
Clustered data-centre load lengthens grid queues and raises local prices.
Reported United States clusters¹⁵
02
MIN → AIC
A mineral or chip export control constrains compute supply.
Reported gallium and germanium, 2023¹⁹
03
AIC → WAT
Cooling demand competes with local water use.
Inferred water-stressed sites²⁵
04
AIC → DAT → FIN
Concentrated cloud and model dependencies create common failures.
Inferred large outages
05
AIC → labour
Automation changes tasks faster than retraining.
Inferred exposure estimates¹⁶

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
Loop	First direction	Second direction
WAT and FOO	Water to food: irrigation · rain-fed yield	Food to water: ≈70% of withdrawals
WAT and BIO	Water to biodiversity: flows · wetlands	Biodiversity to water: regulation · filtration
FOO and BIO	Food to biodiversity: land conversion	Biodiversity to food: pollination · soils
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

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
19Ministry of Commerce of the People’s Republic of China (ed.), Announcement on implementing export control on gallium- and germanium-related items. MOFCOM, Beijing, July 2023.back
29U.S. Department of Commerce, Bureau of Industry and Security (ed.), Implementation of additional export controls: certain advanced computing and semiconductor manufacturing items. Federal Register, October 2022, as updated.back
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back
4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended.back
10Council of Europe (ed.), Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, CETS No. 225. Council of Europe, Vilnius, 2024.back
12United Nations General Assembly (ed.), Resolution establishing the Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on Artificial Intelligence Governance. United Nations, New York, August 2025.back
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025.back
25P. Li, J. Yang, M. A. Islam and S. Ren, “Making AI less ‘thirsty’: uncovering and addressing the secret water footprint of AI models,” arXiv:2304.03271 (preprint), 2023.back

AI and Agentic Systems 2026
Chapter 2

Key findings

Twelve findings carry the AI baseline. They are grouped by coupling, never ranked. Each resolves to its source or, where Nexus analysis is involved, to its stated method.

3 min read1 figure0 tables14 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

2.1Headline finding

ReportedEQL 4NRF-XTECH-AI-GLOBAL-2026-BASELINE-ENE-001

Data centres used about 415 TWh of electricity in 2024, about 1.5% of the world total, and the IEA projects about 945 TWh by 2030, driven mainly by accelerated servers for AI.

Figure 2.1Data-centre electricity use (TWh)

Item	Bar, scaled to the largest value	Value
2024 · observed estimate		415
2030 · IEA base case		945

Source: IEA, Energy and AI¹.

Implication

What would move it

Grid connection queues at the largest clusters, efficiency gains in chips and cooling, and the pace of inference demand from AI agents would each move this finding.

2.2Energy

ReportedEQL 4NRF-XTECH-AI-GLOBAL-2026-BASELINE-ENE-001
Data centres used about 415 TWh of electricity in 2024, about 1.5% of the world total; the IEA projects about 945 TWh by 2030.¹
ReportedEQL 4NRF-XTECH-AI-GLOBAL-2026-BASELINE-ENE-002
Data centres used about 4.4% of US electricity in 2023, with 6.7–12% projected for 2028.¹⁵

2.3Compute

ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-CMP-003
The training compute of frontier models has grown by four to five times a year since about 2010.¹⁴
ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-CMP-004
The cost of using a model at GPT-3.5 level fell more than 280-fold between November 2022 and October 2024.²

2.4Minerals

ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-MIN-005
About 92% of leading-edge logic-chip capacity was in Taiwan in 2019; key mineral refining is concentrated in a few countries.¹⁷¹⁸

2.5Adoption

ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-ADO-006
78% of organisations reported using AI in 2024, up from 55% in 2023; corporate AI investment reached about US$252 billion.²

2.6Labour

ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-LAB-007
Between about a quarter and two-fifths of global employment is exposed to AI, depending on definition.¹⁶³²
ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-LAB-008
Employers expect 170 million jobs to be created and 92 million displaced by 2030, with technology a leading driver.²⁰

2.7Safety

ReportedEQL 3NRF-XTECH-AI-GLOBAL-2026-BASELINE-SAF-009
Reported AI incidents reached a record 233 in 2024, up 56% on 2023.²²⁷

2.8Governance

ReportedEQL 4NRF-XTECH-AI-GLOBAL-2026-BASELINE-GOV-010
Binding AI law has begun: the EU AI Act applies in phases from 2025, alongside a Council of Europe convention and UN governance bodies.⁴¹⁰¹²

2.9Development

ReportedEQL 4NRF-XTECH-AI-GLOBAL-2026-BASELINE-DEV-011
About 2.6 billion people remained offline in 2024, limiting who can benefit from AI.³⁴

2.10Systemic

Expert inputEQL 2NRF-XTECH-AI-GLOBAL-2026-BASELINE-SYS-012
The leading compound pathway for 2027 testing is an advanced-chip supply disruption coinciding with grid constraints at major data-centre clusters. § scenario

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

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025.back
27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended.back
10Council of Europe (ed.), Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, CETS No. 225. Council of Europe, Vilnius, 2024.back
12United Nations General Assembly (ed.), Resolution establishing the Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on Artificial Intelligence Governance. United Nations, New York, August 2025.back
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024.back

AI and Agentic Systems 2026
Chapter 3

Systems and sectors at baseline

4 min read0 figures2 tables11 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

3.1Where each system stood at evidence close

ENE · Energy

Data-centre electricity use

415TWh

about 1.5% of world electricity; about 945 TWh projected by 2030.

Reported2024IEA¹

NAT · National grid

Data centres’ share of US electricity

4.4%

projected at 6.7–12% by 2028 in the national laboratory’s scenarios.

Reported2023LBNL¹⁵

CMP · Compute

Growth in frontier training compute

4–5×a year

sustained since about 2010.

Reported2010–2024Epoch AI¹⁴

ADO · Adoption

Organisations using AI

78%

up from 55% in 2023; corporate AI investment reached US$252 billion.

Reported2024AI Index²

CHP · Chips

Leading-edge logic capacity in Taiwan

92%

a single-region concentration in the AI supply chain.

Reported2019SIA–BCG¹⁷

SYS · Incidents

Reported AI incidents

233in 2024

a record, up 56% on 2023.

Reported2024AI Index²

3.2Domain posture

Each of the twelve coupled domains is read through AI, in words, not scored. The coupling column names the dependency through which AI carries risk into each domain.

Table 3.1Each domain read in words, not scored
Domain	Posture at baseline	Coupling	Status
WAT	Cooling water demand rising at data-centre clusters; global volumes not consolidated.	AIC → WAT	Expert input EQL 2
ENE	About 415 TWh in 2024, projected to about 945 TWh by 2030.	AIC → ENE	Reported EQL 4
FOO	AI in agriculture is early; supply-chain optimisation spreading.	AIC → FOO	Expert input EQL 2
HLT	Clinical and research uses grow; safety and bias concerns documented.	AIC → HLT	Inferred EQL 2
BIO	Monitoring uses grow; physical footprint adds land and water pressure.	AIC ↔ BIO	Expert input EQL 2
CLI	AI aids forecasting; data-centre emissions depend on the power mix.	AIC ↔ CLI	Inferred EQL 3
INF	Data centres become critical infrastructure; grid connection queues lengthen.	AIC → INF	Reported EQL 3
FIN	Corporate AI investment about US$252 billion in 2024; valuations concentrated.	AIC → FIN	Reported EQL 3
DAT	Cloud and model concentration create common points of failure.	DAT ↔ AIC	Inferred EQL 3
AIC	Compute scaling four to five times a year; costs falling fast.	the core system	Reported EQL 3
MIN	Chips and their minerals concentrated in a few countries.	MIN → AIC	Reported EQL 3
TRD	Export controls on chips, tools and minerals since 2022.	TRD → AIC	Reported EQL 4

3.3Sector screens

All sixteen sectors are screened for their exposure to AI, on the same grammar as every edition. The evidence column describes coverage of the baseline evidence, not the performance of the sector.

Table 3.2Sectors: baseline evidence and the exposure to watch
Code	Sector	Baseline evidence	Exposure to watch	Evidence coverage
GS-00	Cross-system	AI couples energy, water, minerals and labour³⁶	concentrated common dependencies	PARTIAL
GS-01	Water	cooling demand at clusters³⁷	local water competition	THIN
GS-02	Energy	about 1.5% of electricity¹	grid strain at clusters	GOOD
GS-03	Food	early adoption	supply-chain dependence on AI tools	THIN
GS-04	Health	clinical and research uses	safety, bias, privacy	PARTIAL
GS-05	Finance	investment about US$252 billion²	valuation concentration	PARTIAL
GS-06	Development	2.6 billion offline³⁴	unequal access to benefits	GOOD
GS-07	Climate	emissions depend on power mix¹	demand growth against clean supply	PARTIAL
GS-08	Disaster	AI in forecasting and response	over-reliance; failures under stress	THIN
GS-09	Infrastructure	data centres as critical assets	connection queues; single points of failure	PARTIAL
GS-10	Cities	clusters in a few metro regions	local grid and water pressure	PARTIAL
GS-11	Biodiversity	monitoring uses	land and water footprint	THIN
GS-12	Earth observation	AI-assisted analysis growing	model and data provenance	PARTIAL
GS-13	Cyber and digital	AI changes attack and defence²¹	automated attacks on infrastructure	PARTIAL
GS-14	Policy	binding AI law has begun⁴¹⁰	implementation capacity	GOOD
GS-15	Science	AI accelerates research	reproducibility and dual use	PARTIAL

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

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
36S. Valadbaygi (ed.), Global Systems Risk & Resilience Report 2026, Baseline Edition. GCRI for the Nexus Consortium, Toronto, 2026. NR-GLOBAL-SYSTEMS-2026-BASELINE-v1.0.0.back
37S. Valadbaygi (ed.), Global Water Risk & Resilience Report 2026, Baseline Edition. GCRI for the Nexus Consortium, Toronto, 2026. NR-SECTOR-WATER-GLOBAL-2026-BASELINE-v1.0.0.back
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024.back
21World Economic Forum (ed.), Global Cybersecurity Outlook 2025. WEF, Cologny/Geneva, January 2025.back
4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended.back
10Council of Europe (ed.), Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, CETS No. 225. Council of Europe, Vilnius, 2024.back

AI and Agentic Systems 2026
Chapter 4

Risks and capability

Four AI risk objects open the portfolio. Order carries no rank. Each is read across driver, exposure, vulnerability, capacity, cascade and consequence.

3 min read0 figures1 table11 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

4.1Risk objects

Each risk is read across the chain from hazard to consequence, with its status and evidence-quality level.

InferredEQL 3RSK-XT01-MIN-001
Advanced-chip supply disruption
Driver
conflict, disaster or export control in a concentrated supply region
Exposure
all AI compute; wider electronics
Vulnerability
about 92% of leading-edge capacity in one region (2019)¹⁷
Capacity
new fabs; stockpiles; diversification
Cascade
MIN → AIC → FIN, INF
Horizon
1–5 years
ReportedEQL 3RSK-XT01-ENE-002
Grid and water constraints at data-centre clusters
Driver
fast load growth in a few regions
Exposure
local grids, water basins, communities
Vulnerability
long connection queues; water stress
Capacity
grid build-out; efficiency; siting rules
Cascade
AIC → ENE, WAT → cities
Horizon
1–3 years
InferredEQL 2RSK-XT01-SAF-003
Failures and misuse of agentic systems
Driver
autonomous agents acting in real systems
Exposure
finance, cyber, critical services
Vulnerability
thin incident data; limited oversight
Capacity
evaluations, monitoring, risk frameworks⁵⁶
Cascade
AIC → DAT → FIN, INF
Horizon
1–3 years
InferredEQL 3RSK-XT01-LAB-004
Labour displacement outpacing transition
Driver
rapid adoption across tasks
Exposure
a quarter to two-fifths of jobs¹⁶³²
Vulnerability
weak retraining and social protection
Capacity
skills programmes; social protection
Cascade
AIC → labour → FIN, HLT
Horizon
3–5 years

4.2Capability ladder

Existence is not readiness. Each of the fifteen capability functions is placed on the ladder by the strongest public evidence for AI at the global level.

Table 4.1Capabilities on the ladder of evidence: listed, exists, demonstrated, tested
Function	Level	Strongest public evidence at baseline
Prevention and mitigation	EXISTS	risk-management frameworks published⁵⁶⁷
Monitoring and observation	EXISTS	incident databases and monitors²⁷²⁸
Early warning	LISTED	no shared warning system for systemic AI failures
Preparedness and planning	EXISTS	national AI strategies widespread
Redundancy and reserves	LISTED	chip and compute redundancy limited
Response	EXISTS	incident response in firms; cross-border response untested
Continuity of critical services	EXISTS	cloud continuity planning; large outages still occur
Recovery	EXISTS	standard IT recovery practice
Adaptation	EXISTS	skills and education programmes expanding
Social protection	LISTED	little AI-specific transition support
Finance and insurance	EXISTS	AI liability and insurance markets nascent
Institutional coordination	EXISTS	summits, UN panel and dialogue created¹²¹³
Technical infrastructure	EXISTS	compute concentrated in few providers
Workforce	LISTED	skills shortages widely reported
Research capacity	ADEQUATE	large research base; safety research small relative to capability³
A capability listed in policy is not a tested one; each rung is the strongest public evidence at baseline.

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

17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
5NIST (ed.), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. NIST, Gaithersburg, MD, 2023.back
6NIST (ed.), Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1. NIST, Gaithersburg, MD, 2024.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
7ISO and IEC (eds.), ISO/IEC 42001:2023, Information technology, Artificial intelligence, Management system. ISO, Geneva, 2023.back
27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
28OECD (ed.), OECD AI Incidents and Hazards Monitor. OECD.AI, retrieved 2026-09-30.back
12United Nations General Assembly (ed.), Resolution establishing the Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on Artificial Intelligence Governance. United Nations, New York, August 2025.back
13UK Government (ed.), The Bletchley Declaration by Countries Attending the AI Safety Summit, 1–2 November 2023. London, 2023.back
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back

AI and Agentic Systems 2026
Chapter 5

Chip supply disruption under grid constraint

A conditional statement under stated assumptions, never a prediction.

2 min read0 figures0 tables2 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

5.1Assumptions

Scenario SCN-XT01-CHG-01: what is assumed, and what is held constant.

Baseline: the 2026 reference state in this edition
Drivers: a disruption to leading-edge chip supply from a single region, while large data-centre clusters face grid limits
Analogues: 2021 chip shortage; 2023 mineral export controls
Interventions: none modelled, the specification describes the unmitigated case
Run state: narrative specification; quantified run scheduled for 2027

5.2The cascade, step by step

1
MIN → AIC
leading-edge chip output falls sharply
2021 shortage
2
AIC → FIN
compute prices rise; investment plans stall
2021
3
AIC → INF
existing clusters run hotter and longer
peak events
4
ENE → AIC
grid limits cap new capacity at clusters
queues¹⁵
5
AIC → services
AI-dependent services degrade or ration
cloud outages
6
TRD → all
export controls and stockpiling amplify the shock
2023¹⁹

5.3The scenario as a table of steps

How the shock in scenario SCN-XT01-CHG-01 would travel through Artificial Intelligence and Agentic Systems‘s coupled systems, step by step, under the assumptions the edition states.

Step	What happens	Source
01 Minerals to ai compute	Leading-edge chip output falls sharply. Analogue or source: 2021 shortage.	Edition XT-01, scenario SCN-XT01-CHG-01, step 1
02 AI compute to finance	Compute prices rise; investment plans stall. Analogue or source: 2021.	Edition XT-01, scenario SCN-XT01-CHG-01, step 2
03 AI compute to infrastructure	Existing clusters run hotter and longer. Analogue or source: peak events.	Edition XT-01, scenario SCN-XT01-CHG-01, step 3
04 Energy to ai compute	Grid limits cap new capacity at clusters. Analogue or source: queues [15].	Edition XT-01, scenario SCN-XT01-CHG-01, step 4
05 AI compute to services	AI-dependent services degrade or ration. Analogue or source: cloud outages.	Edition XT-01, scenario SCN-XT01-CHG-01, step 5
06 Trade to all	Export controls and stockpiling amplify the shock. Analogue or source: 2023 [19].	Edition XT-01, scenario SCN-XT01-CHG-01, step 6

A teaching model of a conditional scenario under the edition’s stated assumptions: not a forecast, not a prediction and not a warning. Sources: AI and Agentic Systems 2026, scenario SCN-XT01-CHG-01 (2026); AI and Agentic Systems 2026, references (NR-CIT-1) (2026).

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

15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
19Ministry of Commerce of the People’s Republic of China (ed.), Announcement on implementing export control on gallium- and germanium-related items. MOFCOM, Beijing, July 2023.back

AI and Agentic Systems 2026
Chapter 6

Lenses and readings

The AI baseline is read through each steward’s lens, then once as a whole. Each lens is a projection of reviewed findings, not a separate argument.

9 min read0 figures6 tables33 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

6.1Three lenses

One edition, read by each steward. The lens of this site comes first.

GRA · KEY F

Capital Meaning

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

GCRI · KEY T

Technical Truth

The evidence supports reading AI as physical infrastructure with fast-rising, clustered demand for power and water, and a highly concentrated supply chain. Evidence is strongest for electricity use and adoption, and weakest for water, inference energy and agentic failures.

ENE-001 · MIN-005 · SAF-009 · SYS-012

GRF · KEY M

Public Meaning

For most people AI arrives as a tool at work, a change in a job or a new data centre nearby. The baseline asks that public debate weigh the physical and social footprint of AI, power, water, work and access, alongside its benefits.

LAB-007 · DEV-011 · ENE-002

NEXUS

Synthesis

AI is scaling faster than the infrastructure and institutions around it. The 2027 programme should measure the physical couplings, test the chip–grid pathway, and build incident evidence for agentic systems before they become critical.

SYS-012 · X-01 · X-04

6.2Global Risks Alliance reading: 2027 AI readiness pack

The registers that open the 2027 global cycle. At baseline each is initialised from this edition.

Table 6.12027 AI readiness pack
Item	Register	State at baseline	First 2027 entry
E1	AI risk state delta	baseline set by this edition	changes against the baseline
E2	WEFH+B state delta	five global system readings set	indicator updates and freshness
E3	Agenda and Portfolio movement	eight agenda items · six proposed objects	admission decisions
E4	Federation delivery register	Canada, Global and Water issued	national readiness chapters
E5	Consortium capability register	national consortia forming	first contributions
E6	Leadership questions and challenges	none yet	first challenge log
E7	Foundry technical state	national environments prepared	first federation cut
E8	Publication candidates	twelve findings carried forward	PUBLISH · HOLD dispositions
E9	Universe readiness	one candidate mission	admission and role spine
E10	Correction register	none	first federation-wide cycle

Publication dispositions

Publish
Hold
Needs evidence
Restricted
Superseded

6.3GCRI reading: the model card

GCRI’s reading of this edition is its model card: what the national model is, what it computes and where it stops. It is printed in full in Annex A, with the methods it belongs to. Annex A, methodology

6.4Global Risks Forum reading: editorial essays

Four signed essays by the Reports editorial team set out the positions this edition takes. They are interpretation, labelled as such; the evidence they draw on is in the findings register.

Every model runs in a building on a grid in a basin. Treating AI as weightless software hides where its demands land: on particular substations, cooling systems and communities. Reading AI by location is the first step to managing it as infrastructure.

EDITORIAL ESSAY · 01

AI has an address

Leading-edge chips are made in very few places, from minerals refined in very few places, and run in very few clouds. Each concentration is efficient; together they form a chain whose weakest link is a single region.

EDITORIAL ESSAY · 02

Concentration is the risk

Systems that act on their own will fail in ways that chat interfaces do not. Before agents become critical to finance, health or infrastructure, the world needs incident evidence about them, not only principles.

EDITORIAL ESSAY · 03

Agents need evidence

About 2.6 billion people are offline. If AI’s benefits reach only the connected, its risks will still reach everyone. Access, skills and language belong in every account of AI readiness.

EDITORIAL ESSAY · 04

Access is the first safeguard

6.5Nexus Standards reading: standards and regulatory watch

Instruments that shape AI development and use, by status on the retrieval date. The edition assesses no one’s compliance.

IN FORCE
ADOPTED
ADOPTED · OPEN FOR SIGNATURE
PUBLISHED · VOLUNTARY

Table 6.2Standards and regulatory watch
Instrument	Body	Scope	Status	Relevance
Regulation (EU) 2024/1689, AI Act⁴	EU	European Union	IN FORCE · PHASED	risk-based obligations
Framework Convention on AI, CETS 225¹⁰	Council of Europe	open to all states	OPEN FOR SIGNATURE	human rights, rule of law
OECD AI Principles⁸	OECD	adherents	ADOPTED · UPDATED 2024	trustworthy AI
UNESCO Recommendation⁹	UNESCO	global	ADOPTED 2021	ethics
Global Digital Compact¹¹	UN General Assembly	global	ADOPTED 2024	digital cooperation
UN scientific panel and dialogue¹²	UN General Assembly	global	ESTABLISHED 2025	evidence and dialogue
NIST AI RMF and GenAI profile⁵⁶	NIST	voluntary	PUBLISHED · VOLUNTARY	risk management
ISO/IEC 42001⁷	ISO · IEC	international	PUBLISHED	management systems
Hiroshima Code of Conduct³⁰	G7	voluntary	PUBLISHED 2023	advanced AI developers

Changed since 2024

EU AI Act prohibitions and general-purpose model rules began to apply⁴
The UN created an AI scientific panel and global dialogue¹²
The Council of Europe convention opened for signature¹⁰

To watch in 2027

Application of high-risk obligations under the AI Act⁴
First outputs of the UN scientific panel¹²
Ratifications of the Council of Europe convention¹⁰

6.6Risk Academy reading: literature and evidence review, science and assessments

The review behind this edition covers water assessments, peer-reviewed hydrology and the institutional reports that bind its findings. Systematic reviews for each agenda item begin in 2027.

Review protocol

Question: What does the published evidence establish about AI as infrastructure and its couplings?
Sources searched: energy and compute analyses; safety assessments; labour studies; governance instruments; peer-reviewed research and labelled preprints
Window: releases available at evidence close, 30 September 2026
Selection: sources that bind or contextualise a finding in this edition
Next step: systematic reviews for agenda items X-01 to X-05 in 2027

Table 6.3Peer-reviewed science and assessments
Reference	Type	Year	Role	Used for
Bengio (ed.)³	assessment	2025	supports	risks of advanced AI
Sevilla and Roldán¹⁴	research note	2024	supports	compute growth
Strubell et al.²³	proceedings	2019	context	training energy
Patterson et al.²⁴	preprint	2021	context	training emissions
Li et al.²⁵	preprint	2023	context	water footprint
Bommasani et al.²⁶	preprint	2021	framework	foundation models
Cazzaniga et al.¹⁶	staff note	2024	supports	labour exposure

Table 6.4Institutional flagships
Reference	Issuer	Year	Role	Used for
Energy and AI¹	IEA	2025	supports	electricity demand
AI Index 2025²	Stanford HAI	2025	supports	adoption, cost, incidents
Data-centre energy report¹⁵	LBNL	2024	supports	US electricity share
Governing AI for Humanity³¹	UN advisory body	2024	context	global governance

Table 6.5Statistics, labour and access
Reference	Issuer	Year	Role	Used for
Generative AI and jobs³²	ILO	2025	supports	exposure
Future of Jobs²⁰	WEF	2025	supports	job churn
Facts and Figures³⁴	ITU	2024	supports	people offline
Technology and Innovation Report³³	UNCTAD	2025	context	development
Global Cybersecurity Outlook²¹	WEF	2025	context	cyber

Table 6.6Supply chains, incidents and electricity
Reference	Issuer	Year	Role	Used for
Semiconductor supply chain¹⁷	SIA · BCG	2021	supports	chip concentration
Critical Minerals Outlook¹⁸	IEA	2025	supports	mineral refining
Export controls¹⁹²⁹	MOFCOM · BIS	2022–23	supports	trade controls
AI Incident Database²⁷	Responsible AI Collaborative	2026	supports	incidents
AI Incidents Monitor²⁸	OECD	2026	supports	incidents
Global Electricity Review²²	Ember	2025	context	power mix

Where the evidence agrees

Data-centre electricity demand is rising fast¹¹⁵
Compute and adoption are scaling quickly²¹⁴
Chip supply is highly concentrated¹⁷

Where it is unsettled

How large inference demand will become as agents spread¹
Whether AI will mainly augment or substitute work¹⁶³²
How much water AI uses at global scale²⁵

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

4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended.back
10Council of Europe (ed.), Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, CETS No. 225. Council of Europe, Vilnius, 2024.back
8OECD (ed.), Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449, as amended in 2024. OECD, Paris, 2024.back
9UNESCO (ed.), Recommendation on the Ethics of Artificial Intelligence. UNESCO, Paris, 2021.back
11United Nations General Assembly (ed.), Pact for the Future, resolution 79/1, annex I: Global Digital Compact. United Nations, New York, 2024.back
12United Nations General Assembly (ed.), Resolution establishing the Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on Artificial Intelligence Governance. United Nations, New York, August 2025.back
5NIST (ed.), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. NIST, Gaithersburg, MD, 2023.back
6NIST (ed.), Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1. NIST, Gaithersburg, MD, 2024.back
7ISO and IEC (eds.), ISO/IEC 42001:2023, Information technology, Artificial intelligence, Management system. ISO, Geneva, 2023.back
30G7 Hiroshima AI Process (ed.), Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems. G7, Hiroshima, 2023.back
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
23E. Strubell, A. Ganesh and A. McCallum, “Energy and policy considerations for deep learning in NLP,” in Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp. 3645–3650, 2019.back
24D. Patterson, J. Gonzalez, Q. Le et al., “Carbon emissions and large neural network training,” arXiv:2104.10350 (preprint, not peer-reviewed), 2021.back
25P. Li, J. Yang, M. A. Islam and S. Ren, “Making AI less ‘thirsty’: uncovering and addressing the secret water footprint of AI models,” arXiv:2304.03271 (preprint), 2023.back
26R. Bommasani, D. A. Hudson, E. Adeli et al., “On the opportunities and risks of foundation models,” arXiv:2108.07258 (preprint), 2021.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
31UN High-level Advisory Body on Artificial Intelligence (ed.), Governing AI for Humanity: Final Report. United Nations, New York, September 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025.back
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024.back
33UNCTAD (ed.), Technology and Innovation Report 2025: Inclusive Artificial Intelligence for Development. UNCTAD, Geneva, 2025.back
21World Economic Forum (ed.), Global Cybersecurity Outlook 2025. WEF, Cologny/Geneva, January 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
19Ministry of Commerce of the People’s Republic of China (ed.), Announcement on implementing export control on gallium- and germanium-related items. MOFCOM, Beijing, July 2023.back
29U.S. Department of Commerce, Bureau of Industry and Security (ed.), Implementation of additional export controls: certain advanced computing and semiconductor manufacturing items. Federal Register, October 2022, as updated.back
27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
28OECD (ed.), OECD AI Incidents and Hazards Monitor. OECD.AI, retrieved 2026-09-30.back
22Ember (ed.), Global Electricity Review 2025. Ember, London, 2025.back

AI and Agentic Systems 2026
Chapter 7

The 2027 agenda

Eight bounded questions open the 2027 AI agenda. Each comes from a baseline finding, names the initiative that will carry it and the first output due in the 2027 edition.

5 min read0 figures6 tables3 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

7.1Questions for the operational series

X-01
Where will data-centre demand exceed grid and water capacity first, and what does that do to local prices and reliability?
INI-XT01-GRD-01 · compute–grid–water; first output: cluster capacity register.
X-02
How exposed is AI capability to a disruption in advanced chips, tools or minerals?
INI-XT01-CHP-02 · supply concentration; first output: single-point-of-failure map.
X-03
How much energy does inference use as AI agents spread, compared with training?
INI-XT01-INF-03 · inference energy; first output: measured estimates by use.
X-04
Which failure modes of agentic systems appear in real incidents, and how are they caught?
INI-XT01-SAF-04 · agentic incidents; first output: incident taxonomy and evidence grades.
X-05
Is AI augmenting or substituting work, by occupation and region, on measured rather than modelled evidence?
INI-XT01-LAB-05 · measured labour effects; first output: evidence review.
X-06
How is the EU AI Act being applied, and what evidence of compliance capacity exists?
To be admitted in 2027; first output: implementation note.
X-07
How does AI change the cyber threat landscape for critical infrastructure?
With GS-13 cyber; first output: scoping note.
X-08
Which AI capabilities in risk management have been tested in real operations?
Technology capability register; first output: rung evidence for 15 functions.

7.2Portfolio proposed for the Q1 2027 alignment

Five initiatives and one Universe mission candidate are proposed for the 2027 AI portfolio. They are programme objects, no partner, funder or member is attached to them in this edition.

Table 7.1Objects proposed for the first operational quarter
ID	Object	Kind	Question	State
INI-XT01-GRD-01	Compute–grid–water capacity	Initiative	X-01	PROPOSED
INI-XT01-CHP-02	Chip and mineral supply concentration	Initiative	X-02	PROPOSED
INI-XT01-INF-03	Inference energy measurement	Initiative	X-03	PROPOSED
INI-XT01-SAF-04	Agentic incident evidence	Initiative	X-04	PROPOSED
INI-XT01-LAB-05	Measured labour effects	Initiative	X-05	PROPOSED
MSN-XT01-2027-01	Chip–grid compound Universe mission	Mission	X-02	Candidate

7.3The pathway

The AI edition reads national readiness chapters, the sector reports on energy, water and cyber, and the global edition through one governed evidence graph. At baseline the federation is designed; Canada, Global and Water are issued.

01
National readiness chapters
AI and compute in each national edition
CANADA ISSUED · 25 further national baselines
02
Sector couplings
GS-02 energy · GS-01 water · GS-13 cyber
GS-01 ISSUED · energy and cyber baselines
03
Technology system model
compute, energy, chips, labour, governance
v0 SPECIFIED · v1 with federation evidence
04
Strategic links
SR-13 strategic technology · SR-04 minerals · SR-10 DPI
PLANNED · strategic baselines
05
Universe
annual proving environment
CANDIDATE MISSION · mandate in 2027
06
XT-01 report
this series
BASELINE ISSUED · 2027 annual edition
07
Evidence and correction
one graph, many projections
OPEN · first correction cycle in 2027

7.4Universe 2027, evidence pack for the AI candidate mission

01
Mandate
02
Core Build · 1 month
03
Action · 1 week
04
Evidence seal
05
Portfolio correction
06
Next mandate

Table 7.2Universe 2027, evidence pack for the AI candidate mission
Item	Record	State	Timing
F1	Annual mandate record	PLANNED	after the Q2 2027 alignment
F2	Core Build record	PLANNED	one month before the exercise
F3	Exercise evidence	PLANNED	during the Action week
F4	Decision packets	PLANNED	exercise state only
F5	Proof receipts	PLANNED	scoped to each recorded check
F6	After-action report	PLANNED	within one quarter
F7	Upgrade docket	PLANNED	routed to the Portfolio
F8	Portfolio renewal	PLANNED	2027 global annual edition
F9	Public synthesis	PLANNED	public-safe extracts only

Mission

ID: MSN-XT01-2027-01 · chip–grid compound disruption
Role spine: Leadership sponsor · Institutional anchor · Mission lead, drawn from national pathways
Receiving actors: to be identified with the mandate

7.5Handoffs and routing

Where the baseline’s findings go next. A handoff routes a finding for awareness, review or work; it transfers no authority.

Table 7.3Handoffs and routing
Finding	Routed to	Purpose	Class
ENE-001 · electricity	GS-02 Energy edition	grid coupling	PUBLIC
ENE-002 · US share	INI-XT01-GRD-01	cluster capacity	PUBLIC
CMP-003 · compute	public record	reference for the 2027 series	PUBLIC
MIN-005 · chips and minerals	INI-XT01-CHP-02 · SR-04	supply concentration	PUBLIC
LAB-007 · labour	INI-XT01-LAB-05	measured effects	PUBLIC
SAF-009 · incidents	INI-XT01-SAF-04 · GS-13	agentic incidents	PUBLIC
DEV-011 · access	GS-06 Development edition	access coupling	PUBLIC
SYS-012 · compound pathway	MSN-XT01-2027-01	Universe candidate	CONSORTIUM

Table 7.4Routing rules by publication class
Class	Where it goes	Who can see it
PUBLIC	the published edition and its evidence bundle	anyone
CONSORTIUM	governed rooms of the federation	consortium members under agreement
CONTROLLED	need-to-know channels	named competent actors only

7.6Feeds and signal register

Table 7.5Feeds watched from 2027
Feed	Producer	Latency	Feeds into
AI Incident Database	Responsible AI Collaborative	continuous	SAF-009
AI Incidents and Hazards Monitor	OECD	continuous	SAF-009
Electricity market reports	IEA	monthly	ENE-001
Compute and model trends	Epoch AI	continuous	CMP-003
Export-control notices	national authorities	per notice	MIN-005
Grid interconnection queues	system operators	quarterly	RSK-XT01-ENE-002
Model and system cards	developers	per release	capability register

Escalation routes

none · keep monitoring
research question
project
risk object
annual publication
signal report

Table 7.6Signals at baseline
ID	Signal	Corroboration	Systems	Route
SIG-XT01-001	Clustered data-centre load lengthening grid queues¹⁵	CORROBORATED	ENE · INF	initiative X-01
SIG-XT01-002	Tightening export controls on chips, tools and minerals¹⁹²⁹	CORROBORATED	MIN · TRD	initiative X-02
SIG-XT01-003	Incidents involving autonomous agents	PARTIAL	AIC · DAT · FIN	initiative X-04

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

15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
19Ministry of Commerce of the People’s Republic of China (ed.), Announcement on implementing export control on gallium- and germanium-related items. MOFCOM, Beijing, July 2023.back
29U.S. Department of Commerce, Bureau of Industry and Security (ed.), Implementation of additional export controls: certain advanced computing and semiconductor manufacturing items. Federal Register, October 2022, as updated.back

AI and Agentic Systems 2026
Chapter 8

Systems in depth

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

5 min read0 figures5 tables17 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

8.1Compute and data centres

AI capability rests on compute, and compute rests on data centres: buildings full of accelerators that need power, cooling, land and connectivity.

4–5×a year

frontier training compute growth¹⁴

≈30% a year

accelerated-server demand growth¹

>280×cheaper

fixed-level use, 2022–2024²

Table 8.1State at baseline
Component	State	Evidence	Status
Training compute	four to five times a year since about 2010	¹⁴	Reported
Inference	demand rising as use spreads; poorly measured	¹	Inferred
Cost	cost of fixed-level use collapsing	²	Reported
Concentration	compute concentrated in few providers and regions	¹	Inferred

Couplings

AIC → ENE electricity
AIC → WAT cooling
DAT ↔ AIC cloud dependence

Evidence to bind in 2027

inference energy by use
utilisation of deployed compute

Related editions

GLB Global Systems
GS-02 Energy
GS-13 Cyber
SR-04 Minerals
SR-13 Strategic technology
XT-06 Semiconductors

8.2Energy and water

AI demand lands on particular grids and basins. The global share of electricity is small; local shares at clusters are large and rising.

415TWh

data-centre electricity, 2024¹

4.4%

of US electricity, 2023¹⁵

≈945TWh

projected for 2030¹

Table 8.2State at baseline
Component	State	Evidence	Status
Electricity	about 1.5% of world use	¹	Reported
Local grids	connection queues at clusters	¹⁵	Reported
Water	cooling demand at water-stressed sites	²⁵³⁷	Inferred
Emissions	depend on the power mix of each grid	¹²²	Reported

Couplings

AIC → ENE load growth
AIC → WAT cooling demand
ENE → AIC power availability

Evidence to bind in 2027

grid capacity at clusters
water use outside the United States

Related editions

GLB Global Systems
GS-02 Energy
GS-13 Cyber
SR-04 Minerals
SR-13 Strategic technology
XT-06 Semiconductors

8.3Chips and minerals

The AI supply chain is among the most concentrated in the world, from mineral refining to chip fabrication.

92%

leading-edge logic capacity in Taiwan, 2019¹⁷

2023controls

gallium and germanium export controls¹⁹

2022controls

advanced-chip export rules²⁹

Table 8.3State at baseline
Component	State	Evidence	Status
Fabrication	leading-edge capacity in one region	¹⁷	Reported
Minerals	refining concentrated in a few countries	¹⁸	Reported
Trade controls	chips, tools and minerals since 2022	¹⁹²⁹	Reported
New capacity	fabs under construction elsewhere	¹⁷	Inferred

Couplings

MIN → AIC chips
TRD → AIC controls
WAT and ENE → fabs

Evidence to bind in 2027

current capacity shares by node
time to recover from a disruption

Related editions

GLB Global Systems
GS-02 Energy
GS-13 Cyber
SR-04 Minerals
SR-13 Strategic technology
XT-06 Semiconductors

8.4Economy and workforce

AI is spreading through work faster than evidence on its effects. Exposure is broad; whether it brings augmentation or substitution is the open question.

78%

organisations using AI, 2024²

25–40%

of jobs exposed¹⁶³²

170 · 92million

jobs created · displaced by 2030²⁰

Table 8.4State at baseline
Component	State	Evidence	Status
Adoption	78% of organisations in 2024	²	Reported
Exposure	a quarter to two-fifths of jobs	¹⁶³²	Reported
Job churn	170 million created and 92 million displaced by 2030	²⁰	Reported
Measured effects	limited causal evidence so far	,	Expert input

Couplings

AIC → labour tasks and skills
AIC → FIN productivity and investment
labour → HLT income and wellbeing

Evidence to bind in 2027

measured employment effects by occupation
access to retraining

Related editions

GLB Global Systems
GS-02 Energy
GS-13 Cyber
SR-04 Minerals
SR-13 Strategic technology
XT-06 Semiconductors

8.5Safety, security and failure modes

Harms are rising with use. Agentic systems that act on their own add new failure modes, while incident evidence remains thin.

233incidents

reported in 2024, a record²

+56%

on 2023²

96experts

in the International AI Safety Report³

Table 8.5State at baseline
Component	State	Evidence	Status
Incidents	record 233 reported in 2024	²²⁷	Reported
Risk taxonomy	malicious use, malfunction, systemic risks	³	Reported
Cyber	AI changes attack and defence	²¹	Inferred
Agentic failures	little systematic evidence	,	Expert input

Couplings

AIC → DAT cyber exposure
AIC → FIN automated decisions
AIC → information integrity

Evidence to bind in 2027

agentic incident taxonomy
evaluation results for deployed agents

Related editions

GLB Global Systems
GS-02 Energy
GS-13 Cyber
SR-04 Minerals
SR-13 Strategic technology
XT-06 Semiconductors

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

14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
25P. Li, J. Yang, M. A. Islam and S. Ren, “Making AI less ‘thirsty’: uncovering and addressing the secret water footprint of AI models,” arXiv:2304.03271 (preprint), 2023.back
37S. Valadbaygi (ed.), Global Water Risk & Resilience Report 2026, Baseline Edition. GCRI for the Nexus Consortium, Toronto, 2026. NR-SECTOR-WATER-GLOBAL-2026-BASELINE-v1.0.0.back
22Ember (ed.), Global Electricity Review 2025. Ember, London, 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
19Ministry of Commerce of the People’s Republic of China (ed.), Announcement on implementing export control on gallium- and germanium-related items. MOFCOM, Beijing, July 2023.back
29U.S. Department of Commerce, Bureau of Industry and Security (ed.), Implementation of additional export controls: certain advanced computing and semiconductor manufacturing items. Federal Register, October 2022, as updated.back
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025.back
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back
27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
21World Economic Forum (ed.), Global Cybersecurity Outlook 2025. WEF, Cologny/Geneva, January 2025.back

AI and Agentic Systems 2026
Annex A

Methodology

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

9 min read0 figures12 tables6 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

A.1Methods

The global baseline uses published values and four simple derivations. Each is shown in full so that any reader can reproduce it.

Table A.1Methods
No.	Derivation	Calculation	Result
(1)	Growth in data-centre electricity to 2030	945 ÷ 415	≈ 2.3×
(2)	Increase in organisational AI use, 2023–2024	78 − 55	23 points
(3)	Net job change expected by 2030	170 − 92	≈ 78 million
(4)	Compute growth over five years at 4–5× a year	4⁵ to 5⁵	≈ 1,000–3,000×

Table A.2Coupling strength scale (matrix and model card)
Strength	Definition
STRONG	a disruption in the source system has historically changed the target system’s service or output within one season, in more than one region
MODERATE	an effect is documented but buffered or slower
WEAK	an effect is plausible but not documented at global scale

Table A.3Status grammar
Label	Rule
Reported	value as issued by its source, with year
Inferred	reasoned from cited evidence; never a measurement
Estimated	Nexus analytical value with stated method
Scenario	conditional on stated assumptions
Expert input	structured editorial judgement awaiting evidence

A.2Model card, Technology System Model v0

Model ID: MDL-XT01-AI · v0.1
Purpose: represent AI as infrastructure coupled to energy, water, minerals, labour and institutions
Family: graph-based system of systems
Inputs: the cited sources of this edition; no proprietary data
Method: expert-specified coupling strengths (STRONG · MODERATE · WEAK) from assessments and analogue events
Validation: none yet, first validation against federation evidence in 2027
Uncertainty: qualitative; strengths are hypotheses, not measurements
Domain of validity: global scale; not for site, company or model-level use
Reproducibility: strength table and sources published in this edition

20couplings

typed in v0

11strong

documented within one season

5moderate

documented, buffered

4weak

plausible, undocumented

A.3The baseline evidence cut

Every Nexus edition is published from a frozen evidence cut. The global baseline cut is documentary: it pins the exact edition of every source cited. Federation cuts begin with the first national Foundry cuts in 2027.

Cut ID: DC-XTECH-AI-GLOBAL-2026-B0
Type: baseline documentary cut
Resolution policy: pinned · as of 30 September 2026
Sources pinned: 40 cited sources, each at the edition and retrieval date in the references
Model runs: none at baseline, analytical values are stated with their method
Digest: computed at freeze and recorded in the publication record
Supersedes: nothing, first cut of the series
Superseded by: the 2027 AI annual cut

Lineage

01
source edition
02
pinned citation
03
finding
04
page
05
publication record

Table A.4From the baseline cut to the federation cut
Element	Baseline 2026	2027 global annual
Sources	cited global publications, pinned by edition	national, regional and sector cuts through one graph
Analysis	stated methods and editorial estimates	typed analysis specifications
Models	Technology System Model v0 specification	model runs with receipts
Findings	twelve baseline findings	findings re-bound to federation evidence
Review	five passes scheduled	five passes on the exact digest

A.4Evidence age and coverage

A 2026 global edition rests on evidence observed in many different years. This page shows each time base, so that no single year label hides older evidence.

Table A.5Evidence age and coverage
Indicator	Observed	Released	Cadence	Next release
Data-centre electricity use	2024	Apr 2025	periodic	next IEA update
US data-centre share	2023	Dec 2024	periodic	next LBNL report
Organisations using AI	2024	Apr 2025	annual	AI Index 2026
Training compute trend	2010–2024	2024	continuous	continuous
Leading-edge chip capacity share	2019	2021	one-off	new studies
Reported AI incidents	2024	Apr 2025	annual	AI Index 2026
Jobs exposed to AI	2024–2025	2024–2025	periodic	new estimates
People offline	2024	Nov 2024	annual	annual

Table A.6Coverage by system at baseline
System	Coverage	Thinnest dimension
Energy	GOOD	inference against training
Water	THIN	use outside the United States
Compute and cost	GOOD	efficiency of deployed systems
Chips and minerals	PARTIAL	current capacity shares
Adoption	PARTIAL	small firms and public sector
Labour	PARTIAL	measured, not modelled, effects
Safety	THIN	agentic incidents
Governance	GOOD	implementation evidence

A.5Evidence cards, two anchor sources

Issued by the International Energy Agency from data-centre, grid and industry data with scenario modelling to 2030. Role: binds ENE-001 and the 2030 projection, labelled as a projection.

EVD-XT01-001

Energy and AI¹

Issued by Stanford HAI from surveys, public datasets and incident databases. Annual. Role: binds adoption, cost, investment and incident findings.

EVD-XT01-002

Artificial Intelligence Index Report 2025²

Table A.7Quality dimensions at baseline
Dimension	EVD-GLB-001	EVD-GLB-002
Completeness	company disclosure incomplete	survey samples skew to large firms
Timeliness	about a year after the data	about four months after the year
Revision	projections revised with new data	series revised annually
Method transparency	methods and scenarios published	methods published by chapter
External validation	compared with national laboratory estimates	widely cited; no formal audit

Evidence cards to follow in 2027

LBNL data-centre report¹⁵, US electricity and water
Epoch AI compute trends¹⁴, method for frontier compute
International AI Safety Report³, risk taxonomy
ILO exposure index³², occupational exposure method

A.6Publication acceptance record

Editorial self-check at evidence close. Every applicable item carries a disposition; the remaining items are review, approval and deposit.

Table A.8Publication acceptance record
Group	Area	Items	Pass	N/A	Pending
D1	Identity and scope	8	8	0	0
D2	Source and data	12	9	1	2
D3	Analysis and models	12	3	8	1
D4	Findings and claims	9	8	0	1
D5	Figures, tables, maps	10	6	2	2
D6	Editorial and citations	10	8	0	2
D7	Review and approval	13	0	0	13
D8	Artefact and design	12	7	0	5
D9	Registry and release	9	0	0	9

Publication gate

Conditions met: 14 of 23 at evidence close
Blocking problems: none recorded
Open conditions: reviews, approval of the exact digest, deposit, registry pointer

Not applicable at baseline

Model runs: first runs in Q1 2027
Maps: plates in Q2 2027
Waivers: none

A.7Status words

Table A.9The status every value carries
Status	What it means
Observed	Measured directly and published by the source.
Reported	Published statistics of an official or institutional source.
Declared	Stated by the body concerned; not independently measured.
Estimated	A Nexus estimate from cited evidence, with its method in Annex A.
Modelled	Produced by a stated model from cited inputs.
Inferred	Reasoned from cited evidence; not measured.
Forecast	A projection by the cited source.
Simulated	A simulation result under stated assumptions.
Scenario	A conditional statement under stated assumptions; never a prediction.
Expert input	Expert judgement, recorded with its basis.
Candidate	Proposed for a later edition; not yet evidence.
Evidence-quality levels (EQL 1 to 5) grade the source behind a value; they are not a score of the subject.

A.8How the edition is made, released and corrected

Every edition is built from a sealed evidence cut, reviewed under three keys, signed and deposited before any site shows it as published; corrections go to Nexus Standards as change proposals and feed the next cut.

Step	What happens	Source
01 Evidence arrives from the pillars	Foundry seals observations and indicator series, Universe hands over after-action and scenario findings, and contributions are screened at every gateway, including this site’s.
02 The cut is sealed	On the cut date the evidence is pinned under one identifier, so anyone can rebuild the edition from exactly what it was built on.
03 The edition is assembled	The edition is drafted from its template and content pack. Agents may draft; they never publish and never decide.
04 The keys turn	The Global Risks Alliance turns the capital key: its steward checks that no finding reads as advice, a rating or an underwriting decision. No edition moves on until every key has been turned by a named steward.
05 The gate opens	A named approver releases one exact digest of the edition; any change after this point is a new version, never a silent edit.
06 Signed and deposited	The PDF, the web edition and the data bundle are signed together with a manifest and an RO-Crate description, and deposited on Zenodo with a concept and a version DOI.
07 Published on five sites	Only then do the five Reports sites show the edition as published, each reading it through its own lens.
08 Corrections go to Standards	Method and correction findings go to Nexus Standards as change proposals; citations, corrections and new contributions feed the next evidence cut.

A teaching model of the process, not a forecast and not a record of any run; dates and counts on this page come from the programme record. Sources: Reports governance at Global Risks Alliance (2026); Zenodo, repository of record for Reports deposits (2026).

Lifecycle, one state at a time: Draft, Computed, Evidence bound, Reviewed, Approved for release, Published, Corrected, Superseded. This edition is a draft.

A.9Glossary and abbreviations

Table A.10Series terms
Term	Meaning
Baseline Edition	the inaugural annual edition that fixes the reference state for the series
WEFH+B	water, energy, food, health and biodiversity, the national nexus core
Coupling	a typed dependency through which one system transmits to another
Protection gap	the share of losses not covered by insurance
Capability ladder	LISTED → EXISTS → ADEQUATE → TESTED → DEMONSTRATED
EQL	evidence-quality level, 1–5
Evidence cut	a frozen resolution of exact sources and versions
NNC · NLC · NWG	National Nexus Consortium · National Leadership Council · National Working Groups

Table A.11Abbreviations used in this edition
Abbreviation	Expansion
AI	artificial intelligence
AI RMF	NIST Artificial Intelligence Risk Management Framework
BIS	US Bureau of Industry and Security
CETS	Council of Europe Treaty Series
GPU	graphics processing unit, used as an AI accelerator
HAI	Stanford Institute for Human-Centered Artificial Intelligence
IEA	International Energy Agency
ILO	International Labour Organization
ITU	International Telecommunication Union
LBNL	Lawrence Berkeley National Laboratory
TWh	terawatt-hour
UNESCO	United Nations Educational, Scientific and Cultural Organization

A.10Technical annex

Table A.12Technical annex
Item	Annex	Contents	Status
A.1	Indicator definitions	the global indicator set, with units and sources	IN THIS EDITION
A.2	Coupling strengths	the twenty-cell matrix and its rationale	IN THIS EDITION
A.3	Scenario specification	SCN-XT01-CHG-01 assumptions and signposts	IN THIS EDITION
A.4	Capability placements	rung evidence for fifteen functions	IN THIS EDITION
A.5	Model diagnostics	first runs and validation	2027
A.6	Sensitivity analyses	influence of coupling strengths	2027
A.7	Cluster notes	cards for the major data-centre clusters	2027
A.8	Review statements	pass-by-pass review record	AT RELEASE

A.1 · indicator definition, data-centre electricity use

Construct: annual electricity consumed by data centres, including cooling
Unit: TWh a year
Method: IEA bottom-up estimate from server, storage, network and cooling loads
Source: IEA, Energy and AI¹
Uncertainty: company disclosure incomplete; range published
Known limits: does not separate AI from other workloads exactly

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

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back

AI and Agentic Systems 2026
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.

7 min read0 figures8 tables16 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

B.1The indicator set

Table B.1Global baseline indicator set
Indicator	Value	Unit	Year	State	Source
Data-centre electricity use	≈415	TWh	2024	Reported	¹
Share of world electricity	≈1.5	%	2024	Reported	¹
Projected data-centre use	≈945	TWh	2030	Scenario	¹
US data-centre share of electricity	4.4	%	2023	Reported	¹⁵
Frontier training compute growth	4–5×	a year	2010–2024	Reported	¹⁴
Fall in cost of fixed-level use	>280×	fold	2022–2024	Reported	²
Organisations using AI	78	%	2024	Reported	²
Corporate AI investment	≈252	US$ billion	2024	Reported	²
Reported AI incidents	233	incidents	2024	Reported	²
Leading-edge logic capacity in Taiwan	92	%	2019	Reported	¹⁷
Jobs exposed to AI, IMF	≈40	%	2024	Reported	¹⁶
Jobs exposed to generative AI, ILO	≈25	%	2025	Reported	³²
Jobs created by 2030	170	million	2025–2030	Reported	²⁰
Jobs displaced by 2030	92	million	2025–2030	Reported	²⁰
People offline	2.6	billion	2024	Reported	³⁴

Indicators, CSV Indicators, JSON

B.2Global source register

The governed source set behind this edition. Each source keeps its authority, coverage, cadence and role; categories describe origin, not truth.

Domains bound by the register

Table B.3Global source register
Source	Category	Coverage	Period	Cadence	Role
Energy and AI¹	intergovernmental	global	2024–2030	periodic	ENE
AI Index 2025²	university	global	2024	annual	adoption
International AI Safety Report³	expert panel	global	2025	periodic	safety
Data-centre energy report¹⁵	national laboratory	United States	2023	periodic	ENE
Compute trends¹⁴	research institute	global	2010–2024	continuous	compute
Semiconductor supply chain¹⁷	industry · consultancy	global	2019	one-off	chips
Critical Minerals Outlook¹⁸	intergovernmental	global	2024	annual	MIN
Labour exposure¹⁶³²	IMF · ILO	global	2024–2025	periodic	labour
Facts and Figures³⁴	UN agency	global	2024	annual	access
Incident databases²⁷²⁸	civil society · OECD	global	continuous	continuous	safety

B.3Regional reading, AI across ten clusters

Regional plates follow with the regional editions. This page gives the regional reading of AI infrastructure and exposure in table form, and ranks no region.

Table B.4Regional reading, AI across ten clusters
Region	Principal hazards	Systems most exposed	Baseline evidence
North America	largest data-centre load; grid queues	power, water, labour	LBNL¹⁵; IEA¹
South America	early adoption; clean-power potential	energy, skills	assessments³³
Europe	binding AI law; grid constraints	governance, power	AI Act⁴
Eurasia	limited public data	access to chips	export controls²⁹
Middle East and North Africa	new data-centre investment; water stress	water, power	IEA¹
Africa	connectivity gaps	access, skills	ITU³⁴
South Asia	large workforce exposure	labour, skills	ILO³²
East Asia	chip manufacturing concentration	supply chain	SIA–BCG¹⁷
ASEAN	fast-growing data-centre hubs	power, water	IEA¹
Oceania-Pacific	small-island access gaps	connectivity	ITU³⁴

B.4Cluster card, leading-edge chip manufacturing

supply chain
single region
chips
minerals

Systems in view: compute, electronics, trade, finance
Defining fact: about 92% of leading-edge logic capacity in one region in 2019¹⁷
Couplings: MIN → AIC chips; ENE and WAT → fabs; TRD → AIC export controls
Capability evidence: new fabs announced in several regions; stockpiles limited
Gaps: current capacity shares; time to recover from disruption
Agenda items: X-02

Leading-edge chips depend on a handful of fabs, tool makers and mineral refiners. Each is efficient; together they form a chain with few alternatives.

SUPPLY

One region, one chain

Chip fabrication uses large volumes of ultra-pure water and reliable power, coupling chip supply to droughts and grid events.

PHYSICAL INPUTS

Fabs need water and power

Export controls on chips, tools and minerals since 2022 add policy risk to physical risk.

TRADE

Controls tighten

B.5How countries appear in the AI edition

Countries enter the AI edition through the readiness chapters of their own national editions, on one card grammar. They are read against peer distributions and thresholds, never ranked.

Table B.5How countries appear in the AI edition
Card field	What it shows	Source
WEFH+B state	each core system against peer distribution bands	national edition
Couplings	the national dependency matrix	national edition
Risk objects	the national risk portfolio	national edition
Capability rungs	the national capability ladder	national edition
Evidence age	oldest and newest source dates	national source register

Editions in the 2026 Baseline programme

26national

countries and unions

10regional

clusters

16global

flagship and sector

14strategic

systems reports

20frontier

technology reports

How to read a country card

Each system is shown against the distribution of its peers, with thresholds where they exist
Couplings and risk objects come from the country’s own national edition
Capability rungs record evidence, not a judgement of institutions
Evidence age is printed, so older national data are visible as such

Table B.6National Baseline Editions in the 2026 programme
Wave	Countries and unions	State
W0	Canada	ISSUED
W1	United States, United Kingdom, France, Germany, Italy, Japan, European Union, Switzerland	PLANNED
W2	Mexico, Brazil, Argentina, Türkiye, Russian Federation, Saudi Arabia, United Arab Emirates, South Africa, Senegal, Kenya, African Union, India, China, Republic of Korea, Indonesia, Singapore, Australia	PLANNED

B.6Licences

Table B.7Licences of each kind of material
Material	Licence
Text	CC BY 4.0
Data	on the licence of each source; Nexus tables CC BY 4.0
Figures	Nexus figures CC BY 4.0; third-party as credited
Maps	boundary source and licence printed on each plate
Metadata	CC0 1.0, as Zenodo publishes it
Third-party data, figures and text	The source’s own licence, credited where it is used; never relicensed
Nexus-authored text, tables and figures: CC BY 4.0 (SPDX CC-BY-4.0). Map plates print their boundary source and licence.

B.7Formats and provenance

Table B.8This edition in every format
Format	Kind	State	What it is	Identifier
Web edition	HTML	On this site	Draft in stewardship review; every value resolves to its source.	NR-XTECH-AI-GLOBAL-2026-BASELINE-v1.0.0
Print edition	PDF	At release	A4 portrait · screen, print and archival PDF, signed with the web edition at release.	NR-XTECH-AI-GLOBAL-2026-BASELINE-v1.0.0
Indicator data	CSV · JSON	On this site	15 indicators with unit, year, status and source; the machine-readable copy ships with the evidence bundle at release.	NR-XTECH-AI-GLOBAL-2026-BASELINE data release v1
Figures	SVG · PNG	On this site	4 on the web edition: headline figure, figure 1.3, coupling instrument, capability ladder.	FIG-XTECH-AI-GLOBAL-2026-BASELINE-*
Slides	PDF	Not recorded	No slide deck is recorded for this edition.
Evidence brief	HTML · PDF	Not recorded	No evidence brief is recorded for this edition.
Dataset with DOI	DataCite	At release	Concept and version DOI are minted at deposit; none is recorded yet.

Provenance

Edition pack: The edition’s content pack (JSON), from which the print and web editions are built
Evidence cut: DC-XTECH-AI-GLOBAL-2026-B0, closed 30 September 2026
Reproducibility: sourced values cited; analytical values reproducible from stated parameters
Self-check: 14 of 23 at evidence close; a fail on any item blocks release
Review: five passes scheduled before release
Publisher of record: GCRI for the Nexus Consortium
Corrections: Registry correction pathway; upheld corrections are published as notices with a new version

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

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024.back
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021.back
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024.back
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025.back
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024.back
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
28OECD (ed.), OECD AI Incidents and Hazards Monitor. OECD.AI, retrieved 2026-09-30.back
33UNCTAD (ed.), Technology and Innovation Report 2025: Inclusive Artificial Intelligence for Development. UNCTAD, Geneva, 2025.back
4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended.back
29U.S. Department of Commerce, Bureau of Industry and Security (ed.), Implementation of additional export controls: certain advanced computing and semiconductor manufacturing items. Federal Register, October 2022, as updated.back

AI and Agentic Systems 2026
Annex C

References and record

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

10 min read0 figures1 table0 notesNR-XTECH-AI-GLOBAL-2026-BASELINE

C.140 sources

NR-CIT-1, editor first. A DOI is linked where the source has one; Nexus records carry a UNI once the catalogue allocates it (NS-1).

1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 6, chapter 8
2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 6, chapter 8
3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025. Cited in chapter 1, chapter 4, chapter 6, chapter 8
4European Parliament and Council of the European Union (eds.), Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal L, 12 July 2024, as amended. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 6
5NIST (ed.), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. NIST, Gaithersburg, MD, 2023. Cited in chapter 4, chapter 6
6NIST (ed.), Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1. NIST, Gaithersburg, MD, 2024. Cited in chapter 4, chapter 6
7ISO and IEC (eds.), ISO/IEC 42001:2023, Information technology, Artificial intelligence, Management system. ISO, Geneva, 2023. Cited in chapter 4, chapter 6
8OECD (ed.), Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449, as amended in 2024. OECD, Paris, 2024. Cited in chapter 6
9UNESCO (ed.), Recommendation on the Ethics of Artificial Intelligence. UNESCO, Paris, 2021. Cited in chapter 6
10Council of Europe (ed.), Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, CETS No. 225. Council of Europe, Vilnius, 2024. Cited in chapter 1, chapter 2, chapter 3, chapter 6
11United Nations General Assembly (ed.), Pact for the Future, resolution 79/1, annex I: Global Digital Compact. United Nations, New York, 2024. Cited in chapter 6
12United Nations General Assembly (ed.), Resolution establishing the Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on Artificial Intelligence Governance. United Nations, New York, August 2025. Cited in chapter 1, chapter 2, chapter 4, chapter 6
13UK Government (ed.), The Bletchley Declaration by Countries Attending the AI Safety Summit, 1–2 November 2023. London, 2023. Cited in chapter 4
14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 6, chapter 8
15A. Shehabi, S. J. Smith, A. Hubbard et al., 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory, Berkeley, CA, December 2024. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 5, chapter 6, chapter 7, chapter 8
16M. Cazzaniga, F. Jaumotte, L. Li et al., “Gen-AI: Artificial Intelligence and the Future of Work,” IMF Staff Discussion Note SDN/2024/001. IMF, Washington, DC, 2024. Cited in the summary, chapter 1, chapter 2, chapter 4, chapter 6, chapter 8
17Semiconductor Industry Association and Boston Consulting Group (eds.), Strengthening the Global Semiconductor Supply Chain in an Uncertain Era. SIA and BCG, Washington, DC, 2021. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 4, chapter 6, chapter 8
18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025. Cited in the summary, chapter 1, chapter 2, chapter 6, chapter 8
19Ministry of Commerce of the People’s Republic of China (ed.), Announcement on implementing export control on gallium- and germanium-related items. MOFCOM, Beijing, July 2023. Cited in chapter 1, chapter 5, chapter 6, chapter 7, chapter 8
20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025. Cited in chapter 1, chapter 2, chapter 6, chapter 8
21World Economic Forum (ed.), Global Cybersecurity Outlook 2025. WEF, Cologny/Geneva, January 2025. Cited in chapter 3, chapter 6, chapter 8
22Ember (ed.), Global Electricity Review 2025. Ember, London, 2025. Cited in chapter 6, chapter 8
23E. Strubell, A. Ganesh and A. McCallum, “Energy and policy considerations for deep learning in NLP,” in Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp. 3645–3650, 2019. Cited in chapter 6
24D. Patterson, J. Gonzalez, Q. Le et al., “Carbon emissions and large neural network training,” arXiv:2104.10350 (preprint, not peer-reviewed), 2021. Cited in chapter 6
25P. Li, J. Yang, M. A. Islam and S. Ren, “Making AI less ‘thirsty’: uncovering and addressing the secret water footprint of AI models,” arXiv:2304.03271 (preprint), 2023. Cited in chapter 1, chapter 6, chapter 8
26R. Bommasani, D. A. Hudson, E. Adeli et al., “On the opportunities and risks of foundation models,” arXiv:2108.07258 (preprint), 2021. Cited in chapter 6
27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30. Cited in chapter 2, chapter 4, chapter 6, chapter 8
28OECD (ed.), OECD AI Incidents and Hazards Monitor. OECD.AI, retrieved 2026-09-30. Cited in chapter 4, chapter 6
29U.S. Department of Commerce, Bureau of Industry and Security (ed.), Implementation of additional export controls: certain advanced computing and semiconductor manufacturing items. Federal Register, October 2022, as updated. Cited in chapter 1, chapter 6, chapter 7, chapter 8
30G7 Hiroshima AI Process (ed.), Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems. G7, Hiroshima, 2023. Cited in chapter 6
31UN High-level Advisory Body on Artificial Intelligence (ed.), Governing AI for Humanity: Final Report. United Nations, New York, September 2024. Cited in chapter 6
32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025. Cited in the summary, chapter 1, chapter 2, chapter 4, chapter 6, chapter 8
33UNCTAD (ed.), Technology and Innovation Report 2025: Inclusive Artificial Intelligence for Development. UNCTAD, Geneva, 2025. Cited in chapter 6
34ITU (ed.), Measuring Digital Development: Facts and Figures 2024. International Telecommunication Union, Geneva, 2024. Cited in the summary, chapter 1, chapter 2, chapter 3, chapter 6
35S. Valadbaygi (ed.), Reports, The Foundational Edition, Volumes I and II. Reports series, Edition 2026–2030, V1.0. GCRI for the Nexus Consortium, Toronto, 2026.
36S. Valadbaygi (ed.), Global Systems Risk & Resilience Report 2026, Baseline Edition. GCRI for the Nexus Consortium, Toronto, 2026. NR-GLOBAL-SYSTEMS-2026-BASELINE-v1.0.0. Cited in chapter 3
37S. Valadbaygi (ed.), Global Water Risk & Resilience Report 2026, Baseline Edition. GCRI for the Nexus Consortium, Toronto, 2026. NR-SECTOR-WATER-GLOBAL-2026-BASELINE-v1.0.0. Cited in chapter 3, chapter 8
38S. Valadbaygi (ed.), Canada Systems Risk & Resilience Report 2026, Baseline Edition. GCRI for the Nexus Consortium, Toronto, 2026. NR-COUNTRY-CAN-2026-BASELINE-v1.0.0.
39International Energy Agency (ed.), Global Energy Review 2025. IEA, Paris, 2025.
40IFRS Foundation (ed.), IFRS S2 Climate-related Disclosures. ISSB, London, 2023.

C.2Publication record

Publication

Title: Exponential Technology Report 2026, AI and Agentic Systems, Baseline Edition
Series: Reports · Exponential Technology Series XT-01
Edition: Baseline Edition 2026 · inaugural annual
Version: v1.0.0 · draft for stewardship review
Publication ID: NR-XTECH-AI-GLOBAL-2026-BASELINE-v1.0.0
Evidence close: 30 September 2026
Publication date: October 2026
DOI: concept and version DOI at deposit

Publisher

Publisher of record: The Global Centre for Risk and Innovation (GCRI), for the Nexus Consortium
Place: Toronto
Editor: Saeed Valadbaygi, Chief Steward and Managing Editor
Stewards: GCRI · technical truth · GRF · public meaning · GRA · capital meaning
Licence: text CC BY 4.0; data on the terms of each source
Language: English (en-GB) source edition
Correspondence: Registry correction pathway · editorial office

Colophon

Format: A4 portrait · screen, print and archival PDF
Typefaces: Space Grotesk (display) · Geist (text) · IBM Plex Sans Italic (standfirsts) · IBM Plex Mono (data)
Design standard: Reports Foundational Edition, Volume I, Part VII
Accessibility: tagged PDF and HTML targets; text alternatives for every figure
Production: Reports editorial team, GCRI

Release path for this version

Table C.1Release path for this version
Step	What happens	When
Draft	editorial self-check complete at evidence close	October 2026
Steward review	five passes on the exact draft (keys T · M · F · N)	Q4 2026
Approval	publication authority approves the exact digest	before release
Deposit	concept and version DOI minted; registry pointer set	at release
Succession	the 2027 AI edition reports changes against this baseline	2027

Current version: v1.0.0 draft, stewardship review
Previous edition: none, first edition of the series
Next edition: NR-XTECH-AI-GLOBAL-2027-v1.0.0

Latest version: resolve the concept DOI or the registry record
Correction route: Registry correction pathway
Notice types: erratum · addendum · supersession · withdrawal

C.3Corrections log

0correction notices. No edition has been released yet, so none can be corrected; the log below lists each edition’s current version.

Edition	Version	Date	State	Change	Notice
XT-01	NR-XTECH-AI-GLOBAL-2026-BASELINE-v1.0.0	October 2026	Draft	Baseline Edition 2026, inaugural annual edition,	None

Notice types: erratum, addendum, supersession, withdrawal. A correction is a new version with a published notice; the earlier version stays resolvable, and no edition is ever edited in silence.

C.4Open record

This edition’s open record

Publication ID: NR-XTECH-AI-GLOBAL-2026-BASELINE-v1.0.0
Series alias: NR-XTC-XT-01-2026-BL
DOI: Concept and version DOI at deposit, on Zenodo with DataCite metadata
Versions: Each correction is a new version (isNewVersionOf); the concept DOI resolves to the latest
Bundle: RO-Crate 1.2 with CITATION.cff, CodeMeta and PROV-O provenance; a signed manifest
Licence: CC BY 4.0 (SPDX CC-BY-4.0) for original material; metadata CC0 1.0

Credit and identity

ORCID: Your iD in the deposit and the contribution record; DataCite can then add the work to your ORCID record
ROR: Your organisation by its ROR identifier, never by a typed name alone
CRediT: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualisation, Writing, review and editing
Account: None needed to contribute; never a password

Find and follow

Search the open indexes for Artificial Intelligence and Agentic Systems; software is cited by its SWHID from Software Heritage, and a quantified scenario run is preregistered on OSF before it is run. Annotation is your choice: Hypothesis opens this page in its own viewer.

Mirrors and harvest

The DOI is canonical. At release each signed bundle is also given an IPFS content identifier and a bloxberg anchor as mirrors; both are listed in the record once they exist. Zenodo exposes every record over OAI-PMH, and reviews and endorsements can arrive through COAR Notify where the repository supports it.

InvenioRDM metadata for a contribution on Artificial Intelligence and Agentic Systems, prefilled from this page

{
  "metadata": {
    "resource_type": {
      "id": "dataset"
    },
    "title": "Evidence on Artificial Intelligence and Agentic Systems for Nexus Reports",
    "publication_date": "YYYY-MM-DD",
    "creators": [
      {
        "person_or_org": {
          "type": "personal",
          "family_name": "Your family name",
          "given_name": "Your given name",
          "identifiers": [
            {
              "scheme": "orcid",
              "identifier": "0000-0000-0000-0000"
            }
          ]
        },
        "affiliations": [
          {
            "id": "ROR identifier of your organisation"
          }
        ]
      }
    ],
    "description": "What it is, where its record is kept and how it bears on Exponential Technology Report 2026: Artificial Intelligence and Agentic Systems.",
    "rights": [
      {
        "id": "cc-by-4.0"
      }
    ],
    "subjects": [
      {
        "subject": "Nexus Reports"
      },
      {
        "subject": "Artificial Intelligence and Agentic Systems"
      },
      {
        "subject": "Exponential Technology Report · XT-01"
      }
    ],
    "related_identifiers": [
      {
        "identifier": "https://globalriskalliance.com/nexus-reports/editions/xt-01/",
        "scheme": "url",
        "relation_type": {
          "id": "issupplementto"
        },
        "resource_type": {
          "id": "publication-report"
        }
      }
    ],
    "version": "1.0.0"
  },
  "custom_fields": {},
  "access": {
    "record": "public",
    "files": "public"
  },
  "files": {
    "enabled": true
  },
  "parent": {
    "communities": {
      "default": "nexusreports"
    }
  }
}

C.5Where this edition sits among the six pillars

Foundry seals the evidence and Universe tests it; Reports publishes it; Standards takes its corrections and Academy teaches from it.

F03 Sealed observations and indicator series Foundry Sealed observations and indicator series behind a profile (closeout).
F04 After-action and scenario findings for Universe After-action and scenario findings for chapters 10 and 13 (after action review).
F05 National Portfolio movement for chapter Foundry National Portfolio movement for chapter 11 (after action review).
Applied in editions Standards Nexus Standards: applied in editions.
XT-01 edition Reports AI and Agentic Systems 2026, published in every format with its sources, citation and corrections.
F06 Method and correction findings as change Standards Method and correction findings as change proposals (correction).
F12 Q4 synthesis informs the next-year Universe Q4 synthesis informs the next-year mandate (renewal).
Taught in the open Academy Risk Academy: taught in the open.

What Foundry and Universe hand to this edition, and what it hands to Standards, Universe and Academy. From the programme spine; XT-01 is a drafted Baseline Edition, so these handoffs run from its first operational quarter.

Handoff	Between	What crosses, and when
F03	Foundry to Reports	Sealed observations and indicator series behind a profile (closeout).
F04	Universe to Reports	After-action and scenario findings for chapters 10 and 13 (after action review).
F05	Foundry to Reports	National Portfolio movement for chapter 11 (after action review).
Ring	Nexus Standards to Reports	Nexus Standards: applied in editions.
F06	Reports to Nexus Standards	Method and correction findings as change proposals (correction).
F12	Reports to Universe	Q4 synthesis informs the next-year mandate (renewal).
Ring	Reports to Risk Academy	Risk Academy: taught in the open.

S.1Abstract

S.2Key messages

S.3Headline finding

S.4At a glance

S.5What each steward reads in it

Technical Truth

Public Meaning

Capital Meaning

Synthesis

S.6What this edition is, and is not

What this edition is

What it is not

What comes next

Notes

1.1The questions this part answers

Key questions

In this part

1.2The argument

1.3Twenty couplings between five systems

Transmission paths

1.4Reinforcing loops

Notes

2.1Headline finding

Implication

What would move it

2.2Energy

2.3Compute

2.4Minerals

2.5Adoption

2.6Labour

2.7Safety

2.8Governance

2.9Development

2.10Systemic

Notes

3.1Where each system stood at evidence close

Data-centre electricity use

Data centres’ share of US electricity

Growth in frontier training compute

Organisations using AI

Leading-edge logic capacity in Taiwan

Reported AI incidents

3.2Domain posture

3.3Sector screens

Notes

4.1Risk objects

Advanced-chip supply disruption

Grid and water constraints at data-centre clusters

Failures and misuse of agentic systems

Labour displacement outpacing transition

4.2Capability ladder

Notes

5.1Assumptions

5.2The cascade, step by step

5.3The scenario as a table of steps

Notes

6.1Three lenses

Capital Meaning

Technical Truth

Public Meaning

Synthesis

6.2Global Risks Alliance reading: 2027 AI readiness pack

Publication dispositions

6.3GCRI reading: the model card

6.4Global Risks Forum reading: editorial essays

EDITORIAL ESSAY · 01

EDITORIAL ESSAY · 02

EDITORIAL ESSAY · 03

EDITORIAL ESSAY · 04

6.5Nexus Standards reading: standards and regulatory watch

Changed since 2024

To watch in 2027

6.6Risk Academy reading: literature and evidence review, science and assessments

Review protocol

Where the evidence agrees

Where it is unsettled

Notes

7.1Questions for the operational series

7.2Portfolio proposed for the Q1 2027 alignment

7.3The pathway