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

Contents
  1. ·Edition home
  2. ·Executive summary
  3. 1AI is infrastructure, not only software
  4. 2Key findings
  5. 3Systems and sectors at baseline
  6. 4Risks and capability
  7. 5Chip supply disruption under grid constraint
  8. 6Lenses and readings
  9. 7The 2027 agenda
  10. 8Systems in depth
    1. 8.1 Compute and data centres
    2. 8.2 Energy and water
    3. 8.3 Chips and minerals
    4. 8.4 Economy and workforce
    5. 8.5 Safety, security and failure modes
  11. AMethodology
  12. BData appendix
  13. CReferences and record
  14. ·Complete edition, for print

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
ComponentStateEvidenceStatus
Training computefour to five times a year since about 201014Reported
Inferencedemand rising as use spreads; poorly measured1Inferred
Costcost of fixed-level use collapsing2Reported
Concentrationcompute concentrated in few providers and regions1Inferred

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
ComponentStateEvidenceStatus
Electricityabout 1.5% of world use1Reported
Local gridsconnection queues at clusters15Reported
Watercooling demand at water-stressed sites2537Inferred
Emissionsdepend on the power mix of each grid122Reported

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
ComponentStateEvidenceStatus
Fabricationleading-edge capacity in one region17Reported
Mineralsrefining concentrated in a few countries18Reported
Trade controlschips, tools and minerals since 20221929Reported
New capacityfabs under construction elsewhere17Inferred

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
ComponentStateEvidenceStatus
Adoption78% of organisations in 20242Reported
Exposurea quarter to two-fifths of jobs1632Reported
Job churn170 million created and 92 million displaced by 203020Reported
Measured effectslimited 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
ComponentStateEvidenceStatus
Incidentsrecord 233 reported in 2024227Reported
Risk taxonomymalicious use, malfunction, systemic risks3Reported
CyberAI changes attack and defence21Inferred
Agentic failureslittle systematic evidence,Expert input

Couplings

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

  1. 14J. Sevilla and E. Roldán, “Training compute of frontier AI models grows by 4–5x per year.” Epoch AI, 2024.back
  2. 1International Energy Agency (ed.), Energy and AI. World Energy Outlook Special Report. IEA, Paris, 2025.back
  3. 2Stanford Institute for Human-Centered Artificial Intelligence (ed.), Artificial Intelligence Index Report 2025. Stanford HAI, Stanford, 2025.back
  4. 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
  5. 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
  6. 37S. Valadbaygi (ed.), Global Water Risk & Resilience Report 2026, Baseline Edition. The Global Centre for Risk and Innovation (GCRI) for the Nexus Consortium, Toronto, 2026. NR-SECTOR-WATER-GLOBAL-2026-BASELINE-v1.0.0.back
  7. 22Ember (ed.), Global Electricity Review 2025. Ember, London, 2025.back
  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.back
  9. 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
  10. 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
  11. 18International Energy Agency (ed.), Global Critical Minerals Outlook 2025. IEA, Paris, 2025.back
  12. 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
  13. 32International Labour Organization (ed.), Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, Geneva, 2025.back
  14. 20World Economic Forum (ed.), Future of Jobs Report 2025. WEF, Cologny/Geneva, January 2025.back
  15. 3Y. Bengio (ed.), International AI Safety Report. Department for Science, Innovation and Technology, London, January 2025.back
  16. 27Responsible AI Collaborative (ed.), AI Incident Database. incidentdatabase.ai, retrieved 2026-09-30.back
  17. 21World Economic Forum (ed.), Global Cybersecurity Outlook 2025. WEF, Cologny/Geneva, January 2025.back

Cite this chapter

Valadbaygi, S. (Ed.). (2026). Chapter 8: Systems in depth. In 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).

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

ReportsContribute

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

Reports is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.

Global Risks AllianceExponential Technology

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

The Global Risks Alliance

Finance · trust · precision

Not for profit

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

Support our work

Who funds us

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

Independence

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

Corrections policy

Licence

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