Provenance -- report_intel_87210a61730c22b9_v2
REPORT report_intel_87210a61730c22b9_v2 (v2, CONDITIONALLY_READY) → INTELLIGENCE OBJECT AI_ENERGY_INFRA intel_87210a61730c22b9 (v6, CONDITIONALLY_READY)
Connectivity vocabulary: CONNECTED (fully resolved), PARTIALLY_CONNECTED (some but not all legs resolved), NOT_CONNECTED (no edge exists in the data), UNRESOLVED_REFERENCE (an id is referenced but not found in any readable store). Never fabricated.
Hypothesis → Claim → Evidence → Source chain
- H1: AI data centers are a significant driver of electricity demand growth in some regions (status: SUPPORTED) hyp_b0a4bbbc9b601728 PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://www.iea.org/reports/energy-and-ai/executive-summary) IEA_ENERGY_AND_AI_EXEC_SUMMARY → SOURCE https://www.iea.org/reports/energy-and-ai/executive-summary PARTIALLY_CONNECTED
- US data centers consumed approximately 176 TWh of electricity in 2023 (about 4.4% of total US electricity consumption), up from 58 TWh in 2014; LBNL projects 325-580 TWh by 2028 (6.7-12% of US electricity). STATUS: 2023 figure is OBSERVED, 2028 figure is FORECAST (kept as separate claim fields, not merged). AI_ATTRIBUTION=PARTIAL: LBNL states growth since 2017 is 'largely due to the growth in AI servers' but the published summary does not disaggregate AI training vs. inference vs. general cloud/non-AI compute -- attribution is explicit-but-aggregate, matching the existing IEA claim's AI_ATTRIBUTION=PARTIAL pattern, not an independent granular breakdown. (type: DESCRIPTIVE, status: OPEN) claim_o1_lbnl_us_dc_2023 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf) LBNL_DOE_2024_US_DC_ENERGY_REPORT → SOURCE https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf) LBNL_DOE_2024_US_DC_ENERGY_REPORT → SOURCE https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- H2: Data center growth matters but AI's independent contribution cannot currently be isolated from the data (status: SUPPORTED) hyp_o0_h2_ai_energy_infra PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://www.iea.org/reports/energy-and-ai/executive-summary) IEA_ENERGY_AND_AI_EXEC_SUMMARY → SOURCE https://www.iea.org/reports/energy-and-ai/executive-summary PARTIALLY_CONNECTED
- US data centers consumed approximately 176 TWh of electricity in 2023 (about 4.4% of total US electricity consumption), up from 58 TWh in 2014; LBNL projects 325-580 TWh by 2028 (6.7-12% of US electricity). STATUS: 2023 figure is OBSERVED, 2028 figure is FORECAST (kept as separate claim fields, not merged). AI_ATTRIBUTION=PARTIAL: LBNL states growth since 2017 is 'largely due to the growth in AI servers' but the published summary does not disaggregate AI training vs. inference vs. general cloud/non-AI compute -- attribution is explicit-but-aggregate, matching the existing IEA claim's AI_ATTRIBUTION=PARTIAL pattern, not an independent granular breakdown. (type: DESCRIPTIVE, status: OPEN) claim_o1_lbnl_us_dc_2023 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf) LBNL_DOE_2024_US_DC_ENERGY_REPORT → SOURCE https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf) LBNL_DOE_2024_US_DC_ENERGY_REPORT → SOURCE https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- H3: Recent demand growth results from multiple factors (manufacturing, electrification, EVs, population, heating/cooling) not primarily AI (status: SUPPORTED) hyp_o0_h3_ai_energy_infra PARTIALLY_CONNECTED
- Approximately 130 GW of capacity-eligible generation projects are stuck, unbuilt, in PJM's interconnection queue; GridLab estimates PJM customers could have saved approximately $3.5 billion in the 2026/27 delivery year had just 10% of queued projects been operational, and the 2026/2027 PJM capacity auction cleared at $16.1 billion, up 10% year over year. STATUS=OBSERVED for the queue size and auction price; ESTIMATED (counterfactual model) for the $3.5B savings figure. AI_ATTRIBUTION=PARTIAL: GridLab attributes PJM demand growth to 'data center additions and electrification' jointly, not AI specifically, and frames this explicitly as a generation/interconnection-queue supply shortage, not a transmission/distribution capacity bottleneck. (type: DESCRIPTIVE, status: OPEN) claim_5787125ab5f0110b PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://gridlab.org/interconnection-bottlenecks-cost-pjm-customers-3-5-billion/) GRIDLAB_PJM_INTERCONNECTION_3_5B → SOURCE https://gridlab.org/interconnection-bottlenecks-cost-pjm-customers-3-5-billion/ PARTIALLY_CONNECTED
- Grid Strategies LLC's national load-growth reports ("Strategic Industries Surging: Driving US Power Demand", 2024, and "Power Demand Forecasts Revised Up for Third Year Running, Led by Data Centers", 2025) identify 'strategic industries' -- semiconductor fabs, battery/EV manufacturing, hydrogen, and LNG (i.e. reshoring/industrial-policy-driven manufacturing) -- as a driver of rising utility load forecasts named independently of, and alongside, data centers. STATUS=OBSERVED at the level of 'a named, non-AI industrial driver exists in utility planning data'; NOT a quantified TWh split versus data-center/AI demand (report body itself was not fetched in the originating pass, only report-title/summary level). This describes a co-occurring structural demand driver, not a finding that it disproves or displaces any AI-attributed demand growth. AI_ATTRIBUTION=NOT_APPLICABLE (manufacturing/reshoring load growth is a non-AI driver by construction; this claim says nothing about AI's share of demand). (type: DESCRIPTIVE, status: OPEN) claim_a11c5cb4c7c4c2d1 NOT_CONNECTED
- no evidence resolved
- Per IEA's Global EV Outlook 2025, global EV charging consumed approximately 180 TWh of electricity in 2024 (about 0.7% of final global electricity consumption), a 60% increase over 2023. STATUS=OBSERVED for the 2024 figure. FORECAST (IEA Stated Policies Scenario): EV electricity demand could grow more than fourfold to approximately 780 TWh by 2030 (about 2.5% of global electricity demand; over 4% in Europe, 3.6% in China) -- kept as a separate forecast field, never merged with the 2024 observed figure. No US-specific TWh figure was given in the fetched section. This is real, independent, non-AI electricity-demand growth (EV charging), roughly half the 2024 observed scale of global data center consumption (180 TWh vs. 415 TWh), and is NOT shown to explain any AI-specific or data-center-specific grid bottleneck (e.g. PJM interconnection delays) -- it is evidence of a parallel/co-driver demand source, not evidence against AI's contribution. AI_ATTRIBUTION=NOT_APPLICABLE (EV charging demand is a non-AI driver by construction; this claim says nothing about AI's share of demand). (type: DESCRIPTIVE, status: OPEN) claim_f5842e6162c8fefc NOT_CONNECTED
- no evidence resolved
- Multiple independent EIA 'Today in Energy' notes (June heat wave increased electricity demand in the eastern and midwestern United States, https://www.eia.gov/todayinenergy/detail.php?id=62410 ; Texas power grid met record-breaking demand for electricity during recent heat wave, https://www.eia.gov/todayinEnergy/detail.php?id=57240) and an Ember analysis ("Powering through the heat: how 2024 heatwaves reshaped electricity demand", https://ember-energy.org/latest-insights/powering-through-the-heat-how-2024-heatwaves-reshaped-electricity-demand/) document that 2024/2025 heat waves drove record or near-record PEAK electricity demand in ERCOT (Texas) and the Eastern Interconnection via air-conditioning load. STATUS=OBSERVED. This is a real, non-AI, weather-driven demand channel operating on PEAK/HOURLY timescales, not the multi-year annual TWh growth trend that the IEA/LBNL AI_ENERGY_INFRA claims track -- it is a genuine alternative explanation for PEAK LOAD STRESS episodes specifically, and must never be merged with annual-baseline electricity-growth claims as if they measured the same thing. AI_ATTRIBUTION=NOT_APPLICABLE (weather-driven peak demand is a non-AI driver by construction; this claim says nothing about AI's share of demand). (type: DESCRIPTIVE, status: OPEN) claim_d641fb1790c3ef53 NOT_CONNECTED
- no evidence resolved
- Approximately 130 GW of capacity-eligible generation projects are stuck, unbuilt, in PJM's interconnection queue; GridLab estimates PJM customers could have saved approximately $3.5 billion in the 2026/27 delivery year had just 10% of queued projects been operational, and the 2026/2027 PJM capacity auction cleared at $16.1 billion, up 10% year over year. STATUS=OBSERVED for the queue size and auction price; ESTIMATED (counterfactual model) for the $3.5B savings figure. AI_ATTRIBUTION=PARTIAL: GridLab attributes PJM demand growth to 'data center additions and electrification' jointly, not AI specifically, and frames this explicitly as a generation/interconnection-queue supply shortage, not a transmission/distribution capacity bottleneck. (type: DESCRIPTIVE, status: OPEN) claim_5787125ab5f0110b PARTIALLY_CONNECTED
- H4: AI-related power impact concentrates in specific regions/markets rather than nationally (status: SUPPORTED) hyp_o0_h4_ai_energy_infra PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://www.iea.org/reports/energy-and-ai/executive-summary) IEA_ENERGY_AND_AI_EXEC_SUMMARY → SOURCE https://www.iea.org/reports/energy-and-ai/executive-summary PARTIALLY_CONNECTED
- Approximately 130 GW of capacity-eligible generation projects are stuck, unbuilt, in PJM's interconnection queue; GridLab estimates PJM customers could have saved approximately $3.5 billion in the 2026/27 delivery year had just 10% of queued projects been operational, and the 2026/2027 PJM capacity auction cleared at $16.1 billion, up 10% year over year. STATUS=OBSERVED for the queue size and auction price; ESTIMATED (counterfactual model) for the $3.5B savings figure. AI_ATTRIBUTION=PARTIAL: GridLab attributes PJM demand growth to 'data center additions and electrification' jointly, not AI specifically, and frames this explicitly as a generation/interconnection-queue supply shortage, not a transmission/distribution capacity bottleneck. (type: DESCRIPTIVE, status: OPEN) claim_5787125ab5f0110b PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://gridlab.org/interconnection-bottlenecks-cost-pjm-customers-3-5-billion/) GRIDLAB_PJM_INTERCONNECTION_3_5B → SOURCE https://gridlab.org/interconnection-bottlenecks-cost-pjm-customers-3-5-billion/ PARTIALLY_CONNECTED
- Per KEPCO data presented at a Korean National Assembly audit (국정감사, cited by Rep. Lee So-young / 이소영 의원), contracted data-center electricity capacity requested nationwide through 2029 totals 14.7 GW, of which 13.5 GW (91.8%) is concentrated in the Seoul capital region (Seoul 3,237 MW/51 facilities, Gyeonggi 8,789 MW/118 facilities, Incheon 1,494 MW/13 facilities). KEPCO's 9th transmission/transformation facility plan reportedly warns that meeting 50% of new data-center demand in the capital region will be difficult due to supply-capacity shortages. AI_ATTRIBUTION=UNKNOWN/INDIRECT: source describes 'data center' (IDC) electricity demand broadly, not an AI-specific workload breakdown. (type: DESCRIPTIVE, status: OPEN) claim_o1_korea_capital_region_dc_demand PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: Korea National Assembly audit material via Rep. 이소영, reported by energydaily.co.kr and kharn.kr (see intel/geographic_evidence/korea_evidence_search_result.json)) KOREA_CAPITAL_REGION_DC_GRID_CONCENTRATION → SOURCE Korea National Assembly audit material via Rep. 이소영, reported by energydaily.co.kr and kharn.kr (see intel/geographic_evidence/korea_evidence_search_result.json) PARTIALLY_CONNECTED
- Data centres consumed 22% of Ireland's metered electricity in 2024 (6,969 GWh), up from 6,335 GWh in 2023 (10% YoY increase) and from 5% of metered electricity in 2015. STATUS: OBSERVED (CSO official statistics), not AI-specific. AI_ATTRIBUTION=INDIRECT: a separate University College Cork academic study (Prof. Hannah Daly) argues AI-driven growth is contributing, but the CSO's own metered-electricity data does not disaggregate AI workload from general colocation/cloud/non-AI data center demand -- the 22% figure is all-data-center, not AI-isolated. (type: DESCRIPTIVE, status: OPEN) claim_3ab07077d739a9a4 NOT_CONNECTED
- no evidence resolved
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- H5: The core problem may be interconnection/transmission bottlenecks rather than generation shortage (status: WEAKENED) hyp_o0_h5_ai_energy_infra PARTIALLY_CONNECTED
- Approximately 130 GW of capacity-eligible generation projects are stuck, unbuilt, in PJM's interconnection queue; GridLab estimates PJM customers could have saved approximately $3.5 billion in the 2026/27 delivery year had just 10% of queued projects been operational, and the 2026/2027 PJM capacity auction cleared at $16.1 billion, up 10% year over year. STATUS=OBSERVED for the queue size and auction price; ESTIMATED (counterfactual model) for the $3.5B savings figure. AI_ATTRIBUTION=PARTIAL: GridLab attributes PJM demand growth to 'data center additions and electrification' jointly, not AI specifically, and frames this explicitly as a generation/interconnection-queue supply shortage, not a transmission/distribution capacity bottleneck. (type: DESCRIPTIVE, status: OPEN) claim_5787125ab5f0110b PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://gridlab.org/interconnection-bottlenecks-cost-pjm-customers-3-5-billion/) GRIDLAB_PJM_INTERCONNECTION_3_5B → SOURCE https://gridlab.org/interconnection-bottlenecks-cost-pjm-customers-3-5-billion/ PARTIALLY_CONNECTED
- Approximately 130 GW of capacity-eligible generation projects are stuck, unbuilt, in PJM's interconnection queue; GridLab estimates PJM customers could have saved approximately $3.5 billion in the 2026/27 delivery year had just 10% of queued projects been operational, and the 2026/2027 PJM capacity auction cleared at $16.1 billion, up 10% year over year. STATUS=OBSERVED for the queue size and auction price; ESTIMATED (counterfactual model) for the $3.5B savings figure. AI_ATTRIBUTION=PARTIAL: GridLab attributes PJM demand growth to 'data center additions and electrification' jointly, not AI specifically, and frames this explicitly as a generation/interconnection-queue supply shortage, not a transmission/distribution capacity bottleneck. (type: DESCRIPTIVE, status: OPEN) claim_5787125ab5f0110b PARTIALLY_CONNECTED
- H6: Efficiency/hardware improvements may offset part of AI workload growth (status: WEAKENED) hyp_o0_h6_ai_energy_infra PARTIALLY_CONNECTED
- Despite Google's hyperscale data centers reaching a power usage effectiveness (PUE) of approximately 1.1 (near the theoretical efficiency limit), Google's total energy consumption increased 3.7x from 2016 to 2022, per peer-reviewed modeling of general cloud computing (Meta/Google infrastructure). STATUS=OBSERVED (within the paper's own dataset). The authors interpret this as a Jevons-paradox dynamic: point-level efficiency gains do not by themselves reduce aggregate energy growth because demand elasticity and reinvestment expand system scale. AI_ATTRIBUTION=INDIRECT: this paper models general cloud computing, not AI-specific workloads, and is evidence against the 'efficiency alone will solve AI energy growth' framing rather than a direct AI electricity measurement. (type: TREND, status: OPEN) claim_d170f52f567053e9 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://arxiv.org/html/2411.11540v1) ARXIV_JEVONS_CLOUD_2411.11540 → SOURCE https://arxiv.org/html/2411.11540v1 PARTIALLY_CONNECTED
- Despite Google's hyperscale data centers reaching a power usage effectiveness (PUE) of approximately 1.1 (near the theoretical efficiency limit), Google's total energy consumption increased 3.7x from 2016 to 2022, per peer-reviewed modeling of general cloud computing (Meta/Google infrastructure). STATUS=OBSERVED (within the paper's own dataset). The authors interpret this as a Jevons-paradox dynamic: point-level efficiency gains do not by themselves reduce aggregate energy growth because demand elasticity and reinvestment expand system scale. AI_ATTRIBUTION=INDIRECT: this paper models general cloud computing, not AI-specific workloads, and is evidence against the 'efficiency alone will solve AI energy growth' framing rather than a direct AI electricity measurement. (type: TREND, status: OPEN) claim_d170f52f567053e9 PARTIALLY_CONNECTED
- H7: Some of the current discourse relies more on forecasts than actual observations (status: SUPPORTED) hyp_o0_h7_ai_energy_infra PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://www.iea.org/reports/energy-and-ai/executive-summary) IEA_ENERGY_AND_AI_EXEC_SUMMARY → SOURCE https://www.iea.org/reports/energy-and-ai/executive-summary PARTIALLY_CONNECTED
- US data centers consumed approximately 176 TWh of electricity in 2023 (about 4.4% of total US electricity consumption), up from 58 TWh in 2014; LBNL projects 325-580 TWh by 2028 (6.7-12% of US electricity). STATUS: 2023 figure is OBSERVED, 2028 figure is FORECAST (kept as separate claim fields, not merged). AI_ATTRIBUTION=PARTIAL: LBNL states growth since 2017 is 'largely due to the growth in AI servers' but the published summary does not disaggregate AI training vs. inference vs. general cloud/non-AI compute -- attribution is explicit-but-aggregate, matching the existing IEA claim's AI_ATTRIBUTION=PARTIAL pattern, not an independent granular breakdown. (type: DESCRIPTIVE, status: OPEN) claim_o1_lbnl_us_dc_2023 PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf) LBNL_DOE_2024_US_DC_ENERGY_REPORT → SOURCE https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf PARTIALLY_CONNECTED
- EXTERNAL_ACQUISITION_RECORD (temporal: UNKNOWN, attribution: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf) LBNL_DOE_2024_US_DC_ENERGY_REPORT → SOURCE https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf PARTIALLY_CONNECTED
- Global data centers consumed approximately 415 TWh of electricity in 2024 (about 1.5% of world electricity consumption), with a regional split of US 45%, China 25%, Europe 15%, and data center electricity consumption grew ~12%/year since 2017 (over 4x the rate of total electricity consumption growth). STATUS=OBSERVED. AI_ATTRIBUTION=PARTIAL: IEA names AI as 'the most important driver of this growth, alongside growing demand for other digital services' but does not isolate AI's specific numeric share versus crypto/general cloud workloads. (type: DESCRIPTIVE, status: OPEN) claim_afe7cc7b19317ee0 PARTIALLY_CONNECTED
Claim → Statistical Series → Source
| Statistical Series | Source | Connectivity |
|---|---|---|
| series_67a90dc183bd74a1 | IEA_ENERGY_AND_AI_EXEC_SUMMARY | CONNECTED |
| series_af37f1207e7ff042 | src_worldbank_api | CONNECTED |
| series_d018d1290acd5f20 | IEA_ENERGY_AND_AI_EXEC_SUMMARY | CONNECTED |