天然气在推动人工智能驱动的能源需求中的作用

This report provides an independent, fact-based assessmenton the role of Gas in meeting incremental electricity demandfrom artificial intelligence (AI)-driven data centres. It reflectsthe International Gas Union’s (IGU) objective of providingtransparent information on global Gas developments togovernments, industry, and civil society. TABLE OF CONTENTS Table of FiguresAbbreviationsKey MessagesIntroductionFigure 1PART I– Demand OutlookUnderlying Data Centre Energy Demand And Compute GrowthFigure 2Robustness of the AI growth trajectoryPART II– Electricity Supply DynamicsFigure 3Figure 4Gas as a Critical Enabler of Electricity Security and Stability of SupplyFigure 5PART III– The Environmental And Societal CasePathways To Low-Carbon Gas SupplyFigure 6A Fragile BalancePART IV– Regional OutlookNorth AmericaEuropeAsia-Pacific (ex-China)ChinaMiddle EastAfricaCONCLUSION– Implications for Policy & Planning34568899111414151616171820212123 FIGURES FIGURE 1Estimated global data centre electricity demand,aggregate projections from leading industry sources2024-20307 FIGURE 2Estimated global data centre electricity demand byworkload, 2025-20308 FIGURE 3Global data centre capacity additions in IEA’s Base Caseand capacity at risk of connection delay due to gridconstraints, 2025-203010 FIGURE 4New U.S. interconnection requests and average lead time,2006–2023 FIGURE 5Selected announced data centres with on-site powergeneration (as of Q2 2025; compiled from press andcompany announcements; illustrative, not exhaustive) FIGURE 6Sources of global electricity generation for data centres(TWh), IEA Base Case, 2020-2035 FIGURE 7Global map of large data centre clusters in 2024 ABBREVIATIONS Australian Energy Market OperatorArtificial IntelligenceBillion Cubic MetersCombined-Cycle Gas TurbineCarbon Capture, Utilisation and StorageContent Delivery NetworkCommercial Operation DateCarbon DioxideDistribution System OperatorU.S. Energy Information AdministrationEngineering, Procurement and ConstructionEuropean UnionFederal Energy Regulatory CommissionGeneral Processing UnitGigawatt / Megawatt (units of power capacity)International Energy AgencyInternational Gas UnionInflation Reduction ActIntegrated Resource PlanISO/RTO - Independent System Operator / RegionalTransmission OrganisationKilowatt-hourLevelised Cost of ElectricityLiquefied Natural GasMinistry of Economy, Trade and Industry (Japan)National Energy Administration (China)North American Electric Reliability CorporationOriginal Equipment ManufacturerPower Purchase AgreementSmall Modular ReactorTransmission System OperatorTerawatt-hour / Gigawatt-hour (units of energy consumption)United Arab EmiratesUltra-High Voltage (transmission lines)Variable Renewable EnergyAEMOAIBCMCCGCCUSCDNCODCO2DSOEIAEPCEUFERCGPUGW / MWIEAIGUIRAIRPISO/RTOkWhLCOELNGMETINEANERCOEMPPASMRTSOTWh / GWhUAEUHVVRE KEY MESSAGES •Artificial Intelligence (AI) represents one of the greatestopportunities of the 21st century, and its transformativepotential hinges on scaling digital infrastructure at speed. •Electricity demand growth is surging as data centres arebecoming the new “industrial load” of the AI economy, andtheir electricity consumption is projected to double to 800-1000 TWh by 2030. •Renewables are expected to provide about half of thedata centres’ electricity consump-tion by 2030. However,their inherent variability creates a mismatch with the flat,24/7 load profile of data centres which requires not onlymore generation but, also, significantly more dispatchablecapacity. •Gas is well positioned to provide the bulk of additional,flexible, dispatchable supply, while contributing todecarbonisation goals. Studies indicate that Gas-firedgeneration for data centres could nearly double by 2035,equal to ~60 bcm of incremental supply. •Gas fulfils a critical, dual role: ensuring the stabilityand security of electricity to the data centres andsimultaneously providing a buffer to the electricity networkso that it can con-tinue to power critical national sectorssuch as healthcare, food processing, construction andmanufacturing. •A new class of energy customers is emerging with the scaleand urgency to transform critical infrastructure investments,creating opportunities for Gas suppliers to co-developreliable, lower-carbon solutions with digital leaders. •Data centres can be built and energised much more rapidlythan new generation and transmission projects which oftenface permitting, approval and construction constraints. Withspeed-to-power as the binding constraint, many developersare increasingly turning to proximate or behind-the-meterGas solutions. •Long-term energy planning must therefore remain fact-based, transparent, and grounded in realistic assumptionsabout electricity infrastructure and rapid expansion of AI-driven technology. INTRODUCTION Artificial Intelligence represents one of the greatest opportunities of the 21st century, withimmense potential for transformative change. Realising this p