人工智能与电网相遇：塑造数据中心的权力格局

Table of contents We’d like to extendour sincere thanksto our panel ofindustry executivesfor sharing theirvaluable insightswith us. Shalabh JhalawadHead – Global Power and Energy Raj RahejaCFO at Camus Energy Saurabh ChauguleGlobal Head - Strategy & Business Cameron McInnesGlobal Head of Engineering atBarclays Development for Energy,Utilities & AI Solutions at AWS Pedro AlmeidaFernandesAn executive in the Iberian gasand power market Inês MuggliMarket Development Managerat SSE plc Jason JacksonProgramme Director at SiemensEnergy Ethan TiaoProduct Manager at Emerald AI We’d like to extendour sincere thanksto our panel ofindustry executivesfor sharing theirvaluable insightswith us. Bijit GhoshManaging Director/CIO and Headof Architecture and Engineeringat Wells Fargo Alyson FreemanDirector, Data Center Kevin MarquardtDirector of Energy Policy andRegulatory Affairs at CyrusOne Rigor Lois AlamarSenior Account Manager atShell Energy Sustainability at Dell Technologies Wilson LaiaHead of LATAM at theInternational Data CenterAuthority (IDCA) Matthieu GallegoOwner at The Blob Company Executive summary For energy providers, AI is emerging as a forcemultiplier for grid planning and reliability: at leastsix in ten power executives in our global surveyexpect >10% improvements in failure reduction,operational productivity, and outage preventionfrom advanced AI analytics. However, only 45% ofutilities today are using AI or machine learning forgrid optimization and reliability, revealing significantopportunity to scale digital and AI-driven operationsto keep pace with booming demand. CapturingAI’s benefits at scale across load forecasting,predictive maintenance, real-time grid control, andbeyond will be critical to powering the next phaseof digital growth. Ultimately, scaling AI will dependon advances in climate tech to deliver reliable,affordable, and sustainable power, while AI itselfwill play a central role in making energy systemsmore efficient and resilient. years. These pressures are playing out differentlyacross regions. In the US, while pockets of excessgeneration remain, data-center-driven loadgrowth is now outpacing available generation anddeliverable capacity.1In Europe, grid constraintshave emerged as the primary barrier in establisheddata-center hubs.2Meanwhile, in APAC, access toenergy itself is widely seen as the most immediateconstraint on further data-center expansion.3 AI turns every data center into a volatileindustrial load: 80% of utilities expect moreextreme and less predictable demand spikes,directly impacting grid resilience The rapid rise of digital infrastructure, especiallydata centers running AI workloads, is strainingpower systems worldwide. A large majority ofindustry leaders – 70% of electricity executivesand 83% of data-center executives – expect high-density AI-led data center sites to significantlyincrease regional power demand within the next3–5 years. Moreover, nearly 80% of electricityexecutives anticipate far greater volatility and load“peakiness” as AI usage grows. The challenge is not only the sheer scale of newdemand but its geographic concentration: massiveclusters of new data centers are straining localgrids. More than half of electricity executives citethis localized load clustering as a major obstacle toreliable service. As a result, electricity companies see a wideningsupply–demand gap: 77% fear that data-centerdemand will grow faster than their ability toexpand supply, and 68% even foresee electricityshortfalls in some regions within the next few Yet AI is not only driving demand; it is alsopart of the solution: More than 60% of utilitiesexpect AI to unlock significant efficiency andoperational gains Executive summary of electricity executives globally (and 79% in US)say they struggle to forecast demand accurately.Over forecasting due to speculative requests canmisallocate capital and create excess capacity: 60%of electricity executives cite the risk of strandedassets due to speculative requests. Grid modernization speed now determines timeto value: 78% of data center leaders point togrid infrastructure construction timelines asthe critical constraint •Thin reliability margins:84% say insufficientreserve margins have become a challenge aslarge always-on loads tighten available buffers;this worry is especially acute in Europe (raised by90% of European electricity executives)•Supply‑chain pressures:74% are facingrising costs and equipment delays for criticalinfrastructure (with European executivesreporting even higher incidence, 78%) Energy companies also recognize that meeting thisbooming demand will require unprecedented speed,but they face a perfect storm of infrastructureand execution challenges that limit their ability torespond quickly. Key constraints include: Data centers turn to on‑site power, reshapingthe utility relationship: 29% have alreadydeployed and 39% plan to add on‑site capacitywithin one to two years How to plan power demand