AIfor

infrastructureresilience

June2025

Tableofcontents

Executivesummary04

Foreword03

1.Introduction06

2.Investinginstability:Whyinfrastructureresiliencematters08

2.1.Growinginfrastructureexposure10

2.2.Risksimpactinginfrastructuresystems10

2.3.Incorporatingresilienceintoinfrastructure14

2.3.1.Plan:thepreventionphase14

2.3.2.Respond:detectionandreactionduringthehazardousevent15

2.3.3.Recover:aftertheincident15

3.LeveragingAIforinfrastructureresilience16

3.1.MeasuringtheeffectivenessofAIforinfrastructureresilience18

3.2.PotentialeconomicbenefitsofAI-poweredresilientinfrastructure19

3.3.AI-enabledinfrastructureresilienceinaction21

3.3.1.Reducingvulnerability:robustplanningandpreventivemeasures21

3.3.2.Mitigatinghazard:real-timedetectionandreactivemeasures24

3.3.3.Timelyoptimalrecovery25

4.UnlockingtheresiliencepotentialofAIforinfrastructure26

4.1.BarrierstotheimplementationofAI27

4.2.Awayforward29

Appendices32

Appendix1.Estimationoftheeconomicvalueofinfrastructure32

Appendix2.Assessmentoftheaveragedirectcostsofdifferenthazards32

Appendix3.CalculationoftheresilienceenabledbyimplementationofAI33

Authors35

Contacts36

DeloitteCenterforSustainableProgress37

Endnotes38

AIforinfrastructureresilience|Foreword

03

Foreword

Aroundtheglobe,infrastructuresystems

areundergrowingpressure—fromextreme

weathereventsandagingassetstothe

demandsoftheenergytransition,urbanization,andacceleratingtechnologicalchange.

Yetamidstthesechallengesliesasignificantopportunity:toenvisionandcreate

infrastructurethatismoreresilient,intelligent,andadaptable.

Artificialintelligence(AI)israpidlytransitioning

frombeingexperimentaltobeinganimportantpartofthesolution.Leadersarerecognizing

AInotjustasatechnicalinnovation,butone

ofthestrategictoolsthatcanbeusedtomake

infrastructuresystemsmoreresilient.Whetherthroughpredictivemaintenance,digitaltwins,orAI-enabledearlywarningsystems,AIishelpingpublicandprivatesectorleadersmakefaster,

smarterandmoreaccuratedecisions—andin

doingsoishelpingtomitigaterisks,reducecosts,lowerrecoverytimes,andmaintainvitalservicestosupportthrivingsocietiesandeconomies.

Examplesarealreadyemerging,liketheuseofdigitaltwinsincityplanningtosimulatefloodoccurrencesindifferentextremeweather

scenarios,demonstratingwhat’spossible

whenadvancedtechnologyisembeddedintoinfrastructurestrategy.

ThepotentialofAIisvast.Withtherightvisionandecosystemcollaboration,itcanhelp

leadersbuildinfrastructurethat’sstronger,

moreefficient,moresustainableandfuture-

ready.Progresscomeswheninfrastructure

stakeholders—includingpolicymakers,planners,operators,investors,technologyproviders,andinsurers—movebeyondexperimentationandpilotstohelpscaleAIadoptionwithconfidence.

Thetimingisright.Ecosystemsareevolving.

Solutionsarematuring.Thevaluepropositionisclear.AIcanbebothatoolforinnovationandastrategicenablerofresilience.

Exploretheinsights,drawinspirationfromtheexamples,andconsiderhowyourorganizationcantakethenextstepforward.

JenniferSteinmann

DeloitteGlobalSustainabilityBusinessleader

CostiPerricos

DeloitteGlobalGenAIBusinessleader

AIforinfrastructureresilience|Executivesummary

04

Executivesummary

Infrastructureisfundamentaltomodernsociety.Itcanshapehowwelive,workandmove,enablingtheflowofpeople,goods,andinformation.Fromenergyandwater,

tohealthcare,sanitation,andtransportation,infrastructurehelpsdeliveressential

servicesthatsupporthumanwell-beingandeconomicresilience.Wheninfrastructurethrives,societiescanflourish.

Toremaineffective,infrastructureshouldcontinually

evolve.Withacceleratingpopulationgrowthandeconomic

development,thecomingdecadeswilllikelydemandawaveofnewinfrastructure—systemsthataremoreexpansive,

intelligent,adaptive,andsustainable.Butasinfrastructure

systemsgrowinsizeandvalue,theyalsobecomeincreasinglyvulnerabletothechangingenvironmentsaroundthem.

Naturaldisastersaloneareprojectedtocauseapproximately

US$460billioninaverageannuallossestoinfrastructureglobally

by2050.1

Forcomparison,naturaldisastershaveresultedin

morethanUS$200billionofaverageannualdamagesgloballyoverthelast15years.

2

Naturalhazardsareexpectedtobecomemorefrequentandintenseinthefutureduetothechanging

climate,significantlyincreasingassociatedlosses.

3

Resilientinfrastructure—soitcanabsorbtheseshocks,bouncebackquickly,andadapt

4

—isimportantascontinuedeconomicandcivildemandsputhighways,powergrids,andwater

systemsundergreaterstress.

5

Makinginfrastructureresilientcanhelpprotectlivesandlivelihoods,keepcitiesrunning,andenableeconomicgrowthdespitepotential

risks.6

Thetransformativepowerofartificialintelligence(AI)hasthepotentialtosignificantlyenhanceinfrastructureresilience.

Infrastructureresilienceunfoldsacrossthreestages—planning(prevent),response(detectandreact),andrecovery—andAI

canofferpowerfultoolsateachstep.Intheplanningphase,

machinelearningcanhelpanalyzeriskdataandsimulate

scenariostoidentifymeasuresthatcanbetakenforpreventionandpreparednesstoimprovefloodresilien

ce7

orusingfire-

resistantmaterials.

8

Duringanevent,AI-drivenearly-warning

systemsandreal-timemonitoringhelpacceleratedetection,

9

andhelpguideemergency

responses.10

Intherecoveryphase,AIcanhelpacceleraterecoverybyprioritizingrepairsthroughpredictivedamageassessmentsandoptimizedresource

allocation.11

Byweavingdata-driveninsightsintoplanning,

response,andrecovery,AIcanstrengthentraditionalresiliencemeasures,reducevulnerabilities,andhelpinfrastructureadaptmoreeffectivelytoevolvingrisks.

Numerousreal-worldapplicationshelpdemonstratethe

effectivenessofAI-enhancedresiliencesolutions.Digitaltwins,forexample,cansimulateandstress-testinfrastructuredesigns,whichcanleadtomoredisaster-resilientassets.AI-powered

predictivemaintenancecanhelppreventtechnicalfailures

andensureoperationalcontinuity—forinstance,appliedtoanoffshorewindturbine,ithasthepotentialtoreducedowntimeby15%,andincreaseannualrevenuesbyupto6%,asoutlinedinthis

report.12

AIcanalsoplayanimportantroleinhazard

mitigation:systemsthatmonitorforestareasforearlysignsofsmokecandetectwildfiresintheirinfancy,enablingsuppressionbeforetheserisksescalate.

,

1314Forexample,adaptingCalifornia’searlywildfiredetectionsystemtoAustralia’sforestscould

mitigateanestimatedUS$100milliontoUS$300millionin

annualdamageswhilerequiringaone-timeinvestmentof

approximatelyUS$300million.

14

,15

Tosupportrecovery,AI

canacceleratepost-disasterdamageassessments,helpingto

restoreservicesandreduceeconomicdisruption.Forinstance,DeloitteConsultingLLP’sOptoAItoolforpost-disaster

inspectionscanmorethandoubleroofreconstructionspeedsbyhelpingtoidentifyrepairneedsafterextremeweatherevents.

12

AIforinfrastructureresilience|Executivesummary

05

ThefindingsinthisreportshowthatintegratingAI-poweredsolutionsforhazardmitigationandvulnerabilityreductionalonecouldyieldapproximatelyUS$70billiongloballyin

annualsavingsindirectdisastercostsby2050—equivalentto15%ofprojectedaveragelosses,complementingother

resilienceoptions.16

WithimprovedAIcapabilities,these

savingscouldexceedUS$110billionannually.

16

Despitethisenormouspotential,thepathtowidespread

implementationofAI-enabledresilienceininfrastructure

systemsischallenging.Obstaclesincludetechnological

limitations,financialconstraints,regulatoryuncertainty,and

institutionalinertia.High-quality,diversedatasetscontribute

toeffectiveAIperformance,yetdataavailabilityandaccuracy

remainoneofthe

majorconcerns.17

Upfrontinvestment

costs—oftenpairedwithuncertainshort-termreturns—can

furtherdeteradoption.

18

Ontheregulatoryandsecurity

front,asAI-specificframeworkscontinuetoevolve,coupled

withcybersecurityandprivacyconcerns,progresscanbe

slow,particularlyinregionswithlimiteddigitalinfrastructure,notablylow-incomecountries.Additionally,ashortageofskilledprofessionalsandorganizationalresistancetonewtechnologiesandwaysofworkingcanhindermomentum.

19

RealizingthepotentialofAItoenhanceinfrastructureresiliencecanrequirecoordinatedactionacrosstheecosystem—from

policymakersandinfrastructureoperatorstotechnology

companiesandthefinancialservicesandinsuranceindustries:

?Policymakersplayafoundationalrolebyhelpingtoshapetheenablingenvironmentforthewidespreadadoption

ofAI.Thiscanincludeplayingaroleinstandardsetting,

offeringeconomicsupportschemes,andmodernizinglegacyinfrastructure.Beyondregulationandeconomicsupport,

governmentscanalsohelpdrivecoordinationacrosstheinfrastructurevaluechain—facilitatingcross-sectorcollaborationandlong-termplanning.

?Infrastructureownersandoperators,manyofwhomarepublicagencies,shouldlooktoembedAIacrosstheplanning,design,andoperationalphasestohelpunlockefficiencygainsandenhanceresilience.Earlyinvestmentsinhigh-impactpilotprojectscangenerateproofpoints,economiesofscale,andacycleofcontinuouslearning.Modernizingsystemstobe

AI-ready,particularlythroughadaptableandexpandableITframeworksandinteroperabilitystandards,isimportant.

?Financialinstitutionsarekeyinovercomingthefunding

gapthatAIsolutionsoftenface.Throughinnovativefinancingtools—suchasresiliencebondsortargetedcreditlinesthat

includeAI—theycanhelpsupportlong-termprojectswith

delayedreturns.TheseinstitutionscanalsoapplyAIinternallytohelpenhanceriskassessmentandinvestmentprocesses,

includingcreditunderwritingandassetevaluations.Asco-

investorsinpublic-privatepartnerships,theycanhelpamplifytheimpactofresiliencestrategiesalongsidegovernments.

?Insurershavetheopportunitytoevolvealongside

infrastructuresystemsbyembeddingAIintotheirservices.ThisincludesdevelopingnewproductstailoredtoAI-enabledassets,offeringpremiumreductionsforsystemsthathelp

integratetrustedAIsolutions,andimprovingriskmodelsthroughadvancedanalytics.Indoingso,insurerscanhelpincentivizetheadoptionofAIforresiliencewhilebetter

managingtheirownexposuretorisksassociatedwithnaturalhazards.

?TechnologycompaniesaretheinnovationenginehelpingtopowerAIdevelopment.Theirroleextendsbeyondsoftware

andalgorithmstoincludeintegratedsolutionsthathelp

combineAIwithcomplementarytechnologiessuchasthe

InternetofThings(IoT)anddigitaltwins.Demonstratingthe

measurableimpactofthesesolutionsonresilienceoutcomesisimportant.Equallyimportantishelpingtoensurethatdigitalinnovationalignswithoperationalgoals,includingmanagingenergyconsumptionthroughalternativeenergysources.

?Architectureandengineeringfirmsplayakeyrolein

embeddingAItoolsintotheplanninganddesignphases

ofinfrastructuresystemsearlytohelpenhancetheir

resilience.Byintegratingtoolssuchasdigitaltwinsduring

planningandhelpingtoensurecompatibilitywithreal-time

monitoringsystemsandpredictiveanalytics,theycanhelp

createsmarter,moreresilientinfrastructure.Theirclose

collaborationwithtechnologyandserviceproviderscanhelpensurethatemerginginnovationstranslateintoscalable,real-worldsolutions.

Coordinatedanddecisiveactionacrossstakeholdersisimportanttohelpbuildinfrastructuresystems,thatarepreparedforthechallengesofachangingworld.

ByforginganecosystemthatisresilienttodisruptionandreinforcedwithAI

acrossthephasesofresilience—planning(prevent),response(detectandreact),

andrecovery—asafer,smarterandmoreresilientfutureawaits.

AIforinfrastructureresilience|1.Introduction

06

1.Introduction

Infrastructurecomprisestheassetsandnetworksthathelpdelivertheessentialservicessupportingmodernlife—fromwater,food,andenergytohealthcare,education,and

communications.

20

Theseassetsincludephysicalsystemssuchasenergygenerationanddistribution,roads,railways,bridges,ports,airports,watertreatmentandsupply,

andwastemanagement,aswellasthedigitalplatforms

thatcontrol,monitor,andoptimizetheiroperation.

21,22

Inmanyeconomies,infrastructureinvestmentmakesupasubstantialshareofGDP—forexample,oversixpercentinChinain2020—anditsvaluecontinuestogrow.

23

Recognizingtheroleofinfrastructuresystemsin

underpinningeconomicgrowthisimportant,especially

whenconfrontedwithdisasters.Sucheventscan

profoundlydisruptsystems,whichcanresultineconomicconsequences.Thecomplexinterconnectionsbetween

infrastructureandthebroadereconomyrevealhow

indirecteffects—suchassupplychaininterruptions,

servicedisruptions,andcommunitydisplacements—candecelerateeconomicactivity.Furthermore,thelong-termimpactsonproductivity,accesstoeducation,andhealthemphasizethenecessityforresilientinfrastructuretohelpalleviatethesechallenges.

Infrastructuresystemsaresubjecttodisasterrisksthatcanentailbothphysicaldamagecostsandservicedisruptions.Risktoinfrastructureemergesfromtheinteractionof

threedimensions

22

—hazard,exposure,andvulnerability(

Figure1

)—whichtogetherhelpdeterminetheriskof

damagewhenadisruptiveeventoccurs.Hazardsarethepotentiallydamagingphysicaleventsthemselves—storms,floods,heatwaves,orearthquakes—whosefrequency

andintensityareincreasing.Exposurereferstothe

presenceandvalueofassetswithinahazardzone,from

powerstationsandpipelinestodigitalcontrolnetworks.Exposurecanincreaseassocietiesinvestmoreheavilyininfrastructure.Vulnerabilitydescribeshowsusceptible

thoseassetscanbetoharm—drivenbyfactorslike

designstandards,materialtypes,maintenanceregimes

andsysteminterdependencies.Byanalyzinghowagivenhazardinteractswithexposed,vulnerableinfrastructure,decision-makerscanhelpquantifyriskandprioritize

investmentsthatcanreduceexposure(forexample,by

relocatingassets),strengthendesignandmaintenancetolowervulnerability,andbuildadaptivecapacity—helpingtoensurethatnewandexistingsystemsremainresilientinthefaceofevolvingrisks.

Engineersandplannerscanembedresiliencein

infrastructurebydesigningandmanagingsystemsto

helpwithstandshocks—absorbingimpacts,responding

swiftlyduringanevent,andadaptingafterwardtohelp

restoreservicewithminimaldisruption.Thiscanbring

significanteconomicbenefits.Thebenefit-to-costratio

(BCR)estimatesforinvestmentsinresilienceexceedthree,andinsomecasescanevenreachashighas50.

24

This

meansforUS$1investedinaresiliencesolution,US$3

toUS$50worthofdamagesandlossescanbeavoided.

AccordingtotheNationalInstituteofBuildingSciences,

eachdollarinvestedinresiliencesavesbetweenUS$4andUS$11indisasterresponseandrecoverycosts.

25

Locatinginfrastructureinplaceslesslikelytoexperiencehazards,andreducingitsvulnerabilitytohazardsthroughbetterdesignorbuildingredundantsystems,canhelpdevelopinfrastructureresilience.

AIforinfrastructureresilience|1.Introduction

07

Figure1.Understandinginfrastructurerisksfordisastermanagement

Hazards

Thepotentialoccurrenceofaneventthatcancausedamage.Itischaracterizedbyitsprobability(howlikelyitistohappen)andintensity(howstrongorsevereitis),asinthecaseof

earthquakes,tropicalcyclones,etc.

Exposure

Referstothepresenceandvalueofpeople,infrastructureoreconomicactivitiesinareasthatcouldbea?ectedbya

Risk

components

hazard.Itquanti?eswhatisatriskwhenahazardoccurs,includingthephysicallocationandcharacteristics.

VuInerabiIity

Referstothetendencyofexposedpeople,assets,orsystemstosu?erharmorlosswhena?ectedbyahazard.

Source:DeloitteGlobalbasedontheassessmentscarriedbyCDRI,

22

UnitedNations,

26

andIPCC.

27

Initsbroaderdefinitionasabranchofcomputersciencethat

enablesmachinestoperformtasksrequiringhumanintelligence,

28

AIistransformingoursocieties.Itisalsorevolutionizingindustriessuchashealthcare,transportation,manufacturing,andretail,byoptimizingsupplychainsandprovidingpredictivediagnostics,

real-timedecision-making,personalizedrecommendations,anddifferenttypesofautomation.

29

Beyondthesetransformations,AIispositionedtohelpstrengtheninfrastructureresilience:itcanhelppredictequipmentdegradationandschedulemaintenancebeforefailuresoccur,

30

usehigh-resolutionweatherandsensordatatoforecastfloodsorheatwavesdaysinadvance,

31

and

deploycomputer-visiondronestoinspectbridgesandpipelinesafteranevent.

32

Bylayeringthesecapabilitiesatoptraditionalresiliencemeasures—suchashazard-basedland-useplanning,robustengineeringstandards,andemergencyresponse

drills

33

—organizationscangainearlierwarning,morepreciseriskassessments,andautomateddecisionsupportthattogetherhelpreducedowntime,limitdamage,andacceleraterecovery.

WhileAIhasdemonstrateditsvalueinoptimizingoperations,

34

andstrengtheningindustrialsystems,

29

thereisstillalackoffocused,

conciseassessmentofitsroleininfrastructureresilience—

especiallyasassetexposureandfrequencyandintensityof

extremeweathereventsgrow.Thisreportaimstohelpfillthisgapbyfirst,identifyingtherisksthreateningtheinfrastructureand

potentialdamages,andsecond,assessingthekeyapplicationsofAItohelpenhancetheresilienceofinfrastructureandtheresultanteconomicbenefits.

Usingadata-driven,model-basedapproach,thisanalysisestimatesboththecurrentandfuturevalueofinfrastructuresystems,as

wellastheaveragelossescausedbymajornaturaldisasters.

Usingexamplesandcasestudiesbasedonempiricalfindings

andmodeledapplications,AI’sresilienceimprovementpotentialisassessedandcalculated.Thefindingsarethencompiledandinterpretedfromadecision-makerlens,toidentifynotonlytheanswertothequestion“what”,butalsoto“how”toharnessAItoenhanceinfrastructureresilience.

AIforinfrastructureresilience|2.Investinginstability:Whyinfrastructureresiliencematters

08

2.Investinginstability:Whyinfrastructureresilience

matters

AIforinfrastructureresilience|2.Investinginstability:Whyinfrastructureresiliencematters

09

Infrastructureservesasabackboneofcommunitiesandsociety.DeloitteGlobal’sanalysisshowstheeconomicvalueofinfrastructurecouldreachUS$390trillionby2050,an85%increasecomparedto2022,whiletheannualaveragelossestoinfrastructurecausedbynaturalhazardscouldmorethandoubleby2050,reachingapproximatelyUS$460billion.Resiliencecancreateimpactineachofits

implementationphasesagainstahazard—planningandprevention,detectionandresponseasit

happens,andrecoveryafterwards—becomingincreasinglyimportanttohelpminimizethese

losses.35

Infrastructurereferstothefundamentalfacilitiesandsystems

servingacountry,city,orotherarea,includingtheservicesand

facilitiesnecessaryforitseconomytofunction.

36

Infrastructurecanbedividedintotwomaincategories:physicalinfrastructureand

digitalinfrastructure.

Physicalinfrastructureencompassesthebuiltenvironment,servingcommunities:transportationinfrastructure,utilities,telecommunications,andsocial(administrativeandpublic)

servicebuildings,includingschools,hospitals,socialhousing,andpublicsafetyinfrastructure.

37

Digitalinfrastructurereferstothedigitalsideofphysicalandsocialinfrastructure,supportingandenhancingitsfunctionality.Itincludesinformationtechnologies,cloudplatforms,software,etc.

38

(

Figure2

).

Physicalanddigitalinfrastructuresystemsareamongsomeofthekeyinvestmentsthathelpsupporteconomicgrowthandsocial

development.

6

Figure2.Physicalanddigitalinfrastructuresystems

Transportation:

roads,railways,portsandairports

Telecommunications

Socialandadministrative

PhysicaI

infrastructure

Energysystems:

power,oilandgas

services:hospitalsand

educativebuildings,publicsafetyandadministrativebuildings

Watersystems:

wastewaterandsewage

Natureandagriculture:

forests,agroforestrysystems,irrigationsystems

Informationtechnologies

Software

Cloudplatforms

DigitaIinfrastructure

Source:DeloitteGlobalanalysisbasedonTheWorldBank

37

,

38

,39

AIforinfrastructureresilience|2.Investinginstability:Whyinfrastructureresiliencematters

10

2.1.Growinginfrastructureexposure

Infrastructuredevelopmentisanimportantpillarfortheglobal

economy.Everyyearcountriesspendbetween0.2%and6%of

theirGDPintransportationinfrastructuredevelopmentalone,

representingmorethanUS$200billionofannualinvestments.

,

4041Infrastructureinvestmentsareexpectedtoreachtrillionsof

dollarsinthecomingdecadestohelpsupportfutureeconomic

developmentandpopulationgrowth.

42

Thetotalestimated

infrastructurevaluefor2050isexpectedtogrowbyapproximately85%,frommorethanUS$200trillionin2022toapproximately

US$390trillionin2050,drivenbytheseinvestments(

Figure3

).

2.2.Risksimpacting

infrastructuresystems

Infrastructuresystemscanbesubjecttoawiderangeofrisksthatcanbecausedbydifferenttypesofhazardsandincidents,includingnaturaldisasters,technicalfailures,cyberthreats,

andsocialinstability(

Figure4

).Acutenaturalshockssuchas

earthquakes,floods,andhurricanescancausesudden,severe,

andextensivedamagetoinfrastructure.

43

Chronicstresses

amplifythefrequencyandseverityofextremeevents,intensifyingnaturalhazards.

44

Concernsrelatedtothehealthandstateof

physicalassets,suchascorrosion,agingcomponents,ormaterialdegradation,cangraduallyundermineperformance,causing

technicalincidentsandfailures.Asinfrastructuresystemsbecomeincreasinglydigital,intelligent,anddata-heavy,cyberattacks

representagrowingrisk,withthepotentialtodisruptoperationsandcompromisesafety.

45

Finally,war/conflicts,geopolitical

tensions,andsocialmovementscanalsoimpactinfrastructure

systemsandcausedamages.Forinstance,thelatestreportoftheRapidDamageandNeedsAssessment(RDNA4)commissionedbytheUkrainianGovernment,theWorldBankGroup,theEuropeanCommission,andtheUNfoundthatthattheRussia-Ukraine

warcausedmorethanUS$520billionofdamagesbytheendofDecember2024,primarilyinhousing,transport,andenergyinfrastructure.

46

Asdifferenttypesofinfrastructureandeconomicsectorsbecomemoreinterconnected,thepotentialimpactofrisksacrossmultipledomainsbecomesmoresevere.

47

Adisruptioninonesectorcanquicklycascadeintoothers,suchaspoweroutagesaffecting

communications,orwatersupplyinterruptionshinderingenergyproduction.