1. McKinsey presentation v2.1中國(guó)第八屆云計(jì)算大會(huì)

SmarteningupwithArtificialIntelligence(AI)Presentation|October2017CONFIDENTIALANDPROPRIETARYAnyuseofthismaterialwithoutspecificpermissionofMcKinsey&CompanyisstrictlyprohibitedMcKinsey&Company2ContentsAIAIisreadytoscalePlayersintheindustrialsectorshouldconsidereightusecasesofAItoachievethenextlevelofperformanceWheredoesChinastandFivepragmaticrecommendationsenableAI-basedperformanceimprovementsMcKinsey&Company3SOURCE:BasedonKurzweil’s"TheLawofAcceleratingReturns"(2001);McKinsey"McKinsey&Company3Exponentiallyfaster,smaller,cheaper,andbetterComputationalpowerComputationalpowerequivalentto…second/ Allhuman 119201940196019802000202020402060McKinsey&Company4+n.GPUproducersandotherhardwareplayersheavilyinvestinAI-specificandready-to-usecomputingdevices/solutions.InvestorsarehypedupforAIapplicationsAIadoptionisreachingatippingpointgivenaconvergenceoffavourableconditionsGrowthDriversCoreCorecomputingtechnologyProgrammingplatform.Masscollaborationsonopensourceplatformslargelypropeldeeplearningandothertechniques..atasetcollectionExplosiveamountofmachine/human-generated,unstructureddataavailableApplicationApplicationanduseMcKinsey&Company5SOURCE:CapitalIQ;Pitchbook;Dealogic;S&P;McKinseyMcKinsey&Company5ternalcorternalcorporateinvestment1TechnologygiantsdominateinvestmentsinAIInvestmentinAI,20161,$billionInvestmentbytechgiantsandothercorporationsAIshareoftotalinvestmentAIshareoftotalinvestment5502–3OtherArivateequityVenturecapitalM&A22–3VC,PE,andotherexternalfunding36–940351Estimateof2016spendbycorporationstodevelopanddeployAI-basedproducts.Calculatedfortop35hightechandadvancedmanufacturingcompaniesinvestinginAI.EstimateisbasedontheratioofAIspendtototalrevenuecalculatedforasubsetofthe35companies.2VCvalueisanestimateofVCinvestmentincompaniesprimarilyfocusedonAI.PEvalueisanestimateofPEinvestmentinAI-relatedcompanies.M&AvalueisanestimateofAIdealsdonebycorporations.“Other”referstograntsandseedfundinvestments.Includesonlydiscloseddataavailableindatabases,andassumesthatallregistereddealswerecompletedwithintheyearoftransaction.Compoundannualgrowthratevaluesrounded.3M&AandPEdealsexpressedbyvolume;VCdealsexpressedbyvalue.McKinsey&Company6SOURCE:CapitalIQ;Pitchbook;Dealogic;McKinseyMcKinsey&Company6NaturallanguageNaturallanguage0.6–0.9MachinelearningMultiuseandnonspecificapplications5.0–7.0Autonomousvehicles0.3–0.5Virtualagents0.1–0.2Machinelearningreceivedthemostinvestment,althoughboundariesbetweentechnologiesarenotclear-cutExternalinvestmentinAI-focusedcompaniesbytechnologycategory,20161$billionAComputerComputervision2.5–3.5SSmartrobotics0.3–0.51EstimatesconsistofannualVCinvestmentinAI-focusedcompanies,PEinvestmentinAI-relatedcompanies,andM&Abycorporations.Includesonlydiscloseddataavailableindatabases,andassumesthatallregistereddealswerecompletedwithintheyearoftransaction.McKinsey&Company7ContentsAIAIisreadytoscalePlayersintheindustrialsectorshouldconsidereightusecasesofAItoachievethenextlevelofperformanceWheredoesChinastandFivepragmaticrecommendationsenableAI-basedperformanceimprovementserallAIindexGIDigitization1technologiesAIspendSupportingdigitalerallAIindexGIDigitization1technologiesAIspendSupportingdigitalassetsdevelopmentOperationsSupplychaindistributionCustomerexperienceandgeneralmanagementWorkforcemanagementAIinworkforceAIresourcesperworkerAIadoptionisoccurringfasterinmoredigitizedsectorsandacrossthevaluechainAIIndexRelativelylowRelativelyhighAssetsAssetsHightechandtelecommunicationsAutomotiveandassemblyinancialservicesesourcesandutilitiesMediaandentertainmentConsumerpackagedgoodsTransportationandlogisticsProfessionalservicesreBuildingmaterialsandconstructionTravelandtourism1TheMGIDigitizationIndexisGDPweightedaverageofEuropeandUnitedStates.SeeAppendixforfulllistofmetricsandexplanationofmethodology.SOURCE:McKinseyGlobalInstituteAIadoptionandusesurvey;DigitalEurope:Pushingthefrontier,capturingthebenefits,McKinseyGlobalInstitute,June2016;DigitalAmerica:Ataleofthehavesandhave-mores,McKinseyGlobalInstitute,December2015;McKinseyGlobalInstituteanalysisMcKinsey&Company8McKinsey&Company91BasedonthemidpointoftherangeselectedbyMcKinsey&Company92Resultsareweightedbyfirmsize.SeeAppendixforanexplanationoftheweightingmethodology.SOURCE:McKinseyGlobalInstituteAIadoptionandusesurvey;McKinseyGlobalInstituteanalysisSectorsleadinginAIadoptiontodayalsointendtogrowtheirinvestmentthemostFutureAIdemandtrajectory1Averageestimated%changeinAIspending,next3years,weightedbyfirmsize29876543210FinancialservicesHightechandtelecommuni-cationsTransportationandlogisticsTravelandtourism ProfessionalservicesHealthcare EnergyandresourcesMediaandeMediaandentertainmentandassemblyConsumerpackagedgoodsFallingbehindEducationConstructionLeadingsectors02468101214161820222426283032CurrentAIadoption%offirmsadoptingoneormoreAItechnologyatscaleorinacorepartoftheirbusiness,weightedbyfirmsize2HightechandtelecommunicationsAutomotiveandassemblyHightechandtelecommunicationsAutomotiveandassemblyFinancialservicesUtilitiesandresourcesTransportationandlogisticsConsumerpackagedgoodsEducationHealthcareBuildingmaterialsandconstructionProfessionalservicesTravelandtourismAIadopterswithaproactivestrategyhavesignificantlyhigherprofitmarginsAIadopterswithproactivestrategy1PartialAIadoptersorexperimentersSelf-reportedcurrentprofitmargin2Differencefromindustryaverage(unweighted)(percentagepoints)-10-5051015Non-adopters201FirmsthatarebigdataandcloudservicesusersandreporttheirstrategicposturetowardAItobe:“DisruptingourindustryusingAItechnologyisatthecoreofourstrategy,”“Wehavechangedourlonger-termcorporatestrategytoaddresstheAIthreatoropportunitydisruption,”or“WehavedevelopedacoordinatedplantorespondtotheAIthreatoropportunitybuthavenotchangedourlonger-termcorporatestrategy.”2Operatingprofitmargin2Operatingprofitmarginforselectedsectorsasashareofturnover,forcontinuingoperationsandbeforeexceptionalitems.SOURCE:McKinseyGlobalInstituteAIadoptionandusesurvey;McKinseyGlobalInstituteanalysisMcKinsey&Company11ThetechnicalpotentialforautomationvariesdependingonthemixofactivitytypesOverviewofkeysectors(Germanyexample)Abilitytoautomate(%)0000SizeofbubbleindicatesshareoftimespentinGermanoccupationsUnpredict-Predict-SectorsbyactivityInter-ableCollectProcessableAutomationpotentialphysical5typeManage1Expertise2face3physical4datadataofactivitiesdefined,physical57264A7264andfoodservicesTransportationandwarehousing6056555151A6056555151RetailtradeManufacturingInformationWholesaletrade1Managinganddevelopingpeople2Applyingexpertisetodecisionmaking,planning,andcreativetasks3interactingwithstakeholders4Performingphysicalactivitiesandoperatingmachineryinunpredictableenvironments5PerformingphysicalactivitiesandoperatingmachineryinpredictableenvironmentsMcKinsey&Company12TheimpactofAIusecasesvariesbyindustryImpactHighMediumLowIndustry(Germanyexample)ProductsProductsandservicesManufacturingoperationsBusinessprocessesAutomotiveOEMs/commer-AutomotiveIndustrialSemi-UsecaseAerospacecialvehiclessuppliersequipmentconductorsAutonomousvehiclesAI-enhancedmaintenancepredictiveCollaborativeandcontext-awarerobotsYieldenhancementinmanufacturingAutomatedqualitytestingAI-enhancedsupplychainmanagementHighperformanceR&DprojectsBusinesssupportfunctionautomationWhilefewoccupationsarefullyautomatable,60percentofalloccupationshaveatleast30percenttechnicallyautomatableactivitiesAutomationpotentialbasedondemonstratedtechnologyofoccupationtitlesintheUnitedStates(cumulative)1ExampleoccupationsTechnicalTechnicalautomationpotential(%)Shareofroles(%)100%=820rolesSewingmachineoperators,gradersandsortersofagriculturalproducts1<5%ofoccupationsconsistof1activitiesthatare100%automatable8263442Inabout60%ofoccupations,least30%ofactivitiesarea4251627391100Stockclerks,travelagents,watchrepairersPsychiatrists,Psychiatrists,legislatorsFashiondesigners,chiefexecutives,statisticiansChemicaltechnicians,nursingassistants,Webdevelopers1Wedefineautomationpotentialaccordingtotheworkactivitiesthatcanbeautomatedbyadaptingcurrentlydemonstratedtechnology.SOURCE:USBureauofLaborStatistics;Afuturethatworks:Automation,employmentandproductivity,McKinseyGlobalInstitute,January2017;McKinseyGlobalInstituteanalysisMcKinsey&Company13McKinsey&Company14ContentsAIAIisreadytoscalePlayersintheindustrialsectorshouldconsidereightusecasesofAItoachievethenextlevelofperformanceWheredoesChinastandFivepragmaticrecommendationsenableAI-basedperformanceimprovementsMcKinsey&Company15SOURCE:CapitalIQ;Pitchbook;Dealogic;S&P;McKinseyMcKinsey&Company15TheUnitedStatesandChinadominatetheAIlandscape,withEuropefallingbehind(1/2)ThemostvibrantAIhubs…SiliconValley.Topglobalhubforstartups–12,700–15,600activestartups–2milliontechworkers.GloballeaderforVCinvestment.Headquartersofmanytophigh-techcompaniesNewYork.LeadinghubforFinancialsandmedia.AItalentpipelinefromuniversitiessuchasCornell.Strongfundingecosystem—2ndintheworldfortheabsolutenumberofearly-stageinvestmentsjing.LeadinginvolumeofacademicresearchoutputinAIcoming.Extensiveinvolvementoftechleaders.AIisgovernmentpriorityn.Longn.Longhistoryofcooperationbetweenscienceandindustry.World-classuniversitiessuchasMITdevelopingadvancedtechnologiesandprovidingatalentpipeline.Globalfinancecenter,supportinginvestmentandfin-techapplications.EuropeanleaderofVCstartupinvestment.Presenceoftophigh-techcompanies.TalentpipelineandresearchexpertisefromuniversitiessuchasUniversityofCambridge,ImperialCollege,andOxford.Hubforelectronicsmanufacturingfirms.Strongexpertiseinhardware.AIisgovernmentpriorityMcKinsey&Company16SOURCE:CapitalIQ;Pitchbook;Dealogic;S&P;McKinseyMcKinsey&Company16TheUnitedStatesandChinadominatetheAIlandscape,withEuropefallingbehind(2/2)…andtheexternalinvestmentbehindtheirgrowth1$billion(estimate)a~0.1~0.12.5–4.0~0.15.0–8.01EstimatesconsistofannualVCinvestmentinAI-focusedcompanies,PEinvestmentinAI-relatedcompanies,andM&Adonebycorporations.Includesonlydiscloseddataavailableindatabases,andassumesthatallregistereddealswerecompletedwithintheyearthetransactionswereannounced.McKinsey&Company171McKinsey&Company17SOURCE:Expertinterviews;McKinseyChinademonstratesstrengthsinAIapplicationanddatasets,butmainlylagsincomputingtechnologyChina’spositionvsUSAGrowthDriversNopresenceNotablepresenceLeadingpositionDescriptionmputingtechnologyormlgorithmectiontionsecases.SignificantlylaggingingeneralHPC1rAI.PotentialinchipsdedicatedtoAIsystemabilities.Inalgorithmadvancesandaccuracy,atparwithworld’sleadingcountryU.S..Lagginginopensourceplatformdevelopmentandcontribution.Leadingindatasetcollection(amountandrichness).Notinfluentialindatasharing.Limitedgovernmentdatasharing.StrongpresenceinAIapplicationdevelopment.Moreinnovationneeded.LimitedroleforuniversitiesSOURCE:McKinsey–smartmovesrequired2017;McKinsey"Futureofmobility"consumersurvey2017McKinsey&Company18Alsofromaconsumerpointofview,Chinaiswellpositioned–AutonomousdrivingexampleNegativePositiveImportanceandhighdemandofusingfullyADChinaGermanyChinaGermany21USA23 2.China2.China 3.Cities Big862Medium1836Small/2528ownSOURCE:McKinseyChinaAISurvey2016McKinsey&Company19tantialtantialLimitedNone78%expertsinChinabelieveAIwillhave‘substantial’to‘disruptive’impactinthenext5yearsTowhatextentdoyouthinkAIwouldhaveadisruptiveimpacttoyourindustryinthenext5years?2artificialintelligencewillgreatlyimpacttheirindustryintelligencewillhavenoimpactMcKinsey&Company201TheinvestmentstakesofseveralAIdealsbyTencent(China)werenotdisclosedandareincludedMcKinsey&Company20SOURCE:CBInsights,McKinseyHowever,theU.S.exceedsChinainthenumberanddiversityofpromisingAIstartupsandmajorM&Adeals,indicatingamorerobustAIecosystemisingAIstartupsNotableM&isingAIstartupsJanJulJanJul33SOURCE:McKinseyChinaAISurvey2016McKinsey&Company21ThebiggestchallengeforAIinChinamightbecommitmentIsAItechnologyastrategicpriorityfortheCEOorC-levelexecutiveteam?%ofrespondentsNotastrategicpriorityoronCEO’sagendaToppriority4346OneofstrategicprioritiesSOURCE:McKinseyChinaAISurvey2016McKinsey&Company222.82.2Data&ComputingcapacityStrategicmaturityTalent&NetworkAlgorithm&AnalyticsOverallcialServices2.82.2Data&ComputingcapacityStrategicmaturityTalent&NetworkAlgorithm&AnalyticsOverallcialServices2.02.42.5aTechnologyRetail&ConsumerGoods3.12.5LeadingInternetplayers(Baidu,AlibabaTencentandJD)2.01.5IndustryplayersdonotfeelreadyforAIduetolackofTalentandAlgorithms&analyticsReadinessscore