AI数据互联互通状况报告-2026年展望（英文版）.pptxVIP

StateofAIData

ConnectivityReport:

2026Outlook

Over200dataandAIleaderssaydatainfrastructureisthebiggestbarriertoAIsuccess

TheAugust2025MITreport,TheGenAIDivide:StateofAIinBusiness2025,1madewavesamong

businessleadersandAIproductownerslargelyduetoitsheadlinestatistic:95%ofgenerativeAIpilotsatcompaniesarefailing.Withtheunprecedentedscaleofinvestmentandthehighexpectationsfor

enterpriseapplicationsoflargelanguagemodels(LLMs),bothGenAIevangelistsandskepticswerequicktoweighinonthedisappointingoutcomesoftheseearlyexperiments.

Whiletheaccuracyofthatspeciicstatisticcontinuestobedebated,thecoreissueitsurfacesisnot:alargeshareofcompaniesarefailingtorealizemeaningfulROIfromtheirAIefforts.Themoreimportantquestionis,why?

Wesurveyed200+dataandAIleaders,bothfromenterpriseswithinternalAIadoptioninitiativesaswellassoftwarecompaniesembeddingAIcopilotsandagentsintotheirproducts.Andhere’swhatwelearned:enterpriseAIisnolongerlimitedbymodels.It’sconstrainedbydatainfrastructureandenterprisecontext.

ThestrongestpredictorofAIsuccessin2026isthematurityoftheunderlyingdatainfrastructurethatdeliversenterprisecontexttothesemodels.

Infact,60%ofcompaniesatthehighestlevelofAImaturityalsohavethemostmaturedata

infrastructure.Andtheinverseisalsotrue:53%ofcompanieswithimmatureAIhaveimmaturedatasystems.

Inthisreport,AImaturityreferstotheextenttowhichanorganizationhasoperationalizedAI,movingbeyondexperimentationtomeasurablebusinessimpact.Ourframeworkconsidersdimensions

suchasmodeldeployment,dataintegrationmaturity,governance,andROItracking.Wecategorizematurityinaive-stageprogressiveframeworkthatdrawsfromEY-Parthenon’sAImaturitymodel:

experimenting,implementing,scaling,optimizi