人工智能转型2026：全球和地区全球样本报告：数字支付和电子商务的商业化、企业准备和规模扩张

IT emerges as the primary focus of function-specific agentic AI use cases,while marketing, finance, and procurement form the next-largest clustersacross horizontal business processes. In 2025, 73% of employees intentionally use general-purposegenerative AI tools at work, reflecting the widespread adoption of AIacross professional settings. Global: Types of AI Tools Intentionally Used at Work, in % of usage, 2025 In Southeast Asia, AI Agnostics prioritize seller protection, AI-poweredtools, and pricing transparency to reduce perceived risks and facilitateAI adoption in E-Commerce, as of February 2025. SEA: Key Areas to Improve Seller Experience Among AI Agnostics, in % of E-Com. Sellers, Feb. 2025 As of January 2025, 89% of respondents in China report semi-regularor regular AI use, reflecting widespread everyday engagement with AI. China:Adoptionof AI, in % of Respondents, Nov 2024-Jan 2025 In 2025, 43% of North American firms report partial GenAI integration,up from 34% in 2024, while full integration rises to 25%, signalinggradual maturity across core operations. North America: Organizational Maturity Levels of Generative AI Adoption, in % of Resp., 2024 & 2025 By 2026, U.S. agencies expect GenAI to reshape how they create,deliver, and position content, with majorities citing impact on creation(76%), marketplace (71%), engagement (69%), and deliverables (62%). U.S.: Expected Areas of GenAI Impact Within Agencies, in % of Respondents, June 2024 In 2025, AI tool usage among employees in Europe varied from 64% inFrance to 78% in Spain, illustrating uneven adoption across majormarkets. Europe: Employee Usage of AI Tool by Country, in % of Resp. Using AI Several Times Weekly, June 2025 As of January 2025, regular AI use in Northern and Western Europevaries widely, with rates above 69% in Baltic and Nordic countries,while several Western economies remain below 53%. North. & West. Europe: Adoptionof AI by Country, in % of Res. Report. Regular Use*, Nov 2024-Jan 2025 In 2024, AI adoption across business functions in the Middle Eastremains limited, with 29-41% of organizations reporting partialimplementation and only 14-28% achieving scaled use. Middle East:AI Implementation Levels by Business Function, in % of Respondents, Oct-Dec 2024 In 2025, AI project failures surge across the Middle East as budgetoverruns of up to 1,000% expose widespread gaps in financialplanning. Middle East: Budget Overruns, Financial Planning Deficits, and AI Project Termination Risks, July 2025 AIBudgetsRoutinelyExceedInitialProjections AccordingtoRolandBerger,artificial intelligenceprojects intheMiddleEastoftenexceedoriginal costestimatesby500%to1,000%.Theseoverrunstypicallystemfrominaccuratebudgetingandinsufficientadvanceplanning.Costescalationbeginsearlyintheprojectlifecycleandcompoundsduringimplementation.Asaresult,financialmanagementhasbecomeoneofthemostcriticalandunderdevelopedcapabilitiesacrossAIinitiativesintheregion. HighTerminationRatesReflectFinancialMisalignment Morethan50%ofAIinitiativesintheMiddleEastareultimatelyterminatedduetobudgetoverruns,asnotedbyRolandBerger.Thishighfailureratehighlightsapersistentmisalignmentbetweenexecutiveambitionandactualprojecteconomics.Thefirmfurtherreportsthatfinancialmanagementisthesecond-largestsourceofexecutivedissatisfactionwithAIprojects,followingonlytalentshortages. FinancialPlanningShortfallsLeadtoSuboptimalROI RolandBergerfindsthatfewerthan15%oforganizationsintheregionproperlyidentify,quantify,andassessthecosts,risks,andexpectedvalueoftheirAIprojects.Withoutthisfoundation,manyinitiativesfailduringimplementationordeliverlimitedreturns.Evenprojectswithhighpotentialstruggletojustifyinvestmentduetoweakfinancialstructuringandunrealisticexpectations. UrgentNeedforRigorousFinancialModellingRolandBergerrecommendsdevelopingrobustfinancialmodelsfromtheoutsetofanyAIinitiative.Theseshouldincluderealistic budgetallocations,contingencybuffers,andlong-termexpenditureplanning.Asnotedinthereport,“AIhasthepotentialtotransformtheMiddleEast’seconomy,butwithoutaccuratecostmodellingandregion-specificplanning,eventhemostpromisingprojectsriskfallingshort.”ComprehensivefinancialdisciplineisessentialforensuringAIdeploymentsdeliverstrategicvalue. In 2025, OpenAI GPT models dominate AI framework adoption in LatinAmerica at 82.1%, with custom in-house models at 22.8% and GoogleVertex AI at 21%, showing dominance and rising alternatives. Latin America: Adoption of AI Frameworks by Provider, in % of Respondents, Mar-Apr 2025 Brazilian companies largely expect AI to reshape their industry, with62% sharing this view, making it the dominant perception despite someremaining skepticism in 2025. Brazil: Corporate Expectations of AI’s Impact on Industry, in % of Respondents, Jan 2025 In West Africa, by 2023, 51% of CEOs reported adopting Generative AIacross their companies, while 48% said it had already reshaped theirtechnology strategies, signaling rapid organizational integration. West