2026年全球金融业人工智能研究报告：决策优势

19 05AI as decision-engine,not cost lever Foreword Active AI use across the finance function has more than doubled in two years. Many organizations now see meaningful business returns, according to our 2026survey. The strongest gains are concentrating in specific places, and a clear gap is opening between organizations seeing performance at scale and those still Across sectors, finance leaders are navigating a sharedset of operational questions: where AI is producing thestrongest performance gains, whether to lead or follow onadoption, how to measure what AI is delivering, how tostrengthen the controls around it, and how to build thetotal workforce, human and AI, that can sustain it. These This report, AI in Finance 2026, examines where thosegains are coming from and what’s driving them. It buildson our 2024 research on AI in financial reporting,expanding the focus to the full finance function —including governance, controls, and the workforce. The Nikki McAllenGlobal Head ofFinance Advisory The dynamic in finance parallels what KPMG's Q1 2026Global AI Pulse found at the enterprise level: AI maturityis outpacing the operating capability to translate it intoperformance. The 2026 finance survey reflects this.Technology and Financial Services organizations areover-represented relative to the market, which means Sebastian StöckleGlobal Head ofAudit Innovation & AIKPMG International One theme stands out. The organizations moving fasteston AI are those that have made trust — governance,controls, human oversight — part of how performancegets built, not a cost of compliance. This sits at the heart Christian StenderGlobal Head ofAI for Tax & LegalKPMG International This report offers a grounded, evidence-based view ofwhere AI is producing performance gains in finance today,where it is not, and what the leaders getting results are Executivesummary The Decision Advantage Performance is not uniform. Organizations withstronger governance and controls report significantlybetter outcomes — in some cases three to six times therate of significant improvement compared to thosewithout. Organizations that formally track AI-relatedKPIs outperform those that do not. Organizations thatare also assurance-ready outperform those with AI adoption across the finance function is broad. More thanthree-quarters of organizations are leveraging AI in financialplanning, reporting and commercial analysis. 71 percentreport AI is meeting or exceeding ROI expectations in theirfinance function. But adoption breadth and exceptionalperformance are not the same thing. The share oforganizations reporting AI is exceeding expectations sits at23 percent — a narrower group than the broader satisfaction 75%vs.30%Active AI use across finance has more thandoubled since 2024 76%of organizations areactively leveraging AI 70%report improveddecision-making quality The operational constraints are consistent. Data qualityand completeness is the most cited barrier andopportunity. Most organizations are upskilling theirexisting teams, but only 28 percent are rethinking the What stands out is where the gains are concentrating. Thestrongest improvements are in decision-making quality,forecast accuracy and responsiveness. These are judgment-heavy areas, not transactional processes. Organizationsdeploying agentic AI report at least 32 percent strongerperformance across key finance metrics, rising to nearly 40 These three categories were measured separately in thesurvey. Adoption maturity, performance outcomes anddeployment patterns differ across each type. Where findings Agentic AIrefers to systems thatcan plan, reason, act and learnautonomously or semi-autonomously AIrefers to the simulation ofhuman intelligence in machinesthat are programmed to think Generative AIrefers to advancedneural networks that learn from largedata sets and create new content, AIasdecision-enginenotacostlever AIasdecision-engine, notacostlever Most assumptions about AI in finance start with efficiency: faster close, fewer errors, lowercost. The data tells a different story. Active AI use in the finance function has moved from The conversation about AI in finance haschanged. Two years ago, the question waswhether AI could deliver. Today the questionis what it should be deployed to do. Theorganizations getting it right are the ones But adoption is moving faster than organizations' ability torealize enterprise-wide value at scale. Traditional ROImeasurement — money in, money saved — does not The organizations pulling ahead are not the onesadopting AI most broadly — most organizations alreadyare. They are the ones reaching the orchestrating phaseof deployment, directing AI into the work where judgment Sebastian StöckleGlobal Head ofAudit Innovation & AIKPMG International Where agentic AI is generating measurable value, itclusters around capacity for growth, responsiveness andimproved customer experience. The finding parallels KPMG's Q1 2026 Global AI Pulse: atthe