AI in Finance Statistics 2026

Artificial intelligence has crossed from pilot to production in financial services. Banks are chasing a value pool worth up to $340 billion a year, fraudsters and the systems that stop them now run on the same generative tools, and the EU AI Act has reclassified credit scoring and insurance pricing as high-risk activities. This page pulls together more than 50 cited statistics from regulators, central banks, consultancies and named institutions to show where AI in finance actually stands in 2026, and where the numbers disagree.

A note on method before the data: adoption figures vary widely between sources because each survey counts a different population. A central bank measuring every firm reports a far lower number than a consultancy surveying large enterprises. Throughout this page, figures are paired with the population they describe, and forecasts are labelled as estimates. Several figures are TheAIDaily compilations that combine two or more primary sources into a single comparable number, with the underlying sources named each time.

AI adoption in financial services

Financial services is now one of the most AI-intensive sectors of the economy, but how intensive depends entirely on who is counted. The most representative figure comes from the US Federal Reserve, which measures every firm rather than a self-selected sample. Its data puts financial-sector adoption at 30%, behind only professional services and almost double the economy-wide rate.

These numbers are not contradictory once the populations are separated. The table below shows why a single headline adoption rate for finance does not exist, and what each survey actually measured.

The cleanest cross-industry comparison also comes from the Federal Reserve, and it is the single most useful adoption stat on this page: finance sits far above the wider economy. The year-on-year momentum is the other half of the story, with KPMG recording a clean doubling of active AI use inside finance functions in just two years.

• Against an 18% economy-wide average, the US financial sector reaches 30% AI adoption, second only to professional services at 33%, which makes finance one of the two most AI-intensive parts of the US economy (Federal Reserve FEDS Notes, 2026).
• Active AI use inside finance functions more than doubled from 30% to 75% between 2024 and 2026, one of the fastest two-year jumps recorded in any sector (KPMG, 2026).
• Agentic AI is the defining 2026 theme: 42% of financial firms are using or assessing AI agents and 21% have already deployed them, while a separate survey found 52% experimenting (NVIDIA and Cambridge CCAF, 2026).
• Reading the surveys together, representative adoption is near 30% across all financial firms but rises to 65% to 80% among large institutions, so the right figure depends entirely on firm size (TheAIDaily, based on Federal Reserve, NVIDIA, KPMG and CCAF).
• Almost no firm is opting out: nearly 100% of financial institutions plan to maintain or increase their AI budgets next year, and 73% of executives call AI crucial to future success (NVIDIA, 2026).

Sources: Federal Reserve FEDS Notes "Monitoring AI Adoption in the U.S. Economy" (Apr 2026); KPMG 2026 Global AI in Finance Report; NVIDIA State of AI in Financial Services 2026; Cambridge Centre for Alternative Finance 2026 Global AI in Financial Services Report; Bank of England and FCA AI in UK Financial Services 2024.

Generative AI in banking statistics

Generative AI is where the money and the attention concentrate. McKinsey's value-at-stake model remains the anchor figure for the whole sector, and it has been repeated so often precisely because no later estimate has displaced it. The adoption curve underneath that number has moved sharply: in 2023, only 1 in 10 banks ran generative AI in production. Two years later, nearly half do.

The $200 billion to $340 billion figure is a 2023 estimate of potential value, not realised value, and most of it comes from productivity rather than new revenue. McKinsey's breakdown shows the value is concentrated in two segments, which matters for any bank deciding where to start.

• Corporate and retail banking hold the bulk of the prize: roughly $56 billion of potential genAI value sits in corporate banking and $54 billion in retail, meaning about $110 billion of the $200 billion to $340 billion pool is concentrated in just two segments (McKinsey, 2023).
• Institutional adoption is now near-total: every surveyed global systemically important bank, insurer and asset manager already applies AI in production or pilot, and 54% of those banks are piloting agentic AI (IIF and EY Annual Survey on AI Use in Financial Services, 2025).
• Revenue expectations are rising fast: 58% of banks anticipate a 6% to 20% revenue uplift from generative AI, and 79% expect a further 6% to 20% of revenue growth over the next two years (EY-Parthenon, 2025).
• Longer-horizon profit models agree on direction: Citi projects AI could lift global banking profits by about 9%, or roughly $170 billion, to near $2 trillion by 2028, while BCG sees retail banks alone unlocking more than $370 billion in additional annual profit by 2030 (Citi GPS 2024 and BCG 2025).

One caution belongs next to every one of these figures: the gap between expecting value and booking it is wide, and the section on ROI below shows how wide.

Sources: McKinsey "Capturing the full value of generative AI in banking" and Global Banking Annual Review 2025; EY-Parthenon Generative AI in Banking survey 2025; IIF-EY 2025 Annual Survey on AI Use in Financial Services; Citi GPS "AI in Finance" 2024; BCG "From Branches to Bots" 2025.

AI in finance market size and investment

Market-size forecasts for AI in finance look contradictory until you read the scope label. A report on "AI in banking" covers far more than one on "generative AI in financial services," which in turn dwarfs one on "AI agents in financial services." The figures below span more than a 10-fold range purely because of what each one includes. All are vendor estimates, not measured data.

Spending data is more grounded than market sizing because it is tracked rather than projected. Banking is the single biggest AI-spending industry in the world, and in Europe financial services is the largest proportional AI spender of any sector.

• Stripping out the scope differences, the broad "AI in finance" market sits around $35 billion to $40 billion in 2025 and is on track for roughly $190 billion to $240 billion by the early 2030s, a 22% to 31% CAGR across the major forecasts (TheAIDaily, based on MarketsandMarkets, Fortune Business Insights, Precedence Research and GMInsights).
• Spending lags the value on the table: banks spent about $27 billion on AI across the Americas and EMEA in 2024, while McKinsey's value model points to $200 billion to $340 billion of annual potential, implying most of the prize is still uncaptured (TheAIDaily, based on IDC 2024 and McKinsey 2023).
• AI dominates venture capital but not through fintech: AI took a record 48% of all global VC in 2025 ($226 billion), and all six largest rounds went to AI companies, while fintech ranked second at 11% ($52.7 billion) (CB Insights, 2025).
• Within fintech, AI deals are still growing: AI-focused fintech funding rose to $16.8 billion from $12.1 billion year-on-year, even as fintech's share of all AI deals slipped to 6.5% as horizontal AI absorbed the capital (CB Insights, 2025).

Sources: MarketsandMarkets, Fortune Business Insights, Precedence Research, GMInsights, Grand View Research (market sizing, all estimates); IDC Worldwide AI Spending Industry Outlook 2024-2025; CB Insights State of Venture and State of Fintech 2025.

ROI and productivity from AI in finance

The most credible evidence for AI in finance is no longer surveys of expectation but the deployments named banks now disclose. Several of the largest institutions in the world report concrete productivity figures, and they point the same way: double-digit efficiency gains and tens of thousands of staff using internal AI tools daily.

Productivity potential is highest in banking of any industry, but turning potential into booked profit is where most programmes stall. The honest picture is a wide gap between expectation and realisation.

• Banking has the most exposed work of any sector: 54% of banking jobs have high automation potential and a further 12% could be augmented, ahead of insurance at 46% and capital markets at 40% (Citi GPS, 2024).
• Expectation runs well ahead of realisation: 71% of finance leaders say AI meets or exceeds ROI expectations, yet only about 6% see payback within a year and 95% of enterprise generative AI pilots show no measurable profit-and-loss impact, so the value leaks between piloting and scaling (TheAIDaily, based on KPMG 2026, Deloitte 2025 and MIT 2025).
• Revenue and cost both move: 89% of financial firms report AI is raising revenue while lowering operating costs, with 64% seeing revenue up more than 5% and 61% seeing costs down more than 5% (NVIDIA, 2026).
• Named-bank disclosures now anchor the case: JPMorgan, Morgan Stanley, Goldman Sachs, Citi and DBS have each put real numbers on internal AI, from Citi's 100,000 hours saved a week to DBS targeting around SGD 1 billion in value, which is harder to dismiss than survey expectations (company disclosures, 2024-2025).

Sources: Accenture "Banking in the Age of Generative AI" 2024; KPMG 2026 Global AI in Finance Report; Deloitte "AI ROI" survey 2025; MIT Project NANDA State of AI in Business 2025; Citi GPS 2024; NVIDIA State of AI in Financial Services 2026; company disclosures (JPMorgan, Morgan Stanley, Goldman Sachs, Citigroup, DBS).

AI in insurance statistics

Insurance has moved from cautious experimentation to broad adoption, and Europe's regulator has the cleanest data. EIOPA's surveys show generative AI spreading quickly through the sector while traditional AI is already embedded across pricing, underwriting and claims.

The investment signal is even stronger than the adoption signal. After years of decline, InsurTech funding rose again in 2025, and almost all of the new money is flowing to AI-centred companies.

• AI now captures nearly all InsurTech capital: AI-centred InsurTechs took about 66% of 2025 funding ($3.35 billion across 227 deals) and a record 95.2% in the first quarter of 2026, as overall InsurTech funding rose 19.5% to $5.08 billion (Gallagher Re, 2026).
• European insurers lean on AI for the back office first: 64% of generative-AI use is back-end productivity such as data extraction and coding assistance, against 36% for customer-facing applications like chatbots (EIOPA, 2026).
• The long-run value case is large but modelled: McKinsey estimates generative AI could unlock $50 billion to $70 billion in additional insurance value, with more than half of claims activities automatable by 2030 (McKinsey, estimate).
• Demand-side pressure is building too: 71% of large businesses across six major economies have implemented generative AI in at least one function, which is reshaping the liability and cyber risks insurers must price (The Geneva Association, 2025).

Vendor market-sizing for AI in insurance varies widely, from about $10 billion to $19 billion in 2025 depending on the firm, all converging on a CAGR near 32% to the mid-2030s. As with the rest of this page, treat those as estimates.

Sources: EIOPA Generative AI survey (Feb 2026) and Report on the digitalisation of the European insurance sector (2024); Deloitte 2025 Global Insurance Outlook; Gallagher Re Global InsurTech Report 2026; McKinsey "The future of AI in the insurance industry"; The Geneva Association "Gen AI Risks for Businesses" 2025; Precedence Research and Market Research Future (market sizing, estimates).

AI in asset management, wealth and trading

In investing, AI has shifted from edge to infrastructure, but managers are deliberately keeping it on a short leash. Adoption is high for analysis and operations and almost non-existent for autonomous decision-making, which is the most important nuance in this whole section.

The leash matters. Managers use AI heavily for efficiency and research but almost never let it make the call, and the measured return improvement so far is small.

• AI is a partner, not a decision-maker: only 5% of asset managers give AI autonomous or semi-autonomous authority, and while 69% report efficiency gains, just 8% report a measurable improvement in investment returns (Mercer, 2026).
• Generative AI is near-universal in hedge funds, climbing from 86% to 95% adoption in roughly 18 months, faster than almost any other finance segment, with 58% expecting to expand its use in investment processes (AIMA, 2024-2025).
• Allocators are rewarding AI capability: 60% of institutional investors say they are more likely to allocate to funds with meaningful generative-AI budgets, and 90% believe it will lift performance for at least some managers within three years (AIMA, 2025).
• Markets are already largely machine-run: more than 70% of spot FX orders on a primary inter-dealer platform are submitted by algorithms, and AI content in algorithmic-trading patents rose from 19% in 2017 to over 50% every year since 2020 (Bank for International Settlements and IMF).

Sources: Mercer "How AI is shaping asset management" 2026; AIMA "Charting the course" 2025 and "Getting in pole position" 2024; Statista Robo-Advisors market outlook 2025 (estimate); J.P. Morgan e-Trading Edit survey 2024; Bank for International Settlements FX and HFT research; IMF Global Financial Stability Report 2024.

AI in payments and fintech

Payments is one of the fastest-moving AI use cases in finance, driven by real-time rails and the need to make fraud decisions in milliseconds. Asia is leading deployment, and the underlying shift to account-to-account payments is creating new surfaces where AI optimises and protects transactions.

The structural backdrop is the rise of account-to-account payments, which now move a meaningful share of commerce and lean on AI for routing and fraud decisions.

• Real-time payments are becoming mainstream: account-to-account payments already account for 19% of e-commerce and 9% of point-of-sale value in 2025, with global A2A value forecast to reach $3.8 trillion by 2030, and AI is the layer making real-time optimisation and fraud scoring possible (Worldpay Global Payments Report, 2025).
• Asia is setting the pace in payments AI, with Singapore at 73% adoption and Vietnam at 57%, ahead of most Western markets on this specific use case (Finastra Financial Services State of the Nation Survey, 2026).
• Fintech is the second-largest AI funding destination behind horizontal AI, taking 11% of all global venture capital in 2025 ($52.7 billion), up 35.5% year-on-year (CB Insights, 2025).
• AI is becoming the default for new fintechs: more than 60 of CB Insights' 100 most promising fintech startups deploy AI, with agents moving from internal workflows toward core financial infrastructure (CB Insights, 2025).

Sources: Finastra Financial Services State of the Nation Survey 2026; CB Insights State of Fintech and Fintech 100, 2025; Worldpay The Global Payments Report 2025.

AI regulation in finance: EU AI Act and DORA

The regulatory picture changed materially in 2026, and any page citing the old dates is now wrong. The EU AI Act classifies core financial activities as high-risk, but the AI Act Omnibus reached a political agreement on 7 May 2026 that pushed the main compliance deadline back by 16 months. For financial firms, the high-risk obligations now bite on 2 December 2027, not 2 August 2026.

What counts as high-risk in finance is specific. Under Annex III of the AI Act, AI used to evaluate creditworthiness or set credit scores is high-risk, except where it is used purely to detect fraud, and so is AI used for risk assessment and pricing in life and health insurance.

• The deadline that matters for lenders and insurers moved: high-risk obligations for credit scoring and insurance pricing now apply from 2 December 2027 after the AI Act Omnibus, 16 months later than the original 2 August 2026 date, with formal adoption expected in mid-2026 (EU AI Act Omnibus, 2026).
• Transparency rules did not move: firms using AI chatbots or generating synthetic content for customers must disclose it from 2 August 2026, and the AI-literacy obligation has applied to every organisation since 2 February 2025 (EU AI Act, 2026).
• Penalties scale with global turnover: prohibited practices carry fines up to €35 million or 7% of worldwide annual turnover, high-risk breaches up to €15 million or 3%, so for a bank with €50 billion in revenue a top-tier breach could in principle reach €3.5 billion (EU AI Act, Article 99).
• DORA has already reshaped third-party oversight: in November 2025 the European Supervisory Authorities named the first 19 critical ICT third-party providers, including AWS, Google Cloud and Microsoft, placing the cloud backbone of finance under direct EU oversight (ESAs, 2025).
• Sectoral rules are spreading beyond Europe: Singapore's MAS issued AI model risk management guidelines in December 2024, and the US Treasury warned in 2024 that generative AI can supercharge fraud at scale (MAS and US Treasury).

Sources: EU AI Act, Regulation (EU) 2024/1689, Annex III and Article 99, and the AI Act Omnibus (political agreement 7 May 2026); EBA "AI Act: implications for the EU banking and payments sector" 2025; ESMA and EIOPA on DORA; ESAs critical ICT third-party provider designation (Nov 2025); MAS AI Model Risk Management 2024; US Treasury "Managing AI-Specific Cybersecurity Risks in the Financial Services Sector" 2024.

Risk, governance and the AI talent gap

Adoption has outrun governance. Regulators and central banks are increasingly explicit that the risks are systemic, not just operational, and the surveys show financial firms often do not fully understand the models they already run.

The systemic concern is concentration. Because a handful of providers supply the models, chips and cloud that finance runs on, regulators worry a single failure could ripple across the system.

• Central banks have named the systemic risks: the Financial Stability Board flags third-party concentration, market correlations, cyber risk and model governance as AI vulnerabilities with systemic potential, while the IMF warns AI could exacerbate market concentration and volatility (FSB 2024 and IMF Global Financial Stability Report 2024).
• Firms run models they do not fully grasp: only 34% of UK financial firms claim a complete understanding of the AI they use, 46% only partial, even though 84% have a named person accountable for it (Bank of England and FCA, 2024).
• Governance is improving but from a low base: the share of financial firms with AI governance frameworks rose from 21% to 36% year-on-year, yet about two-thirds of industry respondents are still not monitoring for bias or discrimination (NVIDIA 2025 and Cambridge CCAF 2026).
• Talent is the binding constraint: AI is the single largest skills gap in UK financial services at 35 percentage points between demand and supply, and workers with specialised AI skills command a 56% wage premium, up from 25% a year earlier (Financial Services Skills Commission 2025 and PwC 2025).
• Data quality, not algorithms, is the top barrier: poor data availability and quality is cited as the leading obstacle to AI adoption by 66% of vendors and 40% of industry firms, ahead of legacy systems and regulation (Cambridge CCAF, 2026).

Sources: Financial Stability Board "The Financial Stability Implications of Artificial Intelligence" 2024; IMF Global Financial Stability Report 2024; Bank of England and FCA AI in UK Financial Services 2024; NVIDIA State of AI in Financial Services 2025; Cambridge CCAF 2026 Global AI in Financial Services Report; Financial Services Skills Commission 2025; PwC 2025 Global AI Jobs Barometer.

AI in finance by region and consumer trust

AI in finance is not evenly distributed. North America leads on adoption, Asia leads on specific use cases like payments, and the Nordics lead Europe. Underneath the regional picture sits a paradox that defines consumer-facing finance in 2026: people use AI for money decisions far more than they trust it.

Trust depends heavily on the task. People are comfortable with AI watching for fraud behind the scenes, and deeply uncomfortable letting it move their money. The bar chart shows how sharply comfort collapses as AI moves from monitoring to acting.

The regional adoption gap is wide, and the place to read Europe's finance sector is the European Central Bank rather than the standard enterprise survey, because the EU's main statistics largely exclude financial firms.

• Usage now runs nearly three times ahead of trust: 55% of consumers use AI for financial decisions while only 19% say they trust it in finance, a roughly 36-point trust-usage gap that is unique to this sector (TheAIDaily, based on TD Bank and YouGov, 2026).
• The country spread is about 35 points: active AI deployment in finance ranges from Vietnam at 74% to Japan at 39%, with only 2% of organisations globally using no AI at all (Finastra, 2026).
• For the EU finance picture, use the ECB: roughly two-thirds of euro-area firms report employees using AI but only about 7% use it significantly, a better gauge than the all-sector Eurostat figure of 20% because EU enterprise statistics largely exclude financial firms (ECB SAFE survey and Eurostat, 2025-2026).
• The Dutch financial sector sits well ahead of the EU average: 59% of Dutch financial firms use AI, the highest of any sector after ICT, against a 20% all-enterprise EU average, illustrating how concentrated finance adoption is even within Europe (TheAIDaily, based on CBS and Eurostat). See our Dutch AI in finance statistics for the national breakdown.
• Comfort is highest for invisible AI: consumers are most at ease with AI in fraud detection (67%), spending tracking (66%) and credit scoring (66%), and far warier of AI giving direct financial advice (TD Bank, 2026).

Sources: KPMG "AI in Financial Reporting and Audit" 2024; Finastra Financial Services State of the Nation Survey 2026; YouGov US survey (Jan 2026); TD Bank and Wells Fargo consumer surveys via ABA Banking Journal 2026; ECB SAFE survey 2026; Eurostat "Use of AI in enterprises" 2025; CBS (Dutch financial sector AI use).

Key takeaways

• Finance is one of the most AI-intensive sectors, but the headline number is a range. Representative adoption is near 30% across all financial firms (Federal Reserve) and rises to 65% to 80% among large institutions (KPMG, NVIDIA, CCAF). Always check which population a figure describes.
• The value is real but largely uncaptured. McKinsey's $200 billion to $340 billion of annual genAI value for banking is potential, not booked. Only about 6% of firms reach payback within a year, and the gap between piloting and scaling is where returns leak.
• AI is now on both sides of fraud. Over half of fraud attempts involve AI, deepfake attempts are up more than 2,000% in three years, and 90% of banks fight back with AI. The world spends about $206 billion on compliance while illicit flows hit $4.4 trillion.
• Investing has high adoption and low autonomy. Hedge funds (95%) and asset managers (55%) have embraced AI for research and operations, but only 5% let it make decisions and only 8% can measure a return benefit.
• The key regulatory date moved. High-risk financial AI such as credit scoring must comply by 2 December 2027, not 2 August 2026, after the AI Act Omnibus. Transparency rules for chatbots still apply from August 2026.
• Trust lags usage badly. Finance is the least-trusted sector for AI (19%), yet 55% of consumers already use it for money decisions. Comfort is high for behind-the-scenes AI and collapses for autonomous money movement.