AI in Financial Services 2026: Adoption, ROI, and Governance Benchmarks

Adoption Is Broad, but Maturity Is Uneven

A significant share of financial services organizations report that AI is at least partially integrated into their operations, with 41% describing full integration—meaning AI is embedded across core processes, not just running in isolated experiments.

That headline figure, however, masks meaningful variation across subsectors. Banks and large financial institutions tend to be further along, supported by larger technology budgets and clearer near-term return profiles. Insurers and wealth and asset managers often face heavier constraints from legacy infrastructure, which slows both deployment and scaling.

The types of AI in use span a wide range of capabilities:

• 69% are using generative AI
• 60% are using prediction AI
• 55% are using language AI
• 51% are using agentic AI

The 51% figure for agentic AI is notable. Autonomous, multi-step AI workflows are no longer a fringe concept in financial services—they are already in active use across a majority of surveyed organizations.

The Barriers Are Structural, Not Motivational

Strong adoption numbers do not mean the path is smooth. The survey identifies three leading inhibitors to AI deployment:

• Security and privacy concerns — 33%
• Data quality and availability issues — 32%
• Integration with legacy systems — 27%

These are not new problems, but AI makes them more consequential. As Erin Sims, Financial Services Senior Analyst at RSM UK, puts it: “Financial services firms are rich with data, but that’s also the challenge. You need a single source of truth. Data governance is critical, but so is treating data like a business risk and a product, with key performance indicators around accuracy, completeness and timeliness.”

The industry’s regulatory environment compounds these challenges. Data flows across third-party vendors and fintech partners create additional governance surface area. Without consistent standards, AI can amplify existing weaknesses rather than resolve them.

Investment Is Substantial and Accelerating

Financial services organizations are committing real capital to AI. Seventy percent plan to invest at least $1 million in AI in the current fiscal year, with 31% expecting to spend $3 million or more.

Current investment is concentrated in:

• AI software and embedded AI solutions — 52%
• Data platforms and infrastructure — 36%
• AI research and innovation — 36%
• Upskilling internal talent — 28%

Looking ahead, 82% of respondents expect AI spending to increase next fiscal year, with 73% anticipating increases of up to 25%.

Satisfaction Is High, but Success Has Prerequisites

Nearly all respondents (98%) report at least some satisfaction with the business value delivered by their AI solutions, and 63% describe themselves as very satisfied. When asked what is driving success, the most common factors cited were:

• Having the right AI technologies for business needs — 39%
• Sufficient infrastructure for AI workloads — 34%
• The ability to measure ROI — 33%
• Established governance and policies — 33%
• A clear and well-defined AI strategy — 33%
• Appropriate AI use cases — 33%

The clustering of these factors at similar percentages suggests that no single element dominates—successful AI deployment appears to require several conditions to be true simultaneously.

AI in the Tax Function: A Standout Case

The survey included a dedicated set of questions for respondents whose organizations are actively using or planning AI in their tax functions. Among financial services respondents in this group, 88% reported that their tax departments already use AI tools—formally or informally.

When asked how AI will change the nature of tax work over the next two to three years, respondents identified four emerging role profiles:

• AI orchestrators: Tax professionals primarily managing, training, and overseeing AI systems — 51%
• Strategic tax advisors: Shift from compliance execution to strategic planning and business partnership — 44%
• Tax data scientists: Focus on data quality, model governance, and analytics — 41%
• Hybrid specialists: Combination of tax expertise and AI skills as a standard expectation — 38%

Tax functions are a logical early adopter. Many tax processes are structured, repeatable, and rule-bound—exactly the conditions where AI performs reliably. The high uptake rate and the clarity of the role evolution described here suggest that tax is ahead of many other enterprise functions in its AI maturity.

Governance as a Competitive Variable

Across the survey findings, governance emerges not as a compliance checkbox but as a direct determinant of AI performance. Organizations that have established governance frameworks and clear accountability structures are more likely to report high satisfaction and measurable ROI.

The practical steps that appear to differentiate successful deployments from struggling ones include:

• Treating data as a business asset and risk: Investing in data governance, standardization, and quality metrics before scaling AI—not after.
• Embedding AI into existing workflows: End-to-end integration rather than stand-alone tools that sit outside daily processes.
• Clarifying ownership and accountability: Defining who owns AI strategy, governance, and ROI measurement across the enterprise.
• Investing in change management and skills: Building cross-functional teams that bring together operations, technology, risk, and compliance.
• Strengthening third-party oversight: Extending governance and controls to vendors and partners that use or provide AI-enabled solutions.

What the Data Actually Tells You

The 2026 RSM survey data presents a financial services sector that has moved past the question of whether to adopt AI and is now grappling with how to scale it responsibly. Adoption is broad, investment is growing, and satisfaction is high—but the organizations reporting the strongest outcomes share a common profile: they invested in data infrastructure first, embedded AI into existing workflows, and built governance frameworks before scaling.

The organizations most at risk are those chasing near-term AI gains while deferring the foundational work—data quality, legacy modernization, and clear accountability structures. The survey data suggests that gap will become harder to close as the pace of AI deployment accelerates.

For teams evaluating AI tools in financial services contexts, the practical implication is straightforward: tool selection matters less than integration depth and data readiness. A well-integrated tool on clean data will consistently outperform a more sophisticated tool bolted onto a fragmented data environment.