AI ambitions hit a Wall: banks grapple with legacy tech as consumer demand surges

By early 2026, AI in banking had decisively moved beyond the pilot and demo stages into delivering tangible commercial impact, as exemplified by U.S. Bank's use of AI-driven Amazon Connect to enhance customer service continuity. Despite initial fears around regulatory pressures and cultural resistance, industry leaders like Prashant Morodi of U.S. Bank acknowledge that these concerns often served as excuses rather than real barriers, with banks increasingly leaning into AI adoption to realize measurable benefits.

While 78% of banks had integrated AI into at least one business function and 84% of leaders anticipated AI's significant impact, only a mere 7% had achieved full enterprise-wide scaling by mid-2026. This gap underscores the challenge of moving beyond narrow, high-value use cases—such as fraud detection, virtual agents, and regulatory tools—toward embedding AI into end-to-end workflows, a transition hindered primarily by legacy core systems, fragmented integrations, and poor data quality rather than the AI models themselves.

Successful scaling of AI in banking hinges on transforming AI from isolated pilots into core infrastructure by connecting it to real workflows, improving data quality, and establishing governance frameworks early. Leading institutions like CoBank and Rocket demonstrate this shift by leveraging AI to preserve institutional knowledge through digital twins and to unify customer data post-acquisition for personalized experiences, signaling growing readiness for scaled AI use and operational transformation.

Achieving enterprise-scale AI adoption requires a fundamental shift in operating models and platform strategies, as highlighted by Rajaram R.K., who emphasizes that AI is now a scaling and operating model challenge rather than a mere technology issue. Banks at the forefront are partnering with firms like Infosys and AWS to move beyond efficiency gains, reimagining entire operations and customer experiences, while regulators such as Singapore’s Monetary Authority actively integrate AI into live environments, collectively accelerating AI-driven modernization across regions.

By early 2026, industry reports such as Temenos's April analysis underscored that modernizing core banking systems is no longer optional but essential for scaling AI and monetizing digital services. While many banks had focused on enhancing digital front ends, neglecting legacy back-end systems has become increasingly unsustainable amid rising demands for real-time services and automation. Experts warn that failure to improve data quality and update these foundational systems risks long-term competitive disadvantage, as technology choices now directly impact innovation capacity and market positioning.

Analyses from May 2026 reveal that outdated core banking infrastructures—characterized by fragmented systems, broken integrations, and siloed data—are the primary bottlenecks preventing banks from moving AI beyond pilot stages. Incremental modernization strategies that connect AI to real workflows, improve data governance, and unify customer, employee, and AI agent interactions through a unified frontline operating system are critical. This approach balances the need for innovation with operational trust and regulatory compliance, transforming AI adoption into a governance and operating model challenge rather than just a technology upgrade.

US Bank’s large-scale collaboration with AWS in May 2026 exemplifies the strategic value of incremental modernization, migrating hundreds of critical applications to the cloud to overcome legacy constraints and enable AI scalability. Integrating generative AI technologies like Amazon Bedrock and Nova Sonic within modernized systems demonstrates an orchestration approach that enhances customer service across voice, chat, and SMS channels while maintaining operational trust. This case highlights how layering new AI capabilities atop existing infrastructure can unlock innovation without sacrificing reliability.

Throughout late May 2026, industry voices including Mladen Vladic and Paymentus’s Garrett Baird emphasized that legacy core banking and payment systems—handling up to 75% of transactions on mainframes—remain indispensable for trust, resiliency, and regulatory compliance. Wholesale replacement is impractical; instead, intelligent modernization through orchestration, middleware, and APIs is the durable path forward. AI emerges as connective tissue that navigates fragmented legacy environments, surfacing relational context and enabling banks to leverage their scale, compliance maturity, and accumulated data as competitive moats. As Visa’s Pismo noted, with roughly 70% of IT budgets consumed by maintaining outdated systems, modernization has shifted from a technical issue to a strategic boardroom priority, where reducing complexity and enhancing speed, simplicity, and intelligence are now key drivers of customer loyalty.

By early 2026, banks were grappling with the critical need to modernize their core banking systems to meet stringent governance and data quality demands essential for scaling AI effectively. The Temenos report underscored that while many institutions prioritized enhancing digital front ends, neglecting legacy back-end systems proved unsustainable as real-time services and automation increasingly required robust governance and cleaner data structures. Experts warned that failure to upgrade these foundational systems risked long-term competitiveness, as technology choices directly impacted banks' abilities to comply with evolving AI governance and regulatory frameworks.

The regulatory landscape, shaped by frameworks like DORA, CRR3, CRD6, and the EU AI Act, has elevated AI from a mere technological endeavor to a comprehensive governance and operational challenge for banks. This shift demands rigorous explainability, oversight, and control, which, combined with legacy system limitations and poor data quality, significantly slow AI scaling compared to non-bank lenders. As one analyst noted in mid-2026, banks must navigate complex layers of model governance, accuracy thresholds, access controls, decision and exception governance, and vendor validation, necessitating a cautious and measured approach to AI deployment.

Successful AI adoption in banking hinges not just on the sophistication of AI models but equally on the quality of data inputs, workflow design, and exception handling mechanisms—a reality often underestimated by lenders. Banks that invest early in building robust governance frameworks and cleansing their data can transform AI from isolated pilot projects into integral operational infrastructure. This comprehensive approach, emphasizing operational design alongside model performance, distinguishes leaders in AI scaling from those stalled at the demonstration phase.

By early 2026, autonomous AI agents have evolved from mere recommendation engines to proactive digital labor capable of executing complex tasks and managing outcomes across banking workflows. These agents learn from data, adapt to real-world conditions, and continuously improve, marking a new era of intelligent automation that fundamentally restructures how banks operate and deliver services.

TD Bank’s launch of an agentic AI system to autonomously manage mortgage lending workflows exemplifies this operational transformation, enabling the bank to reduce manual friction, accelerate approvals, and enhance customer experience. This move from traditional automation to enterprise-level AI deployment signals a broader industry trend of embedding AI deeply into core banking operations, not just customer-facing tools, thereby creating competitive advantages in speed, efficiency, and personalization.

JPMorgan Chase is advancing this frontier by developing long-running autonomous AI agents capable of managing complex workflows across multiple systems for extended periods, a significant leap from short-task AI tools. Although security and governance challenges remain, the bank anticipates corporate deployment in 2026, with early AI-driven tools already boosting private banking gross sales by 20% and potentially expanding client coverage by 50%, illustrating AI’s role in driving sustainable competitive advantage rather than mere cost-cutting.

Collectively, these developments underscore a structural shift in banking where autonomous AI agents are embedded as governed digital labor within core execution models, delivering not only operational resilience and faster decision-making but also preserving institutional knowledge. Banks that embrace this transformation stand to gain significant advantages in revenue growth, cost optimization, and innovation, redefining the future of financial services.

By mid-2026, a clear and pressing gap emerged between soaring consumer demand for AI-powered financial services and the limited AI capabilities currently offered by credit unions. While 75% of SMBs and 59% of consumers expressed willingness to use AI assistants for practical money management tasks—such as bill tracking, budgeting, and credit tips—only about 25% of credit unions provided AI chat support, with even fewer offering AI-driven financial advice or payment tools. This disconnect underscores a strategic imperative for credit unions to integrate AI more directly into their innovation roadmaps to deepen member relationships, attract younger consumers, and engage higher-value SMBs, rather than relegating AI to a distant future feature.

Credit unions face the delicate challenge of modernizing their AI capabilities without undermining the trust and reliability embedded in their legacy systems, which remain central to consumer confidence. As Velera’s Cody Banks articulated, the key is to "modernize without dismantling the reliability consumers already depend on." However, siloed legacy infrastructures restrict real-time data use necessary for AI-driven personalization and fraud detection, creating a bottleneck between consumer expectations for seamless, intelligent services and what current systems can deliver. This has led credit unions to adopt a measured, incremental approach—"a chisel versus a sledgehammer"—to enhance member experience and trust while safeguarding operational continuity.

Strategic partnerships with FinTech firms have become essential for credit unions aiming to innovate rapidly with AI while preserving trust and regulatory compliance. FinTechs excel at moving fast and iterating quickly, but as banks caution, financial institutions must balance speed with the imperative to maintain consumer trust. This collaboration model helps bridge the gap between customer expectations for advanced AI capabilities and the cautious, trust-sensitive deployment environment within credit unions.

Consumer expectations for AI-driven financial products have evolved to prioritize not only personalization but also transparency and accountability, especially among younger generations. According to Plaid’s 2026 report, over half of Americans used AI to manage finances in the past year, with 57% of Gen Z and Millennials viewing AI tools as necessary for effective money management. Trust in AI is bolstered by its 24/7 availability and perceived lack of judgment, with 64% agreeing AI makes financial advice more accessible and 51% preferring AI over human advisors for availability. Yet, despite this growing acceptance, a significant gap remains between these expectations and the current AI product deployment, particularly within credit unions and smaller institutions, highlighting an urgent need to build AI solutions that marry intelligence with trust.

By mid-2026, strategic partnerships between banks and cloud providers have become a cornerstone for accelerating AI adoption and modernizing critical infrastructure. US Bank's expanded collaboration with Amazon Web Services (AWS) exemplifies this trend, as it undertakes one of the largest modernization efforts in financial services by migrating hundreds of applications to the cloud. This partnership not only enhances digital banking experiences through integration of generative AI technologies like Amazon Bedrock and Nova Sonic—enabling seamless human and AI-driven support across voice, chat, and SMS—but also explores innovative applications in fraud detection, compliance automation, and developer productivity, illustrating how ecosystem collaborations drive both operational excellence and regulatory adherence.

Beyond cloud partnerships, collaborations between regulators and financial institutions are pivotal in embedding AI within compliance frameworks and anti-financial crime efforts. Singapore’s Monetary Authority, for instance, is pioneering the use of AI trained on live banking data to bolster regulatory compliance and combat financial crime in real time. Simultaneously, cross-border alliances such as the one between Tencent Cloud and Ryt Bank Malaysia highlight how regional banks in Asia are leveraging AI-powered conversational banking to enhance customer engagement, signaling a broader ecosystem approach that integrates fintech innovation, regulatory oversight, and cloud technology to overcome scaling barriers in AI adoption.