Banks investing in AI see better risk control but lower short-term returns

A new study of large U.S. commercial banks shows that while AI-driven innovation improves core risk management and asset quality, it simultaneously increases operating costs and suppresses profitability in the short term, revealing a structural shift in how financial institutions absorb technological change.

The research, published in the Journal of Risk and Financial Management as "AI Innovation and Bank Performance: Evidence from Patent Activity of Large U.S. Commercial Banks," analyses how AI affects bank performance at an institutional level. Based on data from 31 large U.S. commercial banks between 2015 and 2024, the study uses AI-related patent activity as a proxy for innovation capacity and links it with financial performance indicators across the CAMEL framework, including capital adequacy, asset quality, management efficiency, earnings, and liquidity.

The findings challenge the assumption that AI adoption leads to immediate financial gains, instead revealing a more nuanced trajectory marked by short-term costs and long-term efficiency gains.

AI strengthens risk management but creates short-term profit pressure

The study finds that banks investing heavily in AI innovation experience a measurable improvement in asset quality, particularly through reductions in non-performing loans. This suggests that AI tools are already delivering value in credit risk assessment, loan monitoring, and predictive analytics, allowing banks to make more informed lending decisions and manage portfolio risks more effectively.

These improvements are significant because asset quality is a core indicator of banking stability. By enhancing screening and monitoring processes, AI enables banks to reduce information asymmetry and improve decision accuracy, which translates into stronger loan performance.

However, the same banks also face rising operational costs and declining profitability in the short run. The research shows that AI innovation is associated with higher non-interest expenses relative to income, reflecting the substantial investments required in data infrastructure, computing systems, and specialized talent. At the same time, return on assets declines, indicating that these investments do not immediately translate into financial returns.

This dual effect highlights what the study describes as a productivity "J-curve," where initial investments in advanced technologies lead to temporary performance deterioration before longer-term benefits emerge. Banks must absorb these upfront costs while restructuring their operations to fully leverage AI capabilities.

The findings also reveal that AI innovation is associated with lower capital ratios, not necessarily as a sign of weakness but as a reflection of more efficient balance sheet management. With improved risk assessment tools, banks can optimize capital allocation without compromising regulatory compliance.

Organizational restructuring drives AI's impact on banking performance

The study identifies how AI innovation reshapes bank operations through organizational changes. The research shows that banks with higher levels of AI innovation tend to reduce both employee headcount and physical branch networks, signaling a shift toward more automated and digitally driven business models.

These structural adjustments act as a critical transmission channel through which AI affects performance. By reducing reliance on labor-intensive processes and physical infrastructure, banks can lower long-term operating costs and improve efficiency. However, these changes also involve transition costs, including workforce restructuring and system integration, which contribute to the short-term decline in profitability.

The study's mediation analysis demonstrates that reductions in employee scale and branch networks partially offset the negative financial effects of AI investment. While AI increases costs directly through technological spending, it simultaneously improves efficiency indirectly by streamlining operations.

For example, a smaller workforce is associated with lower cost-to-income ratios and improved profitability, while a reduced branch footprint aligns with the broader shift toward digital banking services. These findings confirm that AI is not just a technological upgrade but a catalyst for organizational transformation.

These changes are not uniform across all banks. Larger institutions, with greater resources and technological capabilities, are better positioned to implement AI-driven restructuring and realize efficiency gains more quickly. Smaller banks, by contrast, may face longer adjustment periods and more limited benefits.

Long-term gains depend on full-scale AI adoption

While patent activity reflects a bank's technological capability, the study shows that real economic benefits depend on whether these technologies are deployed across the organization.

Banks that achieve firm-wide AI adoption are able to mitigate the negative short-term effects of innovation on management efficiency and profitability. In these cases, the initial cost pressures are significantly reduced, and performance begins to improve as AI systems are integrated into core operations.

The findings suggest that partial or fragmented adoption limits the impact of AI, as isolated applications fail to generate system-wide efficiencies. In contrast, coordinated deployment allows banks to leverage economies of scale, standardize processes, and enhance data sharing across business units.

The study also highlights the dynamic nature of AI's impact. Over time, the relationship between AI innovation and operating costs reverses, with costs declining as efficiency gains materialize. Similarly, profitability, which initially declines, begins to recover and improve in the long run. This pattern is particularly evident in large banks, where the scale of investment enables faster realization of benefits. These institutions experience sharper initial cost increases but also achieve stronger and earlier efficiency gains compared to smaller peers.

The research asserts that AI should be viewed as a long-term strategic investment rather than a short-term performance driver. Banks that align technological innovation with organizational restructuring and full-scale adoption are more likely to capture its economic value.

A transformational shift for banking strategy and regulation

Traditional performance metrics may not fully capture the transitional effects of AI adoption, as short-term declines in profitability and efficiency can mask long-term improvements in stability and productivity.

Regulatory frameworks may need to adapt to account for these dynamics, particularly in assessing capital adequacy and operational risk. Supervisors must balance the need to ensure financial stability with the recognition that technological transformation involves temporary disruptions.

For bank managers, the study highlights the importance of integrating AI into broader strategic planning. Investment in AI must be accompanied by organizational changes, including workforce restructuring, branch optimization, and process integration, to unlock its full potential.

The research also raises broader questions about the future of banking. As AI continues to automate routine tasks and enhance decision-making, the role of human labor and physical infrastructure is likely to diminish, leading to a more digital and data-driven financial system. Additionally, the uneven distribution of AI capabilities may widen the gap between large and small institutions, with larger banks gaining a competitive advantage through scale and technological resources.