Financial institutions are already using generative AI to create content and streamline customer support. Machine learning has supported fraud detection in the industry for years. And many organizations in the sector are looking to AI to help them improve productivity and efficiency, accelerate application development, and even explore new business models.
And yet, many companies in financial services expect their AI initiatives to lose money over the next several years, with infrastructure presenting significant bottlenecks.
That’s according to the Industry Report: Financial Services edition of the 2025 Nutanix Enterprise Cloud Index (ECI). The report shows that the financial services sector is prioritizing generative AI, but also illustrates that organizations need to invest in infrastructure modernization and IT upskilling to support long-term success.
“We’re seeing a huge amount of AI investment and optimism in the industry,” said Sean O’Dowd, head of global financial services strategy and solutions at Nutanix.
“But so far, we’re seeing a lower number of institutions saying they’re getting a positive return on investment. There’s still a gap between their planning and their ability to execute and achieve tangible outcomes. A lot of that has to do with skills gaps, as well as the need for modern infrastructure to stand up these solutions.”
Containerization Supports AI Goals
According to the Financial Services ECI report, 56% of banking and insurance organizations have already begun implementing a generative AI strategy, with another 35% reporting that they have a strategy in place but have not yet begun implementation. The top AI goals in the sector include increasing productivity (cited by 53% of respondents), increasing automation and efficiency (51%), and increasing innovation (49%).
“The hunt is on for areas where Gen AI either improves existing use cases, or allows for use cases that were previously harder to execute with traditional machine learning,” O’Dowd said.
“Organizations are looking at anything that is text-based, to take advantage of natural language processing. That includes understanding customer sentiment, as well as looking for insights in public data.”
In insurance, O’Dowd said, Gen AI may help companies leverage huge quantities of photographic or geospatial data to optimize policies, reduce claim fraud, and even guide real-time response to disasters.
“It has really flipped the switch,” he said.
“You’re seeing amazing stories out of the insurance segment, with home insurance policies, for example. Companies are looking at any of the information they can digitize within the four walls of a policyholder’s home, and how that information can be tapped for new efficiencies or revenue.”
Still, 98% of financial services respondents say they face challenges when scaling Gen AI workloads from development to production, and 52% cite IT infrastructure as an area of required investment to improve Gen AI applications—more than any other area.
This emphasis on infrastructure mirrors the findings of a global report from LSEG, which found that 87% of firms have boosted cloud spending over the past two years, and that 91% are advancing their AI initiatives via the cloud.
O’Dowd noted that containerization is especially effective for supporting Gen AI workloads, due to the need for agile data delivery across complex environments. “Given the nature of data gravity and silos, containers provide a nice package to deliver that data in an agile way,” he said.
Financial services institutions are using containerization for 66% of their Gen AI workloads, according to the ECI report. That’s slightly less than the industry-wide average (70%), and 92% of respondents from the financial services sector say their current IT infrastructure requires improvements to fully support cloud-native applications and containers.
In an interview with The Forecast, IBC Bank Chief Technology Officer Michael Chavez expressed the uncertainty that financial institutions face as they try to navigate AI infrastructure decisions.
“We’re trying to figure out where it’s going to run,” he said. “Does it fit better in the cloud? Does it fit better on-prem? We’re leveraging partners to help us explore and build out a roadmap.”
Bridging Additional AI Gaps
After infrastructure, issues around people and data dominate Gen AI areas for improvement.
Around half of financial services respondents in the ECI report cite both IT training (49%) and IT talent hiring (47%) as needed areas for investment. And significant numbers cite cybersecurity (40%), data management (37%), and data governance (33%).
“The ‘skills need’ is an enormous issue,” O’Dowd said.
“You see these bonkers deals that big tech companies are making for AI talent. Banks and insurers aren’t able to offer nearly that level of compensation, so you can imagine the scramble there is to fill that skills gap.”
Data control and security are also extremely important, O’Dowd added.
“The sector is seeing regulation really start to swirl on this, and organizations are trying to think about how to contain these use cases while mandates and regulations are still being formulated. And then, data quality continues to be one of the most important elements of this, because the AI applications are only as good as the information they are trained on. Garbage in, garbage out.”
Over the next year, 39% of financial services organizations expect their AI initiatives to lose money, with 47% expecting to see a gain. Optimism improves as the time window lengthens, with only 27% of respondents saying their organizations will likely lose money on their AI initiatives over the next one to three years.
However, O’Dowd cautioned that current return-on-investment projections are largely the result of speculation, rather than detailed profit-and-loss analyses.
“I think it’s more speculative in nature,” he said. “Those go-forward projects are based on sentiment as they sort our quantifiable metrics, likely things like time savings and improved productivity. However, they’re not just talking. Everyone is organizing and diligently creating plans to move forward. Institutions are standing up governance boards and AI teams to tackle these new challenges. Like when any new technology comes along, using AI requires policies and validation that build over time.”
O’Dowd believes that those institutions that can bridge the distance between boardroom AI ambition and real-world execution will become the industry leaders.
“It’s a competitive battlefield right now, and we’re see who can close that gap the fastest.”
Calvin Hennick is a contributing writer. His work appears in BizTech, Engineering Inc., The Boston Globe Magazine and elsewhere. He is also the author of Once More to the Rodeo: A Memoir. Find him on LinkedIn.
© 2026 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.