According to Nvidia’s new report on AI adoption in financial services, 52 percent of professionals in that sector are now leveraging generative AI tools in their work. That’s up from 40 percent in 2024.
Kevin Levitt says that increase, and AI adoption more broadly, “is all part of the next industrial revolution. Historically, 150 years ago, it was about taking water in and generating power out and, ultimately, producing goods. Now it’s about taking data in and manufacturing intelligence out of AI factories.” Levitt leads global business development in the financial services sector for Nvidia, the Silicon Valley tech company.
We spoke with Levitt in February, just after the release of Nvidia’s 2025 “State of AI in Financial Services” report to learn more about where AI is headed in this key sector.
• • •
What brought about this survey, and how do you hope it will help organizations?
Nvidia sits in a unique position. All aspects of financial services companies that are necessary to build and deploy AI models participate in Nvidia’s accelerated computing platform, from the companies themselves and their developers and data scientists to third parties that help them, such as independent software vendors, global services, service integrators, the CSPs [cloud service providers] and the OEMs [original equipment manufacturers].
We wanted to take what is our global perspective and enable the industry to have a similar lens that ultimately brings focus to how financial services firms are investing in AI, how they’re developing their AI strategies, how they’re prioritizing which use cases and workloads to deliver and develop first. Then use that insight to help both the innovators, as well as those that might be more of a laggard when it comes to leveraging AI, move through their journey as quickly as possible.
To give some perspective, who tends to be the one driving progress when it comes to AI and financial services, whether that’s someone in the organization itself, or one of those third parties that you mentioned?
It’s cross-functional, like any technology that brings about disruptive innovation and opportunity. Every function across a bank, a payments firm, an insurance company, is involved in maximizing how artificial intelligence can create competitive advantage.
That means it’s everything from the front office and teams that are responsible for acquiring new customers to the teams that are responsible for underwriting loan applications, to teams that are responsible for servicing these new customers and getting them onboarded, to fraud teams that are responsible for identity verification and monitoring transaction flow after they’ve become a customer. Those teams are responsible for understanding risk management and identifying which customers might be in peril in terms of their financial journey and which ones are best suited towards new products and services from the bank.
All of this is enabled by AI, and then what it means is that banks will be able to acquire and retain and service their customers faster, cheaper and more effectively.
I was looking at the top generative AI use cases by ROI for financial services, and I noticed that trading and portfolio optimization takes the lead, followed by things like report generation and customer experience. Can you provide some examples of how these use cases change the scope of financial service operations, potentially for the better?
In financial services specifically, we’re generating intelligence which helps banks, insurance companies, payments firms, et cetera, do what they do best, which is taking and using data to generate insights about individuals, about corporations so they can better manage risk, acquire and service customers and so forth.
Where the innovation really takes place is how banks acquire and service customers… [And] that is predicated on data and artificial intelligence…
That’s why AI is such an exciting opportunity in the realm of financial services… There’s not tremendous product innovation when it comes to the development of a new type of mortgage or a new type of credit card. There’s only a certain set of innovation that has happened. Where the innovation really takes place is how banks acquire and service customers as it relates to those products. And all that is predicated on data and artificial intelligence, and now generative AI is helping banks use data more effectively so that they can drive top-line revenue and eliminate costs from their workflows and services.
You mentioned the innovation factor here is generating intelligence. With that in mind, what are some examples of other use cases that you found organizations have yet to tap into on a larger scale that you think would be a real game changer for the industry?
We’re going to see a lot of investment from financial services firms this year in agentic AI, which is about accomplishing specific tasks versus generating intelligence. In some cases, we’ll see multiple agents working together to complete a task. We’re expecting a lot of investment in agentic AI this year and in the years to come.
From a generative AI standpoint, we’re expecting more investment and leveraging of gen AI for synthetic data generation to support the identification of fraudulent activities, because these fraud instances are anomalies. They’re needles in a haystack, and the more needles that we can generate through some depth and synthetic data generation, the better we can train the models to proactively identify them.
We’ll see more use cases for transformers in the context of understanding transactions, predicting the next transaction and making recommendations.
We’re also going to see the use of transformers, which have been applied to unstructured data, such as text to tabular or structured data. We’ll see more use cases for transformers in the context of understanding transactions, predicting the next transaction and making recommendations.
You mentioned agentic AI, and I know that the report tells us that AI agents will continue to flourish in financial services. Meanwhile, so-called “digital avatar and digital human technology,” which I understand to mean more human-like AI in place of humans, is getting better and easier to deploy. How do you anticipate these financial services organizations will strike a balance between tech and human-led operations?
There’s going to be tremendous focus on maintaining a human in the loop, so that as the AI is summarizing reports and summarizing information, generating reports and making predictions, there’s always a partner from the human side making final evaluations and implementations of certain recommendations that are coming from the AI.
All of this is to say that it’s going to allow our most valuable resources, which is our people within the organizations, to focus on higher order, more complex problems, while the AI will be able to handle the more routine and rote tasks that make every function across the bank operate more efficiently and effectively, and give the banks the opportunity to deliver enhanced customer service to build new revenue streams and products and services through all the talented people that they employ today and will continue to employ in the future.
And that leads into my next question. One of the most significant trends the report cites is the increased focus on opening new business opportunities and driving revenue, which rose from 17 percent to 24 percent year over year. How is AI reshaping the revenue generating capabilities and broadening the potential for tapped markets in financial services?
There’s been a lot of talk about not just personalization within financial services, but hyper-personalization, one-to-one messaging. It’s not just in the context of the actual content, but perhaps the imagery and understanding specifically given to an individual customer’s financial journey. What makes sense for them in terms of next best action?
This is not unlike when the internet and then mobile advertising came around, and we became more effective at targeting and servicing customers online through advertising or services like online bill pay. Now with AI, we’re going to be able to enhance every step of that customer journey, such as when a bank presents a recommendation for a new product or service through its recommendation system.
We’ve had banks present at our GTC Technology Conference, showing that an AI-powered recommendation system led to a 60 percent improvement in click-to-conversion rate for new products with existing customers. What that demonstrates is that there’s a tremendous gap between the technologies of yesterday and AI, and their ability to identify new revenue opportunities, and that’s why banks are investing so heavily in artificial intelligence. It’s because, as it shows in the report, the vast majority of banks, over two thirds, are experiencing meaningful ROI from their investments in AI.
What kinds of recommendations is AI making at this point? Because at first glance, I imagine that there are major regulatory issues in recommending investments for people’s retirement accounts.
We’re seeing a lot of recommendation systems operating in partnership with the advisor, so they’re able to take a more holistic look at a given individual’s situation from a wealth management perspective, perhaps from an insurance perspective, as well as other facets and attributes of an individual’s financial health. These agents and AI are able to make recommendations to the advisor, and then the advisor can evaluate them and finalize the recommendation before presenting it to the customer.
We’re seeing a lot of use cases in those wealth management scenarios. We also see them in the context of pure customer service. The call center agent is being assisted with recommendations from an AI that is able to identify sentiment and understand the specific inquiry, and is able to read through documentation and extract the most relevant pieces of information necessary to serve the customer and answer that question in a more timely and effective manner, all of this designed to increase customer satisfaction and, ultimately, retention.
We’ve talked a lot about the opportunities. What are some notable barriers in AI adoption and financial services, and how are the best companies getting past them?
When we talked to the financial services ecosystem about challenges in getting to production with AI, the largest barriers that existed last year came down significantly, notably data issues and privacy concerns. Last year, 49 percent of respondents cited that as a challenge to building and deploying AI applications and financial services; this year, only 33 percent of respondents cited data issues and privacy concerns.
The same went for insufficient data sizes for model training, where that fell from 49 percent to 31 percent year over year. A lack of an AI budget fell by more than half from 31 percent down to 15 percent.
What this indicates is that financial services firms have identified, in partnership with internal regulatory and governance partners, how to effectively identify which data they can use reliably and safely for AI model training, and which use cases have shown the most promise and opportunity for delivering a positive ROI. Banks are astute at where to invest their capital, and they see that these initial forays and artificial intelligence are delivering real returns. That’s why the lack of AI budget is less of a barrier by more than half this year, because there’s such momentum behind opportunities to deploy capital where there’s real ROI, and AI is clearly one of those use cases.
It’s interesting that you use the word momentum, because sometimes I think in metaphors, and I’m quite literally visualizing someone starting a run at a slower pace, and then they’re figuring their stride out and moving faster. It seems like that’s what’s going on here, in regards to the use of AI that companies are really ready to start making impactful change. Is that what you’re seeing as well?
You’re exactly right, and that’s why so many of our customers today are investing in building AI factories within their financial enterprises. …When you have this AI factory that sits horizontally across the bank and those lines of business, then the data science teams have the tooling from the infrastructure up through the software layers to actually build and deploy AI-enabled applications at scale.
Banks today are operating against hundreds of use cases. It will be thousands into the future. They’re really expert at using data, they’re investing in AI factories that give them the opportunity to build and deploy these innovative models, and the returns are real, and so that’s why we’re seeing all this momentum, and the industry really hitting its stride, to fall on your metaphor.
I’m curious if there are any highlights or data points from the study that surprised you, or anything that may shift the landscape for business leaders sparking technological change in their organizations?
The biggest one would certainly be the [decreasing] challenges of building and deploying AI. We didn’t expect it to come down so dramatically, so quickly.
I think the other area of import is the ways in which companies continue to find value from deploying AI across their enterprise. It’s leading to not just operational efficiencies. It’s creating meaningful competitive advantage. It’s improving the customer experience. It’s resulting in greater employee productivity, and it’s opening new business opportunities. Seeing that continued validation for the way in which AI is impacting financial services is giving companies the confidence to continue investing, for those that have been behind, to look externally to partners across the ecosystem that can help them catch up.
Is there anything else you’d add to paint a clear picture of where AI in financial services is headed?
I think we’re going to continue to see the investment in artificial intelligence to fight fraud, and that’s both at the identity verification layer, especially to do a better job of adhering to anti-money laundering and Know Your Customer compliance and regulatory requirements, as well as from a transaction fraud standpoint, doing a better job of reducing false positives, truly identifying the bad actors when it comes to transactions, as well as using AI to fight cyber crime.
We will see a lot of investment in agentic AI to create operational excellence, and last but not least is the continued investment in AI factories to support all of these use cases across the financial services enterprises.
