How AI Is Really Creating Value in Financial Markets | Adena Friedman

Yeah. Well, I do see a J curve, more like an L, a backward L. So it doesn't actually go down. I think that it takes time and investment. I'll say why, to make it so you really get the benefit on the productivity side. And I think it's because first, the technology itself is always going to be ahead of its applications and of its implementation.

And it's an amazing thing, and new tools are being born every single day to make it so you could bring that into the enterprise use cases. But if you're looking at it from a business perspective, someone who wants to use this technology to create productivity, first they have to make sure it's safe and secure. They have to make sure it creates predictable outcomes. And then they also have to, in some cases, particularly the financial industry, make sure that the models are explainable to regulators. So there's kind of this overall governance that has to be set up. The second thing is then bringing the tools in in a way that integrates with your own infrastructure. And some businesses, I mean, I have to say, Nasdaq, we've invested a lot in moving all of our internal infrastructure to cloud and onto modern stacks, but not every...

You were born that way, and we appreciate that. But not a lot of companies are still in that journey. So until your data is in a modern state, until your code, for instance, if you're trying to automate product development, you have to have your code repositories in a very, very good state, you have to have all the documentation of your internal code in a great state in order for it to really be able to automate the coding. So there's those types of infrastructure investments that people are making, but then also I think you also want to make it so that it's integrable into everything else you have. So the more that you can work with partners to do that, I think as opposed to taking every new little thing that's coming out to market, I think you'll find for enterprise use cases, there'll be some real winners of new applications and then there'll be others where they'll take them through the infrastructure providers already using.

Yeah. Well, I think there are so many use cases, but what we've been really focused on is where can we scale it the best. And product development is a huge scalable use case in our business. In our product base, we also provide a lot of software to the financial industry. So we provide software to 3,800 banks or brokers around the world. The fraud side, like really, really amplifying fraud capabilities, automating a lot of the anti-financial crime workloads, automating regulatory reporting, those are very obvious and great scalable use cases. So we're building digital workers into our fraud solutions, we're building digital workers into our regulatory reporting solutions, but then we're also building digital workers to automate coding. And that is a huge use case for us because about a half of our people are in the product and development area of our business.

Yeah, so we own an incredible solution called Verafin, which basically takes data from across 2,700 banks to be able to identify fraud more effectively than pretty much any other provider, because we've become a giant 11 trillion dollar bank, if you think about it in terms of all the assets under custody of all the banks combined that use our solution. We process 1.8 billion transactions a week. We evaluate all those transactions. Any sort of payment is the truth. That data is an incredible asset because it allows us to look at information like, within one bank, they can't see anything that's happening outside their own bank. But with our solution, you can then understand what's happening across the entire financial system. Look for patterns and practices of behaviors of criminal actors. Make sure you can identify them more effectively. You can also take down false positives because you can basically tell the bank, no, that's a good account. We know that's a good account, right? So finding more fraud, taking down false positives, huge part of our value proposition within Verafin.

All of those things. So that is on the algorithmic side of machine learning, and really doing basically Bayesian models. We use computer vision for check fraud, things like that. But then on the GenAI side, it's really on workflow automation, because once you generate an alert, you have to do a whole bunch of workflow around that. You have to research the entity, you have to figure out whether or not you need to report to the regulator, then you need to build a regulatory report. We've automated all of that. So it's a really, really great solution.

I mean, I think every industry is at the beginning of time when it really comes to the potential capabilities of agentic AI. But what we've been able to implement over the last couple of years are three areas that we're focused on. One is within Verafin, on the digital workers, taking an alert, turning it into researching that entity from the alert, completing that research, writing the report and preparing it for the regulators. So all the analyst has to do is review and submit. That is an agentic capability right there. Inside of Nasdaq, we have two areas. One is in our tech ops space. So where we monitor all of our hardware, all of our infrastructure, we've built agentic AI, agentic workers that can do much more deep monitoring and then can take action against the monitoring. So if they see certain anomalous behaviors, the agents can actually take action against those anomalies. So that's just the beginning of what we can do to really build even more automation into our tech ops.

Well, so I look at AI as a long-term transformational technology, and everything that we're investing in right now is to build the infrastructure to allow that technology to thrive and become as great as it could be. And similar to when the railroads were created or even when the internet was created, you had to build an enormous amount of infrastructure to basically move the economy onto a whole new paradigm. And that's what's happening in the AI space. I think that there will, as with any sort of new technological creation, there's going to be winners and losers, but the general trend is this is infrastructure that is going to be needed to meet a potential that we can't even understand yet of what this technology is capable of delivering. And so to us, you have to look at it as a marathon, not a sprint. You have to realize that these capital investments are going to lay a foundation for the next decades of productivity gains. And so to us, these are incredible investments to make.

Yeah. And also realize that in this particular case also the companies that are underwriting these investments are extremely well capitalized. If you look at just the amount of money that the hyperscalers have invested in this technology over the last year, it's about 60% of their annual cash flows, meaning they are giant cash creators and generators. So they're investing it, and then also they have a whole ecosystem of investors who specialize in real estate, specialize in energy, specialize in these areas where they can deploy their capital as well. So it's a very well capitalized investment ecosystem, but it is a big investment.

It is. Our financial industry and our technology industry have become these assets that really have differentiated the United States.

Thanks, Eric. It's great to see you. Thank you.