How Banks Are Modernizing Their Data Strategy for Artificial Intelligence

So, just the idea of cloud bringing all of that together is huge, and you see that across a lot of financial institutions. Once you do that, it naturally leads to, “If we’re going to bring it together, let’s get our data strategy right.”

You start doing things like catalogs and semantic models. You ask, “How many places does this piece of data exist? Can we get rid of 10 of them and get it down to one or two?” That naturally lends itself to the Snowflake platform — whether it’s the AI Data Cloud or simply bringing disparate data together, sharing data across platforms through the cloud and getting rid of ETL, even within a bank. 

BIZTECH: Financial institutions have to prove to customers that they can be trusted with very sensitive data. How do you think about building and maintaining that trust, especially with AI in the picture?

A lot of companies are realizing you don’t get good AI in the enterprise without good data. Unfortunately, many of them have data islands everywhere. It looks like those artificial islands off Dubai — just a lot of islands.

So, how do you bring them together physically over time, and can you bring them together logically in the short term? One of the things we offer is a semantic model within Snowflake Intelligence and Cortex AI that lets you create a virtual lens across those data islands and treat them as a single data set from an AI perspective, even though they’re not physically consolidated yet.

Then, you can put security controls around that semantic model. You still have security at the raw, physical data layer, but you also get it at this logical data layer.

In one scenario, I might have access to that physical table, but from an AI perspective, I might not. So you have both short- and long-term approaches.

The long-term approach is: Bring your data into the cloud and get it clean, once and for all. The cloud and Snowflake flourish in that role.

In the short term, we provide tools to let you build this lens and get some AI value now, without having fully cleaned and consolidated all of your data yet.

Click the banner to sign up for our newsletter and receive more business IT insights.

BIZTECH: Financial institutions also face heavy regulatory and compliance requirements. How does this new approach to data and governance help on that front?

It’s the combination of data governance and compliance in the physical world.

Snowflake lets you bring together unstructured data — documents, attestations, PowerPoints. It can bring in emails that prove I wrote you and said, “I checked the box this month and did step two of this process,” which is what makes the regulator happy.

You can pull that unstructured data into the equation and put governance and security on top of it as well. You can imagine a future where a regulator comes in, and a financial institution says, “Just run our agent and call us if you have any questions.”

We’re a long way from regulators fully trusting that, but you get the idea. That’s the direction it’s heading.

DIVE DEEPER: Get help breaking down financial services data silos.

BIZTECH: Digital payments are changing very fast. What trends are you seeing right now in how money moves?

Real-time payments are a trend. Direct to bank is a trend.

The general premise is speed. Twenty years ago, T+2 was normal. With ACH, you’d post a transaction, and two days later, the money arrives. Now, you’ve got UPI, you’ve got Pix, you’ve got crypto saying, “We can move money 24 hours a day, seven days a week.”

A lot of those systems today are still small-dollar transactions. You and I could build a system in a day to move $12 between us. That’s not rocket science.

Move $12 million. Move $12 billion. Move $12 million in eight seconds instead of two days. Now your window for fraud risk screening and compliance screening has narrowed dramatically.

So you’re right back in a fraud and risk conversation, but now you’re looking at it through an AI perspective, not just rules. If I move $12 and get it wrong, no big deal. If I move $12 million and get it wrong, I’m getting fired. If I move $120 million and get it wrong, our company’s in The Wall Street Journal.

DISCOVER: Get the tech trends impacting financial services organizations in 2026.

That speed and those dollar amounts are what’s driving the broader technology set. At high dollar amounts, how do we do this safely and securely, and how do we protect our investors?

On the treasury side, you’ve got stablecoin, tokenization of money movement and trusted chains between multiple financial institutions. Some things will dominate, but it’s not clear yet which ones.

You can see blockchain trending into institutions, but not as open blockchain — as trusted chains. And by the way, I think Snowflake can play really well there. Why are we posting to a chain when we have a shared data set we can all use? To me, that’s a really simple example.

BIZTECH: As more financial organizations roll out digital payment options, what’s the biggest mistake you see them making?

I think the bigger institutions are moving too slowly, and the smaller institutions are moving too fast.

There are payment fintechs with an enormous fraud loss rate. They’re mostly used for what we call “sin MCCs” — merchant category codes tied to the “sins,” as you know. They do a lot of business, but there’s a lot of dirty money movement in that world.

On the other hand, some of the bigger financial institutions say, “I don’t trust this yet. What can we do? How do we protect ourselves?”

In the age of AI, doing nothing is the one thing you can do wrong. We don’t know exactly where it’s all going, but don’t be a spectator. That usually doesn’t end well.