Wells Fargo is spending billions on artificial intelligence and has handed the keys to Saul Van Beurden, its head of technology. The bank says AI will drive growth, cut costs, and reshape how it serves customers. The question that matters: is any of this actually working?
The answer is more complicated than the press releases suggest. But there are signs Wells Fargo is making moves that go beyond the typical corporate AI hype cycle.
The Van Beurden Strategy: Centralize, Automate, Scale
Van Beurden, who joined Wells Fargo in 2020 after a long stint at Voya Financial, has consolidated the bank’s technology operations under a single umbrella. According to Business Insider, he’s positioned as the central figure in Wells Fargo’s AI-driven growth strategy, reporting directly to CEO Charlie Scharf. That organizational structure matters. At banks where AI efforts are fragmented across business lines, projects die slow deaths in committee. Wells Fargo is betting that centralized tech leadership avoids that fate.
The bank has been explicit about the numbers. Wells Fargo spent approximately $9 billion on technology in 2024, a figure CEO Scharf has referenced in earnings calls. A significant portion of that is now directed toward AI and machine learning applications. Not all of it is generative AI — much of the spend goes toward modernizing legacy infrastructure, migrating workloads to cloud platforms, and building the data pipelines that make AI deployments possible in the first place.
That’s the unsexy part nobody wants to talk about. You can’t run large language models on mainframes from 1997.
Van Beurden has pushed the bank toward what he calls an “AI-first” approach to internal operations. Fraud detection. Document processing. Customer service routing. These aren’t flashy applications. They’re the kind of incremental automation that actually moves the needle on operating costs at a bank with over 200,000 employees.
Wells Fargo reported in its Q4 2024 earnings that noninterest expense fell to $13.9 billion for the quarter, down from $15.8 billion in the same period of 2022. The bank has attributed part of that reduction to technology-driven efficiency gains, though it’s difficult to isolate AI’s specific contribution from broader headcount reductions and branch consolidation. As reported by Reuters, the bank’s profit rose meaningfully in the period, aided by disciplined cost management.
So the cost story checks out. Partially.
Where the Skepticism Is Warranted
Here’s the problem. Every major bank is telling a version of this same story. JPMorgan Chase has over 2,000 AI and machine learning use cases in production, according to its own disclosures. Bank of America claims its virtual assistant Erica has handled more than 2 billion client interactions. Citigroup is deploying AI across its institutional client operations.
Wells Fargo isn’t leading this race. It’s trying to catch up.
The bank remains under an asset cap imposed by the Federal Reserve in 2018 following its fake-accounts scandal. That cap, which limits Wells Fargo’s balance sheet to roughly $1.95 trillion, constrains growth in ways that AI can’t fix. No amount of machine learning will persuade the Fed to lift that restriction faster. And until it’s gone, the bank’s revenue growth potential remains structurally limited compared to peers.
Van Beurden’s AI push is partly a response to this constraint. If you can’t grow the balance sheet, you grow efficiency. You squeeze more revenue per dollar of expense. It’s a rational strategy, but it’s also a strategy born of limitation, not ambition.
There’s another issue. Wells Fargo has been quieter than competitors about specific AI deployment metrics. JPMorgan publishes detailed breakdowns of its AI initiatives. Wells Fargo’s disclosures have been more general. When a company talks about AI transformation without providing granular data, industry veterans have good reason to raise an eyebrow.
I’ve covered technology in banking long enough to know the difference between a real deployment and a PowerPoint deployment. The former shows up in operating metrics. The latter shows up in investor day presentations and then quietly disappears.
To be fair, some of Wells Fargo’s AI work has produced tangible results. The bank has deployed AI-powered tools for commercial lending underwriting that have reportedly reduced processing times. Its virtual assistant, Fargo, powered by Google Cloud’s AI capabilities, launched in 2023 and has been expanded across its customer base. According to American Banker, the tool handles routine banking inquiries and has reduced call center volume for certain transaction types.
Real product. Real users. That counts for something.
But the competitive gap remains. JPMorgan spent $17.1 billion on technology in 2024 — nearly double Wells Fargo’s outlay. Scale matters in AI. More data, more compute, more engineers generally produce better models and faster iteration cycles. Wells Fargo is fighting with a smaller budget against a rival that has made technology spending a core strategic priority for over a decade under Jamie Dimon.
The Bottom Line for Industry Watchers
Wells Fargo’s AI strategy under Van Beurden is real but constrained. The bank is making genuine progress on operational efficiency through automation and machine learning. The organizational structure — centralized tech leadership with direct CEO reporting — is the right architecture for execution. And the cost reductions are showing up in earnings.
But this isn’t a bank that’s about to leapfrog its competitors through AI. The asset cap limits growth. The technology budget is smaller than key rivals. And the bank hasn’t yet demonstrated the kind of AI-driven revenue generation — new products, new markets, new pricing models — that would signal a true inflection point.
What Van Beurden is doing is competent, focused, and incremental. For a bank still rebuilding trust after years of scandal, that might be exactly the right approach. Boring works. Especially when the alternative is overpromising and underdelivering — something Wells Fargo has painful experience with.
The industry insiders who dismiss this as pure hype are wrong. But so are the analysts projecting AI-fueled transformation. The truth is in between, and it’s measured in basis points of efficiency improvement, not in breathless narratives about the future of banking.
Sometimes the most honest thing a bank can do with AI is use it to process documents faster. And that’s fine.
WebProNews is an iEntry Publication