When the Bank Becomes the Data Platform: What Gen AI Maturity Really Means for Treasury
Published: July 15, 2026
Banks are pouring money into the technology behind AI right now. According to Gartner's 2026 Data and Analytics Trends report, 79% of banking leaders plan to increase funding for data and analytics this year, and nearly one third expect that funding to jump by 25% or more over last year. A separate survey of at least 450 banking IT leaders, commissioned by the London Stock Exchange Group, found that almost nine out of 10 banks increased cloud spending over the past two years, and 91% said cloud is now central to how they support AI.
That is a good sign. It means banks have stopped treating AI as a shiny side project and started funding the unglamorous stuff underneath it: clean data, enough computing power, someone watching how the system behaves. Think of a home renovation. Anyone can hang a light fixture. Investing in the wiring behind the walls is what enables you to trust the lights will turn on. That matters if you're a corporate treasurer, managing a company's cash, payments, and financial risk, because your bank's AI capability is quickly becoming the infrastructure on which your own cash visibility and payments will run.
Why banking AI is harder than it looks
I spent 30 years running large technology and customer service operations inside Amazon, Visa, Capital One, USAA and Microsoft. One fact I learnt: investing in AI is not the same as earning people's trust in it. A system can be impressive in a lab and fall apart the moment real money and real regulation are involved.
Banking may be one of the hardest places on earth to roll out AI well; not for lack of talent, but because innovation, regulatory oversight, and operational stability must all work together, with customer trust sitting on top, the hardest part to earn and the easiest to lose. A shopping website can install a flawed recommendation feature and quietly fix it later. A bank cannot. If an AI system misjudges how much cash a company will need next week, or misroutes a payment during a stressful market moment, that mistake lands on a treasurer's desk the same day.
What this looks like for you as a treasury client
Most of the conversation about AI in banking is still built around the retail customer: a faster chatbot, a friendlier app, a helpful nudge about spending. As a treasurer, you need almost the opposite: fewer surprises, the ability to see exactly why the system did what it did, and a system comfortable saying, “I'm not sure enough to act on this” instead of guessing.
Treasury teams handle enormous amounts of financial data: cash across dozens or hundreds of accounts, multiple currencies, intercompany loans, tools to hedge currency, or rate swings. Done well, AI can genuinely help: faster cash forecasting, earlier detection of suspicious activity, better use of idle cash. But because the numbers are so large, a wrong AI decision doesn't cost you a bad recommendation, it can cost real money. A bank simply bolting AI onto an existing treasury portal, without earning your trust in it, deserves a healthy dose of scepticism.
Three questions to ask any bank about their AI
When your bank says it is investing in data and cloud infrastructure for AI, here is what should be happening behind that statement.
- Is the data trustworthy? AI is only as good as the information behind it: account records, transaction histories, market data and risk models, often sitting in systems never designed to talk to each other. A bank that has genuinely invested here can explain why its AI flagged a particular account. If it cannot, that's a red flag.
- Is the infrastructure resilient, not just powerful? Cloud gives banks the horsepower AI needs, but the real question is what happens when things go wrong. Can the bank isolate an AI issue without it spilling into your payments? That determines whether the tools still work on the day you need them most.
- Is there a human in charge? Clear rules about who can access what, and a person who signs off before anything moves money or changes your risk exposure. AI should support the bank's existing controls, never bypass them. A bank that cannot show you the path from an AI flag to a human decision isn't ready for your treasury relationship.
Trust is earned in the explanation, not the accuracy score
Regulation and operational stability can be checked against a rulebook. Trust cannot. It is built moment by moment, especially when something goes wrong. One pattern holds true everywhere: people do not lose trust in a system because it makes a mistake. They lose trust when they cannot explain the mistake, or quietly starts behaving differently without telling anyone. A treasurer will forgive an AI tool for flagging a normal payment as suspicious. What they will not forgive is a tool that cannot say why. The banks getting this right treat “can you explain this?” as just as important as “is this accurate?” Regulation, done properly, is not the enemy of trust. It is the scaffolding for it.
The bottom line
The investment banks are making in data and cloud infrastructure for AI, shown in the research above, is encouraging but a foundation is not the finish line. The real test is whether banks bring treasurers into that infrastructure with real transparency built in from day one, not added on later because someone complained. Banks that treat trust as something to be engineered, not marketed, are those with which treasury teams will choose to build their future.
Sources: Gartner, Top Data & Analytics Trends for 2026 (banking and insurance funding figures); London Stock Exchange Group / Phase 5 survey of 450+ banking IT leaders on cloud and AI adoption.