M&T Bank Just Plugged 15,000 Employees Into Enterprise AI — And It’s Not Just Some Dashboard Toy
M&T Bank is making a move that most folks in finance aren’t ready for. They’ve rolled out AI copilots to over 15,000 workers across customer service, software coding, and risk management. Yeah, seriously. After more than a decade of grinding through tech modernization, they’re finally deploying this at scale. When you ask people What is AI, they usually picture fancy chatbots — but this is full-on workflow integration in a multi-billion-dollar bank. No joke, this is a total shift from pilot projects to the real deal.
Here’s the thing: M&T didn’t just buy a SaaS license and call it a day. They spent years rebuilding their tech backbone — data platforms, cloud migration, legacy system cleanup — before they even thought about rolling out AI. The bank is now building its own specialized models on top of major AI Models, so they’re not just running whatever comes packaged with a vendor. That kind of investment? Totally different from what most mid-size banks are doing right now. Some are still figuring out how to get people to use basic automation tools.
The cost side of this is gonna be wild to watch. Once you’re running copilots for 15,000 people, understanding how AI Tokens and inference costs scale becomes a massive operational question. If M&T nails this — if they can prove the ROI on custom model development versus just paying per-token for commercial APIs — every other regional bank is gonna scramble to do the same thing. The infrastructure bet alone could separate winners from laggards over the next few years. Kinda crazy to think about what five years looks like if more banks take this route instead of just bolting AI onto outdated systems.
- Bigger than pilot season: M&T went from testing AI tools to deploying them company-wide for 15,000+ workers. Most banks are still stuck in proof-of-concept mode. This shows what happens when you actually commit.
- Build vs. buy is getting real: Instead of relying purely on commercial AI products, M&T is investing in internal model development. That’s a bold stance — and probably way cheaper at scale than burning through API credits forever.
- The debt is the moat: A decade of tech modernization came first. Cloud infrastructure, clean data, stable systems — that stuff had to be in place before AI could actually work. Banks that skipped those steps are gonna hit walls real fast.