The €600 Million Lesson in Saying No

Deutsche Bank just signed a 10-year deal with Thought Machine worth €600 million. The goal: take the bank's core banking systems from fifteen down to two, with €300 million in projected annual savings by 2028.

Sit with that number for a second. Fifteen. At one of the largest, best-capitalized banks in the world — an institution that can build almost anything it wants, or buy almost anything it needs — the honest answer to "how many core systems do you run?" was fifteen.

That's not a technology story. It's a governance story, and it's the same story I see at companies with three orders of magnitude less revenue and about €599.9 million less budget to fix it.

The failure isn't technical. It's structural.

Nobody sets out to run fifteen core systems. It happens one reasonable decision at a time: an acquisition that came with its own stack, a regional team that needed something faster than corporate could deliver, a "temporary" system that outlived the person who approved it, a migration that got 80% done and then quietly stopped being anyone's priority.

Each of those decisions was defensible in isolation. The fifteenth system wasn't a mistake — it was the fifteenth in a long line of individually-reasonable yeses, with nobody ever holding the job of saying no.

That's the pattern I'd bet on for Deutsche Bank, and it's the pattern I see constantly in growth-stage fintech and healthtech companies: not a single catastrophic decision, but the slow accumulation of tools, systems, and one-off processes that nobody has the standing — or the incentive — to retire.

Why "just consolidate" doesn't work

If the fix were obvious, Deutsche Bank wouldn't need a decade and €600 million to execute it. A few reasons consolidation is harder than it looks from the outside:

Every system has a sponsor. Someone championed each one, built their team's workflow around it, and will treat its retirement as a referendum on their judgment. Killing a system is rarely a technical negotiation — it's a political one.

Sunk cost gets relabeled as risk aversion. "We can't migrate off this now, it's too risky" is sometimes true and sometimes just sunk cost wearing a risk-management costume. Distinguishing the two requires someone willing to ask the uncomfortable question out loud.

Nobody owns the whole picture. Individual teams own individual systems. Almost nobody owns the total count, the total maintenance burden, or the total drag on execution speed. What isn't owned doesn't get managed.

"Just in case" is a permanent state. Every legacy system has a plausible edge case that justifies keeping it alive "just in case." Plausible edge cases are infinite. At some point someone has to decide that coverage of the last 2% isn't worth the tax on the other 98%.

The version of this at your company

You don't need fifteen core banking systems to have this problem. You need:

Two CRMs because sales never fully migrated off the old one. A reporting tool the ops team built around three years ago that nobody outside ops can access. A project management tool that engineering uses, a different one that the rest of the company uses, and a spreadsheet that everyone actually uses to keep track of both. A vendor contract nobody remembers signing that auto-renews every January.

None of these individually will sink you. Collectively, they're a tax on every new hire's onboarding, every cross-functional project, and every "why does this take so long" conversation you have with your board.

What actually fixes it

Not a framework. A person with explicit authority to say no — and the standing to make it stick.

That means naming an owner for tool and system count, not just system performance. It means treating "we're keeping this just in case" as a claim that requires evidence, not a conversation-ender. It means building retirement into the plan when you adopt something new, instead of treating consolidation as a separate initiative you'll get to eventually. And it means accepting that killing a system that someone built and defended will occasionally require a harder conversation than building a new one ever did.

Deutsche Bank is paying €600 million and committing ten years to solve a version of this problem. Most companies don't have that budget or that runway — which is exactly why the job of saying no needs to start now, while the fix is still a conversation and not a decade-long transformation program.

Julia works with growth-stage fintech and healthtech founders and product leaders as a fractional CPO and executive coach, helping them build the operational discipline to scale without the fifteen-systems problem. If this sounds like where your company is headed, get in touch.

Source: Deutsche Bank Selects Thought Machine for Core Systems Overhaul, FinTech Futures, August 28, 2026.