The 94% Problem: Why Core Banking Needs AI and Trust

Banks have been chasing core modernization for more than a decade. But IBM’s latest report, The 94% Core Banking Problem (co-authored by Paolo Sironi), delivers a striking number: 94% of banking modernization projects run late, over budget, or fail to deliver real business value.

It’s not just a question of efficiency. If banks keep modernizing without rethinking their models, they risk becoming utilities. Essential, but interchangeable.

This is where AI enters the picture. Not as another cost-cutting tool, but as the catalyst to redefine how banks create value, compete with BigTech, and preserve trust.

Why modernizations fail

Despite massive investment, modernization efforts often stall. The main culprits?

• Hidden dependencies in legacy systems that slow migrations.
• Security demands that add cost midstream.
• Talent shortages in cloud and AI expertise.

The result: more than half of CIOs saw little or no benefit from their modernization spend.

The lesson: swapping out technology isn’t enough. Banks need to rethink how they operate, not just what systems they run on.

AI: Accelerator and risk

According to IBM, 92% of banks plan to use AI to accelerate development by 2028. From decoding decades of legacy code to automating documentation, AI can turn “hidden knowledge” into explicit insights and speed up modernization.

But there’s a catch. Only a third of CIOs are embedding risk controls when scaling AI. Without trust, faster change just means faster fragility.

The opportunity: AI can reshape value creation, but governance and transparency must come first.

Where banks are starting

The first areas of modernization are telling:

• Customer onboarding (47%)
• KYC/AML compliance (44%)
• Fraud detection (42%)
• Payments (40%)

These are domains where data, trust, and client experience intersect. They’re also the areas where visibility into ownership structures, compliance signals, and financial ecosystems makes a tangible difference.

That’s why open, connected data is becoming a strategic enabler: not as a new “system of record,” but as the connective tissue that allows banks to modernize confidently.

Scaling AI across the enterprise

Banks are shifting from isolated AI pilots to enterprise-wide strategies. Governance models are evolving from centralized “AI labs” to hybrid approaches that combine strategy with local adaptation.

And on the horizon: agentic AI. Already, 42% of banks are piloting, and 17% plan to go live by 2026. This leap toward autonomous decisioning will test banks’ ability to balance speed with accountability.

The question: will banks use AI to reinforce trust and differentiation or let it push them closer to utility status?

Lessons CIOs are learning

From IBM’s survey of 700 executives, four lessons stand out:

• Get the fundamentals right: Cloud and AI, with realistic expectations.
• Speak a common language: Align business and tech around outcomes.
• Move in increments: Avoid risky “big bang” overhauls.
• Make trust non-negotiable: Embed risk management by design.

The bigger picture

AI is no longer just about efficiency. It is redefining the very fabric of banking, forcing institutions to rethink how they create value, build trust, and deliver human advisory in an AI-first world.

The 94% statistic is a warning. But it’s also an opportunity. The banks that combine modular architectures, AI acceleration, and trust by design will thrive.

For us at openthebox, the report resonates deeply: many of the pain points it highlights—data silos, compliance complexity, lack of visibility—are the same frictions we see across the financial ecosystem. Addressing them isn’t just about modernizing systems, but about enabling better decisions, earlier signals, and more resilient growth. Wondering how you compare? Take the quick survey to find out.

And the banks that embrace that shift? They won’t be utilities. They’ll be leaders.