Risk Management & IT Strategy

The Hidden Costs of Sticking with COBOL

Jain Sekhar
Data Engineer
December 03, 2025

The world runs on code, and for a surprising chunk of the global economy, that code is over 60 years old. If you checked your bank balance, booked a flight, or filed for insurance today, there is a very high probability you interacted with COBOL.

For decades, the adage for CIOs was simple: "If it ain't broke, don't fix it." But in 2025, that logic is becoming a dangerous gamble. While the code itself is stable, the ecosystem around it is crumbling. The costs are no longer just line items on an IT budget—they are becoming existential risks.

95%
ATM Swipes
Percentage of worldwide ATM transactions powered by COBOL.
$3 Trillion
Daily Commerce
Estimated daily value processed by these legacy systems.

The 4 Hidden Costs

1. The 'Graying' Talent Cliff

The most immediate hidden cost is human. The average COBOL developer is over 55. Universities stopped teaching this decades ago, creating a massive supply-and-demand imbalance.

  • The Consultant Premium: Companies are forced to hire retired developers back at 2-3x their previous salary just to keep the lights on.
  • The Knowledge Gap: New hires spend weeks acting as "archaeologists," deciphering undocumented spaghetti code instead of shipping features.

2. The High Price of "Fragility"

Legacy systems are often described as "robust," but a better word is "brittle." One wrong line of code can cascade into catastrophic failure.

Delta Air Lines (2024)$500 Million

Aging IT infrastructure prevented a simple reboot of crew tracking systems, dragging recovery out for days.

UK Banking OutagesReputation Damage

Barclays and HSBC suffered blackouts traced back to legacy cores unable to handle 24/7 mobile volume.

3. The 'Innovation Tax'

This is the cost you don't see on a balance sheet. Every dollar spent patching a legacy system is a dollar not spent on innovation.

  • Fintech Agility: Competitors launch features in weeks. COBOL cores take 18 months just to ensure a new feature doesn't crash the mainframe.
  • The AI Barrier: Modern AI speaks Python and JSON. "Wrapping" old COBOL in APIs adds latency and kills real-time performance.

4. Cloud Incompatibility

"Lifting and shifting" a mainframe to the cloud is notoriously difficult.

Data locked in a mainframe flat-file format from 1985 is hard to analyze. You cannot easily feed 40 years of customer transaction data into a modern analytics engine, and the hardware costs for physical mainframes (cooling, floor space) are massive compared to cloud-native competitors.

The Path Forward

The fear of migrating is valid. However, sticking with the status quo is no longer an option. Successful examples prove it’s possible to modernize without a "Big Bang" disaster.

DBS Bank (Singapore)

Re-architected into a "22,000-person startup," moving away from legacy heavy-lifting to become the "World’s Best Digital Bank."

Commonwealth Bank of Australia

Successfully migrated core banking to the cloud, reducing IT expenses and enabling AI-driven personalization.

The Verdict

Sticking with COBOL feels safe because it works today. But the hidden costs—stifled innovation, inability to hire talent, and the risk of prolonged outages—are compounding interest every day.

The question for leaders is no longer:

"Can we afford to replace COBOL?"

It is quickly becoming: "Can we afford NOT to?"