Why mainframe modernisation finally pays back
New TCO models have flipped the cost curve. The payback period is 22 months for most cores.
The TCO model has shifted
For two decades, mainframe modernisation was an exercise in faith. The numbers were always close to neutral on a five-year horizon, and CIOs would (rationally) defer the spend. That shifted in 2024 — and the gap is now wide enough that the financial case writes itself.
What changed
Three things changed at once: cloud-native databases finally caught up to MIPS-grade workloads, the talent market for mainframe operators tightened to the point where retention is unaffordable, and CDC tooling (Debezium, Kafka Connect) matured to a level where dual-write is no longer a research project.
The 22-month payback
On the seven core modernisation programmes we have shipped since 2023, the median payback period was 22 months. That includes the migration cost, the parallel-run cost, and a 15% buffer for in-flight risk. In the most favourable case it was 14 months; in the toughest, 31.
How to design the business case
Anchor the case on three lines: (1) run-cost compression year over year, (2) time-to-market for new products, and (3) talent-cost avoidance for the next decade. We have rarely seen a modernisation business case fail at the second item.
Where to start
The strangler-fig pattern. Pick the smallest domain that produces externally visible value, build it side-by-side, run it live-parallel for a meaningful window, and only then cut over. Treat any “big bang” cutover proposal as a red flag in the steering committee.
Related articles.
The strangler-fig pattern, applied to ERPs
5 minThe strangler-fig pattern, applied to ERPs
Read articleThree patterns that make AI agents production-safe
4 minThree patterns that make AI agents production-safe
Read articleFinOps after the GPU bill: what we learned
7 minFinOps after the GPU bill: what we learned
Read articleWant to discuss this?
A senior partner will respond within one business day.