Market & Evidence
Don't take our word for it.
Here are the numbers: how much critical code is locked in place, and what happens when people try to rewrite it. All public, all named, all sourced.
The scale of what's locked in
600–800B+
lines of COBOL still in production worldwide
50–100B
lines of PL/I locked into legacy estates
74%
of legacy modernization projects never completed (Mainframe Modernization Business Barometer)
80%
of AI initiatives abandoned before production (Gartner)
Named, sourced, public
Transformation programs in trouble
Every entry below rewrote or converted the code. That is exactly what we never do — it's why these failure modes don't apply to the platform.
| Program | Cost | Outcome | Source |
|---|---|---|---|
| US Veterans Affairs — EHR Modernization | $13.8B spent, $49.8B projected | Paused after safety and reliability failures (2018–2025) | GAO; Senate VA Committee |
| US Social Security Administration | $B-scale, ongoing | Decades-old core still in COBOL after repeated attempts | Washington Post; WIRED |
| UCLA — Financial system replacement | $132M+ | Years late, severe operational disruption | Daily Bruin; LA Times |
| California DMV — Modernization | $208M+ | Cancelled after years of missed milestones | LA Times |
| Texas Office of the Attorney General | $400M+ | Child-support system rebuild years late, over budget | Texas Tribune |
| TSB Bank (UK) — Core Banking Migration | £330M+ losses, £48.7M fine | Severe, sustained outages (2018) | UK FCA / PRA / Parliament |
| US state unemployment systems | $100Ms combined | Collapsed under pandemic load mid-modernization | US DOL OIG |