ECONOMICS + ARCHITECTURE
Technology Obsolescence Management
The relevant question is not whether technology is old. It is whether keeping it destroys more economic value than changing it.
Technology obsolescence is usually managed as an inventory of versions, end-of-support dates and replacement projects. That is necessary, but incomplete. For executives, obsolescence becomes material when an asset increases operating cost, amplifies risk, slows change or blocks strategic options.
Obsolescence should be managed as a capital-allocation problem — not as a technology-age problem.
By José Ñáñez
Technology Advisor · Board Member
Published April 7, 2026 · Updated August 24, 2026 · 10 min read
THE MANAGEMENT ERROR
Age is not the decision variable
An old system is not automatically obsolete. If it remains supported, resilient, economically efficient and does not constrain required change, replacement may destroy value rather than create it.
The inverse is also true. A relatively new asset can already be economically obsolete if vendor support is ending, control costs are rising, critical skills are disappearing or the architecture makes every business change slower and more expensive.
The decision should follow economic drag and strategic constraint — not the calendar.
EXTERNAL EVIDENCE
Legacy becomes expensive long before it stops running
Public-sector evidence is not a benchmark for a bank or enterprise, but it makes the mechanism visible. In 2025, the U.S. Government Accountability Office identified 11 critical legacy systems most in need of modernization. Among them, eight used outdated languages, seven operated with known cybersecurity vulnerabilities and four had unsupported hardware or software.
Outdated languages
8/11
Known cyber vulnerabilities
7/11
Unsupported hardware / software
4/11
The same GAO review reported that those 11 systems ranged from roughly 23 to 60 years old and collectively cost about $754 million annually to operate and maintain.
Source: U.S. GAO, GAO-25-107795, July 2025.
NIST treats unsupported components as an explicit control issue: organizations should replace components when support is no longer available or implement justified risk mitigation and alternative support when replacement is not feasible.
ECONOMIC DRAG
What obsolescence actually costs
The visible technology bill is only one part of the burden. The economic case should capture the incremental cost created by keeping the current state.
DECISION EQUATION
Economic drag = excess run cost + change friction + risk & control burden + opportunity cost
Excess run cost
Premium support, scarce skills, custom maintenance, duplicated tooling, inefficient infrastructure and higher incident effort.
Change friction
Additional architecture, testing, integration and delivery effort required every time the business needs to change the legacy environment.
Risk & control burden
Compensating controls, unsupported components, remediation effort, resilience exposure and expected loss from operational or security events.
Opportunity cost
Revenue, capacity or strategic options delayed because the architecture cannot adopt new products, channels, data, cloud or AI capabilities efficiently.
A legacy asset becomes economically obsolete when the cost of preserving the current state starts consuming the capacity required to change it.
DISPOSITION
Modernization is not the default answer
A mature portfolio does not replace everything. It chooses the least expensive intervention that restores an acceptable combination of economics, risk and strategic flexibility.
01
RETAIN
Keep the asset when economics remain sound, support is viable and the architecture does not materially constrain the business.
02
CONTAIN
Keep the asset temporarily, isolate it, strengthen controls and prevent new dependencies while preparing a later transition.
03
MODERNIZE
Change selected components, interfaces, infrastructure or runtime while preserving valuable business logic.
04
REPLACE
Move to a different platform when the current asset cannot economically meet future requirements.
05
RETIRE
Remove the asset when its capability is redundant, unused or no longer worth its operating and control burden.
The correct decision is the one that minimizes total economic burden while preserving required business capability and control.
OPERATING MODEL
From technical inventory to capital agenda
The original framework remains valuable, but the sequence should be driven by evidence and economics rather than by replacement volume.
00
Amnesty and zero moment
Create a complete starting point. Teams should be able to expose unsupported assets, exceptions, hidden dependencies and deferred maintenance without turning the inventory exercise into a blame mechanism.
01
Inventory and business criticality
Map assets, dependencies, vendor support, skills, business processes, data flows and consequences of failure.
02
Quantify economic drag
Estimate excess run cost, change friction, risk and control burden, and the strategic cost of delayed capabilities.
03
Choose disposition
Retain, contain, modernize, replace or retire. The decision should be explicit and owned, not left as an indefinite technical backlog.
04
Fund the portfolio
Allocate capital according to avoided burden, risk reduction, dependency removal and strategic optionality — not only asset age.
05
Close the loop
Track benefits, decommission the legacy state, update the inventory and prevent today’s target architecture from becoming tomorrow’s unmanaged legacy.
OBSOLESCENCE REDUCTION CURVE
From current technology state to target currency
The S-curve starts from the organization’s real state. If 80% of the technology estate is current today, 20% is obsolete. The program does not start from zero: it attempts to close the gap between that current 80%, the selected target and the maximum realistically achievable level.
NORMALIZED LOGISTIC MODEL
R(t) = 1 / (1 + e^(−k(t − t0)))
H(t) = H0 + (L − H0) × [R(t) − R(0)] / [1 − R(0)]
q = (H − H0)/(L − H0) → r = R(0) + q[1 − R(0)] → t = t0 + ln[r/(1 − r)]/k
H(t) represents technology currency over time. H0 is current currency; L is the maximum achievable level; k controls transformation steepness; and t0 is the moment of maximum velocity. Normalization guarantees H(0) = H0 and makes the curve progressively approach L.
Current technology baseline (H0)
Current share of the technology estate that is not obsolete: supported, current and within the institutional standard. This is the baseline from which the program starts. Its complement is current obsolescence: 100% − H0.
Target technology currency
Technology state the organization wants to reach through the program.
Maximum achievable currency (L)
Maximum technology-currency level the organization considers realistically achievable. It does not need to be 100%.
Transformation velocity (k)
Curve-steepness parameter. A higher k concentrates correction into less time; it does not directly mean “k × 100% per month”.
Maximum-velocity point (t0)
Month when the base curve reaches its maximum slope. This is the program’s highest correction-capacity window.
Current obsolescence = 100% − H0
20.0%
Direct complement of H0 = 80.0%
Target obsolescence
5.0%
Minimum residual obsolescence
1.0%
Correctable gap
19.0 pp
Gap closed at target
78.9%
How to read k operationally
The table uses the selected H0, L and t0. k is interpreted through how long the model takes to close the central part of the gap —from 20% to 80%— and through peak improvement velocity expressed in percentage points of technology currency per month.
| k | Profile | 20% → 80% gap | Peak velocity |
|---|---|---|---|
| 0.08 | Slow transformation | 26.2 months | 0.53 pp/mo |
| 0.12 | Moderate transformation | 18.8 months | 0.71 pp/mo |
| 0.18 | Strong transformation | 13.6 months | 0.95 pp/mo |
| 0.25 | Very high transformation | 10.4 months | 1.25 pp/mo |
| 0.30 | Exceptional transformation | 8.9 months | 1.46 pp/mo |
Profiles are illustrative model calibrations, not market benchmarks. Peak velocity changes with the actual H0 → L gap and with t0.
Real-world references: context, not k benchmarks
There is no defensible public source publishing an obsolescence k for Nubank or another bank. Organizations report scope and timelines, not a complete time series of percentage of obsolescence corrected. Banco Pichincha has communicated a three-year plan to modernize and migrate more than 380 cloud applications; ANZ reported roughly 40 applications moved into a new CI/CD framework in eight months; and Nubank has published a one-year migration plan for its Colombian market. These references help size execution capacity but should not be artificially translated into k.
Time to target
20.1 mo
Peak velocity
0.95 pp/mo
t0 = 12 mo · H(t0) = 88.4%
Last +1 pp / +2 pp
1.6 / 2.9 mo
Technology currency trajectory
% technology currency
Preparation · <20% of gap closed
6.8 mo
Inventory, architecture, prioritization, budget, contracts and readiness. The organization begins closing the gap but has not yet reached its highest velocity.
Execution · 20–80% of gap closed
13.6 mo
The organization moves through the central part of the correctable gap. The slope rises, peaks around t0 and then starts slowing.
Stabilization · >80% of gap closed
0.0 mo
The hardest assets, dependencies and exceptions remain. Near L, every additional point of technology currency takes proportionally more time.
Executive interpretation
The organization starts at 80.0% technology currency, equivalent to 20.0% current obsolescence. The target is 95.0% technology currency, equivalent to 5.0% residual obsolescence. That requires closing 15.0 pp of a total correctable gap of 19.0 pp, or approximately 78.9% of all available improvement.
Preparation extends approximately to month 6.8 and the central execution zone to month 20.4. Maximum velocity occurs around month 12 at 0.95 pp/mo. The target is reached around month 20.1. Near that target, the last percentage point requires approximately 1.6 mo and the last two percentage points require 2.9 mo.
The target sits in the central execution phase. It does not yet require entering the highest marginal-time zone.
How much of the gap do we actually need to close — and how much additional time are we willing to fund to pursue the final points?
Monthly plan reference
The table translates the curve into a monitoring path. Month 0 starts exactly at the current state; subsequent rows show estimated monthly improvement, cumulative technology currency and residual obsolescence.
| Month | Phase | Monthly improvement | Technology currency | Obsolescence | Gap closed |
|---|---|---|---|---|---|
| 0 | Preparation | — | 80.00% | 20.00% | 0.0% |
| 1 | Preparation | +0.38 pp | 80.38% | 19.62% | 2.0% |
| 2 | Preparation | +0.44 pp | 80.81% | 19.19% | 4.3% |
| 3 | Preparation | +0.49 pp | 81.31% | 18.69% | 6.9% |
| 4 | Preparation | +0.56 pp | 81.87% | 18.13% | 9.8% |
| 5 | Preparation | +0.62 pp | 82.49% | 17.51% | 13.1% |
| 6 | Preparation | +0.69 pp | 83.18% | 16.82% | 16.7% |
| 7 | Execution | +0.75 pp | 83.93% | 16.07% | 20.7% |
| 8 | Execution | +0.81 pp | 84.75% | 15.25% | 25.0% |
| 9 | Execution | +0.86 pp | 85.61% | 14.39% | 29.5% |
| 10 | Execution | +0.91 pp | 86.52% | 13.48% | 34.3% |
| 11 | Execution | +0.94 pp | 87.45% | 12.55% | 39.2% |
| 12 | Execution | +0.95 pp | 88.40% | 11.60% | 44.2% |
| 13 | Execution | +0.95 pp | 89.36% | 10.64% | 49.2% |
| 14 | Execution | +0.94 pp | 90.29% | 9.71% | 54.2% |
| 15 | Execution | +0.91 pp | 91.20% | 8.80% | 58.9% |
| 16 | Execution | +0.86 pp | 92.06% | 7.94% | 63.5% |
| 17 | Execution | +0.81 pp | 92.87% | 7.13% | 67.8% |
| 18 | Execution | +0.75 pp | 93.63% | 6.37% | 71.7% |
| 19 | Execution | +0.69 pp | 94.32% | 5.68% | 75.4% |
| 20 | Execution | +0.62 pp | 94.94% | 5.06% | 78.6% |
| 21 | Stabilization | +0.56 pp | 95.50% | 4.50% | 81.6% |
The table extends through the first full month after the selected target is reached. It can be used as a baseline for plan-versus-actual monitoring.
Phases are defined over the correctable gap (L − H0), not over absolute technology currency. The 20% and 80% thresholds are illustrative operating references, not regulatory standards.
Deterministic planning model. “Technology currency” should come from a consistent institutional methodology — ideally weighted by criticality, support status, risk, cost and dependency. The S-curve represents a possible trajectory, not a forecast.
EVIDENCE DISCIPLINE
The hardest number is not modernization cost. It is the cost of keeping.
Modernization investment is usually visible because it appears in a project budget. Legacy drag is distributed across support contracts, incident teams, architecture exceptions, slower releases, control work, scarce skills and business initiatives that take longer than they should.
That asymmetry biases decisions toward inaction: the cost of change appears as one visible number, while the cost of staying is fragmented across many budgets and periods.
A credible obsolescence agenda therefore needs evidence. Every material estimate should be traceable to operating cost, incidents, delivery effort, support status, dependency data or a documented business constraint.
If the cost of keeping is not measured, “do nothing” will almost always look cheaper than it really is.
RESEARCH CONTEXT
External evidence should inform the decision — not replace institutional economics
NIST SP 800-171 Rev. 3 — Unsupported System Components ↗
Requires replacement of unsupported components or documented mitigation / alternative support when replacement is not feasible.
U.S. GAO — Critical Decades-Old Legacy Systems ↗
2025 review connecting legacy systems with operating cost, outdated languages, unsupported components and known cybersecurity vulnerabilities.
U.S. GAO — Technology Modernization Fund ↗
2026 review showing that modernization savings can be material but often arrive later than the initial investment, reinforcing the need for explicit benefit tracking.
THESIS
Obsolescence is a portfolio of economic decisions
The objective is not to eliminate old technology. It is to prevent technology from silently consuming capital, operating capacity and strategic optionality. Some assets should remain. Some should be contained. Others should be modernized, replaced or retired. The discipline is to know why — and to make the economics visible before inertia makes the decision for you.