Enterprises blow nearly a third of software budgets on upgrades they never asked for

A Censuswide survey of 2,018 senior IT decision-makers in the US and UK found organizations spent an average 28% of software budgets on vendor-driven upgrades, migrations and licensing changes; the most recent such change cost an average $12.87 million. Respondents also reported retired hardware losing 1.65 years of remaining useful life on average, and 90% said vendor-driven costs contributed at least moderately to delays, pauses or cancellations of AI or data projects. The research was commissioned by Origina, which sells independent software support, and 91% of respondents backed legislation protecting enterprise software customers.

Analysis

The strategic tension is between near-term maintenance monetization and long-term customer value capture. Forced upgrades can protect incumbent vendors’ renewal and migration revenue, but if they consume the budget and engineering capacity that would otherwise fund new applications, they may ultimately slow expansion into customers’ IT stacks and encourage selective use of independent support, open standards, or software replacement. The customer response is unlikely to be a wholesale exit: critical-system risk makes vendor dependence durable, while procurement teams may target the most expensive contracts first.

There is a second-order capacity effect for systems integrators and internal IT teams. Migration work can support services demand, but scarce engineers diverted to mandatory maintenance may delay higher-value AI and data deployments; that is a headwind to near-term realization of enterprise AI spending, not evidence that AI demand itself has weakened. Early hardware retirement may pull refresh purchases forward, but should not be read as durable hardware demand without evidence of replacement orders.

The survey is commissioned by an independent-support vendor and relies on respondent estimates; it is a signal of customer frustration, not audited evidence of industry-wide spend or a direct read-through to any named vendor. Near term, no broad software trade is justified. Over 1–3 months, watch earnings commentary for rising support exceptions, delayed deployments, or renewal concessions. Over 6–18 months, actual legislation or procurement changes could pressure incumbent maintenance economics, but the stated policy preferences are not law and the proposals’ scope and timing are uncertain. The contrarian point: lock-in may sustain revenue longer than customer dissatisfaction suggests; the more investable risk is slower incremental adoption and higher scrutiny of renewal value, not an imminent collapse in installed-base revenue.

Key Decisions for Investors

• Do not trade the software sector on this survey alone. Treat it as a watch item for enterprise software vendors including IBM, Oracle, SAP, and Microsoft—not as evidence that any one was implicated. Reassess if earnings calls show weaker renewal rates, larger support discounts, or customers moving workloads off-platform.
• Monitor independent enterprise-software support providers and third-party maintenance as a potential beneficiary category, but require verified customer wins, renewal retention, and evidence that vendor contracts permit the substitution before taking exposure; the article does not establish those economics for Origina.
• Track whether mandatory migration work is delaying customers’ AI/data projects. A sustained pattern of implementation deferrals would argue against extrapolating near-term enterprise AI software revenue, while continued project starts and stable deployment timelines would falsify that concern.
• Treat proposed customer-protection legislation as a 6–18 month policy optionality, not an imminent catalyst. Escalate the risk only if a concrete bill advances and materially limits support termination, upgrade mandates, or maintenance terms; otherwise the survey’s advocacy should not drive valuation changes.