For 2025, most technology decision-makers expected their budgets to rise—but often by too little to fund every priority. The practical response was to reallocate: protect essential skills, data, security and modernization work; find money in cloud and software sprawl; and release funding for uncertain initiatives in stages. These were expectations recorded in 2024 planning coverage, not a forecast for 2026 or proof of what organizations ultimately spent in 2025.
What “moderate increases” meant for 2025
Forrester’s 2025 planning material, as reported in CIO’s August 22, 2024 feature, said 91% of surveyed global technology decision-makers expected their IT budgets to increase. That headline did not mean most organizations had ample discretionary cash: approximately four in ten expected growth below 5%, a similar share expected 5% to 10%, about 8% expected more than 10%, and roughly 9% anticipated a flat budget or a slight decline. These are the article’s account of Forrester’s planning figures, not measured 2025 outcomes. CIO’s 2025 budget-planning coverage
A nominal increase can still mean less purchasing power. The CIO article cited a 3.3% inflation projection for 2025 from the July 2024 World Economic Outlook; that was a contemporaneous planning assumption, not a statement of the inflation rate ultimately realized. Wage pressure, software price changes, expanding cloud consumption and new AI-related costs could also absorb budget growth. The useful question for a CIO was therefore not simply how much the budget rose, but what costs would grow automatically and what business outcomes the remaining funds could buy.
Where CIOs expected to protect or increase investment
Skills and operating capacity
Personnel represented nearly 35% of IT budgets in the Forrester figures cited by CIO; category definitions can differ between organizations. Training was a priority, particularly for AI and other strategic technologies. Buying software without the people and operating practices to use it well can add complexity rather than capability.
#1 Best Overall
Budget plans should distinguish hiring, reskilling, contractors and managed services. They should also account for the capabilities behind production technology: data stewardship, platform engineering, security, model governance and change management. AI training is not just instruction in prompting; teams need to know how data is accessed, how outputs are assessed and who is accountable when a system is used in a business process.
Software, platforms and automation
Software accounted for about 21% of IT budgets in the CIO article’s account of Forrester data. Forrester projected software spending to grow at a 10.5% compound annual rate through 2027. That was a forecast, not an observed growth rate, and it implied a funding trade-off: if software grew faster than the total IT budget, other spending would need to shrink, become more efficient or grow more slowly. Forrester’s 2025 technology executive budget guide
Before adding another application, CIOs can check whether existing products overlap, whether purchased seats are being used and whether a wider platform would actually lower total cost. A platform replacement should include migration, integration, security, administration, support and exit costs—not just its subscription price. Applications with no active owner may remain in place simply because no one has taken responsibility for reviewing or retiring them.
AI and data foundations
Forrester reported that 92% of technology decision-makers planned to increase budgets for data management and AI. Its guidance emphasized governed, high-quality data; scalable architecture; security; skills; knowledge management; and AI governance. Forrester’s executive summary
Recommended Free Tools
AI investment is broader than model access or a subscription. A credible budget may need to cover:
- Data cleansing, cataloging, access controls and lineage.
- Model and vendor evaluation, security and privacy reviews, and legal, compliance and procurement work.
- Integration into existing applications and workflows, plus monitoring and evaluation after launch.
- Training, change management and a clear operating owner.
- Usage, compute and inference-cost controls, with a plan to contain or retire pilots that do not work.
These costs should be assessed against a specific use case and workload rather than hidden in a general “AI” line item.
Modernization and technical debt
Outdated systems and technical debt can increase maintenance costs, constrain delivery and create resilience or security risks. Forrester’s 2025 planning guidance described leading organizations as allocating 10% to 30% of their budgets to technical-debt reduction. That is an observation about leading organizations, not a universal target. Forrester’s executive summary
Prioritize work by its effect on business-critical services, outage or compliance exposure, maintenance burden, delivery speed and ability to support revenue or strategic capabilities. Refactoring is not valuable merely because a system is old; the investment needs a credible link to lower risk, lower cost or better outcomes.
Rank #3
Where to recover funding without creating new risk
Review cloud consumption as an operating process
Forrester advised identifying redundant and unnecessary cloud services; CIO’s article described cloud sprawl as a consequence of rapid adoption without a mature strategy. A practical review starts with ownership and visibility, not an across-the-board spending cut:
- Inventory accounts, subscriptions, regions, clusters, databases and storage, and assign each material cost to an application or business owner.
- Look for idle, oversized, duplicated and non-production resources. Review storage-retention and data-transfer charges as well as compute.
- Compare on-demand usage with committed or reserved capacity where workload patterns and contract terms justify it.
- Set budgets and alerts, then track whether owners act on them; alerts alone do not reduce costs.
- Measure AI workloads separately and define placement rules across public cloud, colocation and on-premises environments.
Moving a workload back on-premises is not automatically cheaper. Utilization, latency, compliance, resilience, internal skills, migration costs and existing commitments all affect the decision. Cost reductions also need reliability guardrails: indiscriminate rightsizing or shutdowns can impair production or disaster recovery.
Automate only after examining the work
Forrester recommended aggressive automation of manual processes, while CIO’s article described Freshworks’ focus on automation and AI to reduce employee workload and complexity. The useful first distinction is whether work is repetitive and rules-based, requires human judgment, or should be eliminated or redesigned before any tool is applied. Automating an undocumented or broken process can scale its errors and create more review work than it removes.
Assess software and virtualization contracts before cutting or switching
The CIO article and Forrester pointed to concerns about VMware licensing and contracts after price increases associated with Broadcom’s acquisition, and Forrester recommended evaluating alternatives. The sources do not establish one uniform increase applicable to every contract, edition or region. Nor does a lower license price by itself prove that a migration will save money.
Compare renegotiation, rightsizing and a phased exit against the full transition burden: retraining, hardware compatibility, application certification, backup and disaster-recovery changes, management tooling, contract termination terms and operational risk. A migration can make sense, but the business case should include temporary duplication and the cost of supporting the destination platform.
Why smaller, staged investments can beat a big-bang program
Forrester analyst Christopher Gilchrist told CIO that organizations were likely to divide long transformation programs into smaller, temporary strategic efforts rather than fund one sweeping five-year initiative. The point is not to abandon a long-term destination. It is to release money in evidence-based stages while preserving shared architectural direction.
- Choose a bounded capability. Identify a business problem and the users, process and systems in scope.
- Set the baseline and target. Name the business owner, current performance, intended result and measurement period.
- Authorize a limited discovery or pilot. Set a time limit and budget ceiling, along with security, data and architecture requirements.
- Make a scale-or-stop decision. Use evidence against the target and specify the decision date before the pilot starts.
- Reuse what works. Carry successful components into the next increment under shared standards for data, identity, integration, security and observability.
These controls help avoid two opposite failures: committing to a large program before its value is demonstrated, and accumulating disconnected pilots that become another layer of technical sprawl.
Budget AI as a portfolio, not a blanket category
Each proposed AI initiative should be classified by maturity and evidence. The goal is to fund foundations and valuable use cases while keeping speculative work reversible.
Best Value
Increase or defend
- Data quality, access governance and security that support multiple use cases.
- Integration, evaluation and monitoring needed to operate systems in production.
- Use cases tied to a measurable revenue, productivity, service-quality or risk outcome.
- Training and workflow changes that help employees adopt useful tools safely.
Pilot with a defined ceiling
Forrester’s architecture and delivery guidance identified areas such as AI agents, specialized language models, API developer portals, AI PCs, AI cost management and edge intelligence for investment or experimentation. A pilot should have a named user and process, a data and governance path, a budget ceiling, a success measure and a production decision date. Forrester’s architecture and delivery guidance
Deprioritize or stop
- Generic demonstrations with no accountable owner, defined process or success metric.
- Duplicate copilots or overlapping pilots bought by different departments without a shared review.
- Projects with no workable route to secure data access, integration or production support.
- Tools that create more checking and exception handling than useful work removed.
Four budget postures, not one universal prescription
Individual company examples in CIO’s reporting illustrate different choices; they are executive perspectives, not representative market data.
- Invest for growth: SnapLogic’s CTO argued that organizations pursuing growth need to spend above the industry average and described AI opportunities across finance, sales, marketing and HR workflows. That perspective does not establish that higher spending will produce the same return elsewhere.
- Prioritize productivity: Freshworks emphasized automation and AI as ways to reduce employee workload and complexity.
- Expand selectively: Counslr anticipated a possible increase of up to 25%, an individual company expectation that illustrates how business model and circumstances can depart from the broad pattern.
- Maintain and improve: Barco ClickShare expected no increase because near-term initiatives were already planned and focused on incremental improvements to meeting spaces and collaboration.
The relevant posture depends on whether a company is funding growth, protecting service levels, improving productivity or preserving options amid uncertainty. An example from one firm is not a budget benchmark for another.
A decision test for the next budget meeting
For every material initiative, identify the business owner, baseline, target, measurement period, total cost, budget ceiling and scale-or-stop date. Include implementation, integration, training, security, support and exit costs. Then decide whether to increase, defend, reduce, pilot or stop:
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errors| Decision | Use it when | Test before approval |
|---|---|---|
| Increase | The initiative has a credible path to measurable growth, productivity, service quality or resilience. | Is there a baseline, an accountable owner and a fully costed operating plan? |
| Defend | Underfunding would expose a business-critical service or capability to unacceptable risk. | What specific failure, disruption or strategic constraint would reduced funding create? |
| Reduce | Spend is unused, duplicated or poorly aligned with business value. | Will the cut affect critical workflows, reliability, security or create shadow IT? |
| Pilot | Potential value is plausible but key assumptions remain unproven. | Can the test be bounded, reversible and judged against a target on a set date? |
| Stop | No owner, measurable outcome or credible route to production can be identified. | What evidence would justify continuing, and has the initiative met it? |
Apply the same scrutiny to trade-offs: AI versus foundational data work, cloud flexibility versus consumption cost, automation versus workforce and service changes, modernization versus continuity, and vendor consolidation versus dependency risk. Incremental funding works best when those choices remain connected to one long-term architecture and business strategy.
What the 2025 planning picture does—and does not—show
The figures and company examples describe planning expectations reported in 2024 for fiscal-year 2025. They do not establish what organizations ultimately spent, what inflation or vendor costs turned out to be, or which investments delivered returns. Their durable lesson is about budget discipline: modest growth is an allocation problem. CIOs have to choose what outcomes merit new money, recover avoidable costs without undermining reliability, and require evidence before uncertain initiatives scale.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
