In today’s climate, business leaders no longer ask whether they should adopt AI because that’s already happening.
Across industries we’re seeing organizations that are integrating AI tools into their processes achieving notable benefits, such as enhanced efficiency and a variety of improved business outcomes.
EMEA Field CTO at Apptio, an IBM Company.
For many organisations, the real dilemma is balance: how to stay competitive in the AI era while managing costs, resourcing and intellectual property.
Frequently, AI-driven projects are initiated from the business units themselves, and they work with IT to deliver it.
Yet many still lack the information needed to evaluate technology spend decisions properly. That gap between growing costs and unclear ROI is now shaping boardroom debates across industries.
The conversation has shifted from “what can AI do?” to “what value is it delivering, and at what cost?” to “are we trying to use AI on the right kinds of projects?”
Managing trade-offs without cutting muscle
Scaling AI requires careful trade-offs, not just in trimming budgets, but in deciding where to reallocate resources without disrupting core business operations.
The key to making these decisions lies in achieving visibility. Many organizations rely on ROI as a guiding metric for investment decisions or cost-benefit analysis.
However, these metrics often operate in silos, communicated differently across finance, IT, and operations. As a result, many organizations stop evaluating ROI once a project is underway, making it difficult to accurately track and realize the full value of AI investments.
A single taxonomy and shared data source is essential. Otherwise, leaders can end up talking past each other: finance worrying about capital expenditure versus operational while IT measures utilization rates and uptime.
When measuring the value delivered by AI, it needs to be translated into business metrics that show the cost against the business outcomes being achieved.
With generative AI workloads notoriously compute and energy-hungry, forecasting spend precisely is already a challenge. Businesses need a unified view to decide where to cut, where to double down, and how to ensure AI projects align with strategic goals.
Counting the real costs of AI
Unlike past technology rollouts, AI is not a simple one-time capital investment. Data from Apptio shows that over 90% of organizations expected technology budgets to rise this year, with AI one of the most significant new spend drivers and I expect we’ll see similar sentiments going into 2026.
It brings continuous costs in IT infrastructure, energy, people, and processes. Training models and running inference requires massive compute power, often hosted in energy intensive data centers. Specialist AI talent is scarce and expensive. And all the while, boards are asking how these outlays translate into measurable ROI.
AI serves diverse functions within organizations, including data analysis, process automation and fraud detection or cybersecurity. While these are highly impactful applications, scaling them requires absolute clarity about costs and benefits. Leaders need to distinguish between the spend involved in training large foundation models versus embedding third-party services into existing processes.
Here, Technology Business Management (TBM) frameworks can help. By linking IT spend directly to business outcomes, leaders can spot waste, prioritize high-value projects, and prevent AI from repeating the same overspend patterns many businesses encounter with the cloud.
Rethinking Data
Where data should live has become one of the most pressing questions when scaling AI projects. Boards are increasingly nervous about intellectual property loss, regulatory compliance, and the risks of feeding sensitive datasets into third-party systems.
The cloud remains indispensable for scalability, but there is growing recognition that not every workload belongs there. Some companies are pulling specific processes back on-premises to regain predictability, strengthen compliance, and control long-term costs.
This isn’t about turning away from the cloud; it’s about using it more strategically. A hybrid approach – balancing cloud agility with on-premises control – is fast becoming the default.
Final takeaways for business leaders
AI is here for the long haul, but success depends on treating it with the same discipline as any other strategic investment. Four principles stand out:
1. Prioritizing visibility: If businesses don’t take stock of how many investments are being made and how projects are performing, AI spend can increase but with limited ROI.
2. Taking a hybrid approach: Cloud strategies are not a one size fits all; by looking at hybrid models, IT teams can still benefit from scalability but also secure and retain control over data.
3. Staying on top of costs: Implementing AI is not a static cost; there are many elements that need to be continuously monitored and reviewed. By factoring these in early, businesses can better control spend.
4. Looking at the full picture: Teams must ask themselves, are we linking IT investments to measurable business outcomes? Value isn’t just money saved, its increased productivity, better decision making and customer outcomes. It’s important that business leaders look at the full spectrum when measuring success and reward initiatives that do deliver.
The goal of any technology leader is to make tech investment decisions that deliver value and help support wider business objectives. Whether AI, cloud, or any other innovation, that goal never changes.
With many more innovative projects expected, technology, business and finance leaders will need to partner closely to prove value and increase internal expertise.
On the road to AI ROI, the most successful companies will be those that know how to manage trade-off, invest pragmatically and smartly manage data.
Check out the best IT automation software.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
