Investment in any technology has typically been decided based on returns in the investment. But maybe ROI is not the key consideration when organizations innovate with AI… or is it?

IBM’s recent ROI of AI study found that among those not currently achieving positive ROI from AI, only about 1 in 4 expect to see positive ROI in the next year. 

Should CIOs press the panic alarm? Not if AI implementation is motivated by innovation rather than ROI alone. Or maybe it’s time we revisit ROI metrics for investment in technologies such as AI…

We discuss these issues, as well as some key findings from the ROI of AI report such as the importance of open-source in AI and the future of AI in Aisa Pacific, with Shanker V Selvadurai, Vice President and CTO, IBM Asia Pacific.

31% of companies surveyed in IBM’s recent ROI of AI study say their AI investments are driven more by innovation, compared to 28% that are more ROI driven; 41% indicate that their organizations are equally innovation and ROI-driven. In your opinion, what does this say about how organizations are measuring ROI on AI?

Shanker: The study underscores a pivotal shift in how organizations perceive AI ROI. While 28% still prioritize immediate financial returns, a notable 31% view innovation itself as a form of return, and 41% balance both innovation and ROI. This indicates a broader understanding that ROI encompasses more than just cost savings. It includes strategic advancements like agility, product differentiation, and time-to-market improvements.

Moreover, only 11% of organizations expect immediate returns within two years, while nearly 60% anticipate realizing business impact over a two to five-year horizon. This long-term perspective highlights a growing emphasis on factors such as faster software development, rapid innovation, and productivity gains as critical ROI metrics, moving beyond traditional financial measures.

How should an organization develop a holistic view of AI and more relevant ROI metrics?

Shanker: Developing a comprehensive AI ROI framework involves expanding beyond traditional cost-saving metrics to include:

  • Time-to-market and productivity gains: In 2024, only 15% of IT decision-makers ranked quantifiable savings as their top AI metric. Instead, faster software development and reduced troubleshooting time have become more prominent indicators of success.
  • Innovation momentum: Organizations that layer quick wins, like automating routine workflows, on top of longer-term strategic AI initiatives, such as generative AI-powered product design, tend to achieve more substantial payoffs.
  • Customer and employee impact: Improvements in experience often lead to tangible business outcomes. IBM’s Client Zero initiative, for example, used AI to automate 280,000 HR tasks and save 1.6 million hours annually — boosting both productivity and employee satisfaction. Such gains, while sometimes seen as “soft,” are increasingly being treated as core to AI ROI.

To operationalize this, cross-functional teams (including finance, IT, data, and business units) should co-create an AI scorecard. This scorecard should track both tangible metrics (e.g., hours saved, cost avoidance) and intangible ones (e.g., time-to-insight, Net Promoter Score uplift), aligned with short-, mid-, and long-term milestones.

It’s also crucial to maintain a clear vision of the desired end goals. Beyond traditional ROI, foresight into what constitutes a successful AI strategy — like increased customer satisfaction — adds depth to the evaluation process, even if such outcomes are challenging to measure in the short term.

With 60% of IT decision makers reportedly using open-source ecosystems as an AI tool source, do you expect that to change in the near future? How and why is open source crucial for AI adoption, especially in Asia Pacific?

Shanker: The reliance on open-source ecosystems is poised to continue and even deepen, particularly in the Asia Pacific region. Open source is vital for AI adoption due to several factors:

  • Cost efficiency and flexibility: Open-source models eliminate license fees and can be fine-tuned with proprietary data, offering a significant advantage for midsize enterprises across Asia Pacific.
  • Rapid innovation and local adaptation: Crucial in a diverse region like Asia Pacific, open-source models can be tailored for local languages and contexts. Tools like InstructLab make it easier to customize models with domain-specific data, improving accuracy and accelerating deployment.
  • DevOps integration: Incorporating open-source tools into DevOps workflows streamlines processes, automates repetitive tasks, and fosters greater collaboration, accelerating development and deployment cycles.

IBM advocates for openness as the foundation for scaling AI responsibly, combining community-driven innovation with enterprise-grade governance.

For 2025 and beyond, in what areas are Asia Pacific organizations continuing to invest in AI, and in what areas are they reducing investments? Why?

Shanker: Investment patterns in Asia Pacific largely align with global trends, with organizations continuing to prioritize areas such as data quality management, product and service innovation, and IT operations. These domains directly support trustworthy analytics, enhanced customer experience, and operational efficiency — key enablers of measurable business value and AI ROI.

Meanwhile, some areas such as marketing, and certain IT operations, are seeing comparatively lower levels of investment. This doesn’t signal a pullback, but rather a strategic shift towards functions where AI delivers clearer, more immediate impact on innovation and productivity.

Asia Pacific leaders are focusing their efforts on initiatives with strong business alignment over a 2–5-year horizon. They’re deploying smaller, specialized open-source models that offer the right balance of performance, cost, and governance. Increasingly, they’re also turning to agentic AI to orchestrate complex workflows and improve adaptability.

Ultimately, success in AI hinges not on how much is invested across the board, but on how intentionally those investments are aligned with long-term business outcomes, and that starts with recognizing that the future of AI is open.