AI investment: Where the real ROI lies
By Nihar Dalmia, GenAI Market Activation Leader for Deloitte Canada
A few weeks ago, a new Deloitte report came out highlighting the paradox between the swell in organizational expenditure on AI technology and elusive return on investment (ROI) across Europe and the Middle East.
If you haven't read the article, it makes some unmissable points. Read it here: AI ROI: The paradox of rising investment and elusive returns.
Before we get into my point of view on AI and ROI, I want to give you a snapshot of the narrative the article lays out.
Here’s my key takeaways from the report:
- 85% of organizations increased AI investment in the past year and 91% plan to increase it again — yet only 6% saw payback within one year.
- ROI on typical AI use-cases takes around 2 to 4 years, far longer than the 7-to-12-month expectation for many tech investments.
- Key barriers to achieving ROI include benefits that are intangible (e.g., better vendor relations), fragmented data/infrastructure, rapidly evolving tech outpacing metrics, human adoption issues, and AI being entangled with broader transformation.
- Organizations continue to invest because they see AI as a strategic imperative, but this is often driven by fear of falling behind and belief in long-term value and not because the short-term returns are clear.
- High-performing ‘AI ROI leaders’ differentiate themselves by rethinking business models and taking a human-centred approach, using tailored ROI metrics, and mandating broad AI fluency.
- There is a distinct difference between generative AI and agentic AI. Generative AI can deliver measurable ROI within about a year. While agentic AI (autonomous end-to-end process systems) is more complex and, most expect to realize that ROI within 1 to 5 years.
Contact us
And this is what the article says you should do differently:
- Treat AI as a business-transformation lever, not just a tech upgrade
- Allocate meaningful budget and make AI a strategic priority
- Embed human-centred design and change-management
- Elevate governance and executive ownership
- Use tailored ROI frameworks and measure over realistic timeframes
- Lay strong data foundations and architectural readiness for scale
But for me, there’s more to the story...
We’re living through a moment of AI hype and pressure to spend. But the returns, so far, remain elusive. Across industries, there’s a widening gap between how fast organizations are investing in AI and how slowly they’re preparing to actually capture value from it.
I see it every day. Organizations are racing ahead with big investments in generative and agentic AI, but their readiness lags. That exists as data quality, governance, process redesign, or culture. When you move fast to invest but slow to prepare, you create a gap where value leaks out.
That’s the paradox at the heart of the piece. The value of AI is real. But for many, it’s fragmented, unevenly distributed, and still years away from being realized.
In my experience, a lot of organizations view ROI as a financial calculation
But that’s a narrow lens for a technology transformative and long-term AI. These investments don’t fit neatly into quarterly payback periods. Measuring ROI on AI is more like measuring the return on the Internet in 1998. The benefits take years to materialize and often show up in places you didn’t anticipate.
In government, (where I spend a lot of my time working with clients), return means something very different. AI might not shrink budgets or boost revenues, but it can process applications faster, reduce errors, and improve service delivery. Those outcomes may not translate into financial gain, but they create immense human value.
There’s also a self-reporting bias at play. Executives tend to see AI’s value in cost savings and efficiency. But ask the people actually using these tools, and you’ll hear something different: they’re more creative, more productive, and more fulfilled at work. They can do more with less and enjoy their jobs more while doing it. That’s ROI, too. It’s just harder to measure.
What I think the article may underplay...
The report hints at, but doesn’t fully unpack, the biggest issue: organizational readiness.
Technology is rarely that core problem. It’s people, processes, and governance that determine whether AI succeeds or stalls. ROI is elusive not because the model fails, but because integration and scaling do. Leaders underestimate how much change management, cross-functional collaboration, and executive alignment it takes to turn pilots into powerhouse programs, or AI tools into true digital teammates.
Governance and incentives matter as well. Without the right frameworks for oversight, accountability, and risk appetite, organizations pull back before realizing value. And unless culture evolves to embrace experimentation, the best models will sit idle.
AI adoption isn’t a sprint. The real winners will treat their AI workforce the way they treat their human workforce: hire it, train it, and improve it over time. It’s a long-term commitment, not a one-time procurement.
If there’s one thing I’d leave you with, it’s this:
Invest in AI...but invest just as much in your organization’s readiness to use it.
That means building cross-functional teams that break down silos between tech, data, and business. Creating governance frameworks that enable risk-taking, not stifle it. Upskilling your workforce to work with AI, not around it. And treating adoption as an ongoing transformation, not a project with an end date.
If you’re not ready to lead, be a fast follower. Learn from others, build partnerships, and avoid reinventing the wheel. The next phase of AI maturity won’t be about who spends the most. It’ll be about who integrates, scales, and sustains value the best.
So, before you invest another dollar into AI investment, ask; “If I expect my technology to perform, am I preparing my organization to change just as fast?"
That’s where the real return lies.
Nihar Dalmia
