Earnings calls for major technology companies have always drawn scrutiny, but growing talk of an AI bubble has put technology spending — and performance — under a microscope. As the market awaits reports from Microsoft, Meta and Apple, among others, CIOs at many enterprises may also be feeling the heat to show ROI.
After an initial flush of excitement and exploration, companies and their stakeholders are no longer as willing to fund AI ventures without evidence of results. But what happens when that expectation clashes with a CIO’s long-term roadmap?
“In the early days of enterprise AI, from mid-2010 until 2021, companies were happy to invest in AI initiatives regardless of the outcome,” said Quentin Reul, director of global AI strategy and solutions at expert.ai. “Most enterprises were in an ‘investigative’ mode in terms of what problems AI would solve. Now, most public companies have faced pressures from their investors to deliver real value from AI initiatives.”
From a business perspective, that shift makes sense: After years of signing large or even blank checks, executives want to see where the money has gone and what it has bought them. But AI investment does not always take a simple path forward, and it can take a while to generate the clear financial returns that investors seek
When earnings calls raise questions about AI ROI, how can CIOs answer?
A new playing field
For vendors working closely with CIOs and IT teams on AI strategy, it’s clear the conversation is changing. Andrew Hillier, co-founder and CTO of Kubex, an AI infrastructure optimization platform, recalls earlier interactions with customers who were excited by AI’s potential, where the focus was almost entirely on speed and scale: How fast can we deploy AI? How many GPUs can we provision?
“Now, we’re hearing, ‘How do we prove these investments are paying off?'” he said. “AI is moving out of innovation budgets and into operations budgets. This brings much more scrutiny.”
It’s not that previous conversations ignored the financial piece; it’s that today it is front and center for boards, executives and CIOs alike. The message from the top is clear: A successful AI initiative is not just one that works smoothly, but one that has measurable impact on the company’s bottom line. This naturally influences what projects can be greenlit and where future investments are made.
One challenge of this new environment is that the novelty of a new AI capability is no longer sufficient to justify IT spend. Reul described how, in earlier phases of enterprise adoption of AI, the launch of an AI-enhanced workflow on the back end or AI-optimized experience on the front end could position a company as a digital innovator. Even if the tool’s impact on operations was minor or unproven, the willingness to experiment and be first could offset any costs through a boost to company image — but no longer.
Related:IT Leaders Fast-5: Craig Williams, Ciena
It can be difficult to close an AI project when investors are asking for returns, but CIOs today need to be even more discerning about which initiatives truly create value. AI can quickly shift from a source of profitable competitive advantage to a cost center if it’s used for the sake of being used. And as the technology becomes better understood, earlier roadmaps may no longer be appropriate.
Reul warned against companies perpetuating flawed AI investments simply “because of previous investment, even though there is little hope for the initiative to deliver on its initial promise.”
Sometimes, a pivot may be needed. In other circumstances, patience really is rewarded.
The AI runway: Why you can’t rush results
One consolation for enterprise CIOs is that they are not alone in facing increased pressure to prove AI returns. Demonstrating ROI on AI initiatives has been consistently difficult in recent years. In some cases, this is because the project is still too early in its implementation to generate the results that AI evangelists promise: greater efficiency, automation, speed.
Ling Zhang, founder and data and AI strategy consultant at Grow to Your Fullest, attributed this disconnect between board expectations and IT results to a fundamental misunderstanding of what AI investment represents: a transformation over time, not a flashy show.
“Picture an airplane,” she said. “If the airplane is scaled to accommodate more people, the runway and taxiing will have to be longer, right? If you work on short-term returns or quick shows, it will not work.”
Asking investors to wait longer for returns is risky, especially if they have already fronted significant capital for these AI initiatives. But Zhang emphasized that taking a shortcut simply isn’t an option for effective AI. From organizing data and securing infrastructure, to figuring out governance and priming teams to adopt the new tools, the foundational work required for AI can’t be rushed. And it falls increasingly on the CIO to enforce this methodical approach, even over a longer timeline.
“People like chasing shiny objects,” she said. “But everything needs a foundation — and you build the foundation.”
One way that CIOs can keep finance aligned during this stage is to be thoughtful with resource allocation when laying the groundwork. Hillier says many customers are realizing the importance of sufficient infrastructure to successful AI implementation, but the right configurations must still be tightly connected to specific use cases.
“Building the infrastructure and capabilities to enable future AI use cases is obviously important in this AI era and arms race,” he said. “But if the infrastructure spend grows faster than visible business adoption, finance will start asking tough questions.”
The metrics that can make the difference for enterprise CIOs
There is expert consensus that abandoning a long-term AI roadmap to placate boards and investors with quick wins is a disastrous idea. There are simply too many connecting pieces that need to be fine-tuned. But that doesn’t mean CIOs can’t offer stakeholders some insights and metrics to keep them confident in the larger plan.
Zhang emphasized the importance of a holistic approach and getting buy-in from all sides before beginning an AI project, which she said has to start with the CIO. She recommends starting any initiative by asking the business side directly what they hope to see, in material terms, so that any and all AI metrics are calculated against that goalpost.
Hillier agrees that CIOs need to translate their technical gains into language that resonates with financial perspectives: “Finance recognizes value when you can tie it to actual dollars and cents,” he said. “Even tying back to the potential loss of dollars and cents with operational risk reduction showcases value: What would an outage or incident have cost us?”
That translation can be trickier if AI was approached with a more experimental mindset. Reul has seen many organizations try to find or create a use for AI in their operations, rather than starting with a problem for AI to solve. This can make it harder to track AI’s impact and value. However, Reul and Hillier both emphasized that highlighting clear cost-savers is an effective way to reassure stakeholders.
“For most public companies, AI initiatives tied to financial revenues, such as increased number of sales or subscription retention, are more likely to survive increased scrutiny,” Reul said. Internal AI initiatives, by contrast, face greater risk unless CIOs can tie them to measurable cost savings, such as a reduction in external software licenses or a reduction of workforce.
Making intangible value tangible
Depending on where an organization is in its AI journey, there may be substantial numbers to back up ROI — or there may be minimal support. If it’s early days, compelling storytelling around the roadmap can be enough to win over the financial side of the business, Zhang said, as long as it includes signposts along the way that will speak directly to revenue creation. However, if it’s been one or two years and there is minimal material gain to show for it, CIOs should rethink their approach, she said.
“They need to communicate where the money is being spent and what the result is, but the result doesn’t necessarily have to be, ‘Oh, I already made X dollars,'” she said.
Consider intangible improvements that might be worth celebrating, Zhang proposed. If an AI initiative is customer-facing, for example, map its impact to customer satisfaction and the benefits of this over time. When AI is implemented to automate internal workflows, don’t forget the human benefit: Saving employees time and stress, so they can devote more care to their families, can lead to higher job satisfaction — and greater productivity.
Pressure as motivation, not paralysis
Hillier sympathizes with organizations that are trying to demonstrate the more ephemeral benefits that they “know are real but are difficult to quantify.” He gave the example of using AI for competitive analysis: “It makes the team smarter and more productive, but it’s hard to measure. You can’t point to specific cost reductions.”
However, this investor pressure can be a positive influence on AI initiatives, according to Hillier. Rather than a constraint, the need to demonstrate ROI may actually prompt more effective decision-making.
“We don’t see anyone wavering on whether they need to be doing something with AI, just that they want to make sure that they have a handle on the impact on budgets,” he said. “The pressure to justify spend is certainly there, and it’s forcing better discipline and companies to think about the right infrastructure configuration.”
Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.