In 2025, big tech has spent more on AI than the US government has spent on education, jobs and social services. So perhaps claims of the demise of AI came too early, and instead, we are entering the era of AI ROI?
Conservative estimates claim that major tech firms have lavished more than $155 billion on AI development projects, with the race to be the leader in AI likely to accelerate to hundreds of billions for each firm every year.
With this much being invested, it is no wonder that the large vendors are pushing the AI agenda – and pushing back against evidence that the AI hype cycle might be about to burst.
Last month, TechInformed reported on early signs of a slowdown in AI adoption and what that might mean for the industry. Industry figures acknowledged that enterprises are less interested in large-scale pilot projects, instead focussing on how to make money.
A recent MIT NANDA report reveals a surprising and troubling trend – 95% of organisations report zero return on investments into Generative AI projects.
Nitesh Bansal, CEO of R Systems and a veteran digital product engineering leader, believes this divide comes down to approach. “The enterprises seeing returns are those that focus on embedding AI into core workflows to solve immediate business problems, rather than treating AI as a moonshot project,” he says.
R Systems has worked on numerous AI projects with its 300 international clients, but one challenge the industry is facing is moving from pilot phase to full adoption.
Neeraj Abhyankar, VP, data & AI at R Systems, explains: “Many enterprises rush into GenAI projects with high expectations, vague objectives, and underprepared data infrastructures, resulting in tools that, unfortunately, become difficult to scale or provide long-term value.”
Successful GenAI adoption and implementation are contingent on strong, clean data ecosystems, industry-specific design, robust governance, and seamless integration into practical workflows, he adds. “Many projects fail when they’re treated as innovation showcases rather than used for business tasks that can provide tangible ROI.”
“The AI hype bubble isn’t bursting, it’s maturing. Today we’re seeing enterprises expecting more from their investments, wanting to have a clear picture of ROI that aligns with KPIs and operational necessities and ensure the models they are deploying are in line with ethical standards and are able to be governed appropriately.”
Second wave
Experts told TI that we are entering the second wave of GenAI projects, and this wave will come with a focus on use cases, where AI improves workflows and produces measurable outcomes.
“A big lesson from that first wave is that most AI systems can process data but fail to capture the human side like context, emotion and intent,” explains Neurologyca chief strategy officer Marc Fernandez. “When it is clear from the task or the prompt that human context adds value, the difference in results is significant.”
He points to a number of examples including; wellness, where AI can distinguish between stress and focus; soft skills development, where it can track confidence or communication improvements; AI agents and robotics, where adapting to human cues is essential; education, where personalization goes beyond right and wrong answers; security, where intent recognition can improve threat detection; and personalized recommendations, where AI can adjust not only to preferences but also to someone’s state of mind.
“This shift makes adoption look slower on the surface, but it shows companies are building more sustainable strategies for how AI fits into their business,” adds Fernandez. “The conversation has moved from experiments to impact and from hype to results. The companies that succeed will be the ones that use AI not just to process information but to support how people actually work and make decisions.”
Squirro’s founder and CEO, Dorian Selz, claims that the idea that the AI bubble is about to burst misses the bigger picture.
“Yes, the craze for AI pilots will be cut sharply. Many such pilots have been started without a clear-cut business case.”
“Thus, what we’re seeing is a shift from hype-driven pilots to value-driven adoption. Enterprises are moving past experimentation and focusing more on ROI, integration, and risk management, all of which takes time and discipline. And is harder than many think.”
Additionally, companies are assessing the regulatory setup of their activities. Without clear and practical frameworks, many large organisations will hold back from scaling AI projects because of concerns around liability, compliance, data privacy, and ethical use, he claims.
“These elements are prerequisites to build trust and certainty, both needed for enterprise-wide adoption in the long run.”
“We continue to see strong demand for AI solutions that deliver measurable outcomes, particularly where governance and transparency are built in from the start. Far from bursting, the AI cycle is entering a more mature phase that prioritises more sustainable and responsible growth over short-term excitement.”
Experimental hype
The shift away from pilots to more ROI focused projects is actually a huge opportunity for the industry, argues Publicis Sapient, head of data science and AI at Simon James.
James claims the recent slowdown reflects AI’s maturation from “experimental hype into strategic value-driven deployment” and adds that this is why there are fewer flashy announcements, but “more substantial, sustainable integration”.
“The increasing adoption of generative AI for a widening range of tasks, the building of smarter agents, the disruption to search engines and the unpreparedness of many businesses create a once-in-a-generation opportunity to disrupt the status quo in many markets,” says James.
“Those brands that can adapt faster will have an advantage that may start slowly and imperceptibly but can snowball dramatically. The most forward-thinking organisations are already adjusting their strategies accordingly, focusing on practical application to business challenges rather than getting caught up in hype cycles.”
James adds that organisations with clear, auditable AI systems can move faster, take bigger bets and scale more aggressively “because stakeholders trust the technology.”
“While competitors get stuck in endless committee discussions about AI risks, transparent AI deployers are capturing market share. The question for business leaders is no longer whether to adopt agentic AI, but how to thoughtfully integrate increasingly capable AI systems into their operations.”
Media madness
Digitate chief customer officer Ugo Orsi argues the media has helped to contribute to the hype bubble, especially due to GenAI’s ability to generate both visual and written content.
On the ground, however, he claims genAI has been “confirming its capabilities” for enterprises.
“Gen AI has the capability to disrupt white-collar jobs and radically change office work. In a future working environment workflow, processes will no longer be strategic to managing people, as 80% of office work can be replaced by AI and Gen AI.”
“Now enterprises have realised it, they are taking time: why embark on such a change, if not forced. As of now, there is no compelling reason to drive AI/ Gen AI adoption at scale.”
Peter van der Putten, director of the AI Lab, Pegasystems and an assistant professor, AI, Leiden University, sums up the new era, claiming any signs of a bubble bursting are actually just a course correction in expectations.
“There is nothing more seasonal in technology than AI hype, with its many winters and summers. But this is with the emphasis on the word ‘hype’ – the overinflated expectations of AI within financial markets.”
“The underlying development and adoption of the technology has been steadily on the rise. So whatever whiff of AI wintery conditions appears this will be a course correction in these expectations, rather than adoption coming to a freezing halt.”
He argues that not-so-successful AI will be within enterprises that run “tons of lab projects and proof-of-concepts rather than scaling up.”
“These become fragmented AI projects, where intelligence is disconnected from both corporate strategy as well as from high-value and volume processes, workflows and interactions.”
Van der Putten suggests that enterprise AI will flourish when it is harnessed to workflows and with a clear sense of the desired outcome.
“This is when the business problem or opportunity comes first in plans for AI adoption, and businesses are open not just to the new AI kids on the block, such as agentic AI, but also keep an eye open for leveraging good-old-fashioned analytical AI capabilities such as process mining, automated decisioning and predictive analytics.”
Read Part 1 of our AI Hype feature HERE
