Judging by the relentlessly upbeat marketing from Silicon Valley’s technology giants, you would think the generative AI (GenAI) hype-train was still on a smoothly inclining track. Yet there is widespread sentiment among executives that it is difficult to extract value from AI technology.
If 2024 was the year of GenAI experimentation, 2025 was meant to be the year of scaling. However, rolling out this technology across the organization is proving difficult. Gartner predicts that 30% of GenAI projects will be abandoned after proof of concept by the end of 2025. Many describe this period of disenfranchisement as the ‘GenAI winter.’
This isn’t the first time we have seen this predicament. It echoes the difficulties that many businesses experienced during earlier waves of digital transformation (for example, the introduction of cloud). An initial burst of enthusiasm is cooled by the reality of organizational complexity.
The disillusionment stems from two distinct but often conflated challenges: transitioning from pilot to scale and measuring the performance of AI initiatives. To implement AI with success, businesses must untangle these and address each in turn.
Outdated legacy technology, a lack of human skills, cultural aversion to change and increasingly strict regulation all make scaling AI tricky. Yet IMD’s analysis of the world’s largest 300 businesses uncovers multiple success stories and a blueprint for effective scaling.
For example, automotive giants Volkswagen and Mercedes-Benz have deployed AI-powered software to make life much easier for the driver. Volkswagen’s AI copilot tool provides personalized driver convenience and assistance systems while Mercedes-Benz uses AI to optimize vehicle performance via over-the-air updates.
In the manufacturing sector, Siemens deploys AI to predict which machinery components need maintenance while GE Aerospace uses AI for quality control.
The success stories we identified show that advance planning is paramount. As soon as an AI pilot shows potential, it is important to map out the challenges that could derail scaling. These could include integrating AI with existing systems or reluctance among users to adopt the tech. Highlighting these potential barriers well in advance of scaling allows sufficient time to address them.
Creating dedicated teams, each focused on scaling a specific use case, can also work well. The chief digital officer of a major consumer goods company recently told me that, in their organization, as soon as an AI initiative shows promise, a dedicated group begins to address legal, cybersecurity and compliance issues. This lays the groundwork for widespread adoption.
Organizations should remember that ‘scaling’ is not a blanket concept. A tool such as Microsoft’s Copilot could be scaled to literally every employee within an organization. But scaling a tool that helps developers code more effectively means rolling it out only to those undertaking that specific task. The more widely a specific AI tool is implemented, the greater the range of challenges that emerge.
It’s crucial to anticipate and address skills gaps as soon as possible. To this end, many businesses have launched AI upskilling programs. Frequently, these teach employees GenAI basics, such as how to prompt. As important as this is, organizations must ensure that they are also fostering the specific competencies required to support the scaling of AI, such as data governance, cybersecurity and risk management.
A Gartner report found that almost half of businesses lacked specific success metrics for digital initiatives.
A lack of robust performance measures exacerbates executive disillusionment with AI. A Gartner report found that almost half of businesses lacked specific success metrics for digital initiatives. Many rely on usage rates, which, while important, do not capture overall outcomes.
A two-pronged approach can offer a holistic assessment of AI maturity. First, businesses should assess whether they have the foundations in place to integrate AI at scale. Second, they need to find an effective way of measuring the business outcomes of specific AI initiatives.
Organizations should assess their preparedness to use AI across five dimensions:
Measure the benefits of specific AI deployments across the following performance categories:
By addressing these separate but often conflated issues, business executives can start to see genuine benefits from using AI. Spring is only months away!
As investment in AI accelerates, access to emerging technology skills is becoming a decisive driver of career progression and pay. New research shows that structural features of tech roles, not women’s choices...
Spotify CHRO Anna Lundström drives AI readiness, culture, and personalized well-being to empower employees and sustain Spotify’s innovative, human-centric growth....
Transform your AI adoption strategy from a high-stakes gamble into a portfolio of calculated moves using this plan, write Faisal Hoque and Paul Scade....
Organizations should let departments choose the right balance between humans and AI, and focus on outcomes, not rigid processes, to keep pace with rapid technological change. ...
As investment in AI accelerates, access to emerging technology skills is becoming a decisive driver of career progression and pay. New research shows that structural features of tech roles, not women’s choices...
Spotify CHRO Anna Lundström drives AI readiness, culture, and personalized well-being to empower employees and sustain Spotify’s innovative, human-centric growth....
Transform your AI adoption strategy from a high-stakes gamble into a portfolio of calculated moves using this plan, write Faisal Hoque and Paul Scade....
Organizations should let departments choose the right balance between humans and AI, and focus on outcomes, not rigid processes, to keep pace with rapid technological change. ...
Explore first person business intelligence from top minds curated for a global executive audience