Organizations point to three main reasons these projects drag on, along with practical ideas for fixing them. Let’s count down to the top culprit — game-show style — since multiple answers were allowed.

But before we get to that, a quick reality check: Cloud isn’t new, but AI is. Cloud practices have matured over time, while AI is still in its early stages. Ideally, enterprises should be applying lessons learned from years of cloud adoption to their emerging AI efforts. So far, we’re seeing progress on the cloud front — but not nearly enough crossover into AI project management.

Figure 1. Why Cloud and AI Projects Falter — and How to Set Them Right.

More than half of enterprises (56%) admit they lack the essential in-house skills needed for cloud and AI initiatives — and acquiring those skills is taking far longer than they expected. What’s surprising is that many didn’t see this coming. They either assumed their existing teams were fully equipped (the most common case) or believed they could easily bring new expertise on board. Figure 1. Why Cloud and AI Projects Falter — and How to Set Them Right.

In reality, most organizations had the wrong mix of skills. They employed capable developers and operations specialists — but what they truly needed were architects. Without strong architectural expertise, projects stalled early in the planning phase. Teams struggled to connect new technologies to real business goals, leaving them unable to design workable, outcome-driven solutions.

During early planning and business case reviews, 69% of enterprises discover that their project requirements simply can’t be met as defined, forcing a round of redefinition before real work even begins. This raises a crucial question: Is the problem with the project itself, or with how the requirements are set? According to enterprises, it’s almost always the latter.

For cloud projects, the issue often starts with a common misconception among senior leaders — the belief that “the cloud is always cheaper.” Despite growing awareness of cloud repatriation (moving workloads back to on-premises data centers), unrealistic cost-saving assumptions still surface early on. The good news is that most IT teams now know how to assess cloud initiatives for cost and value, so unworkable business cases are usually caught before major investment.

For AI projects, the challenges are more fundamental. Both executives and departmental leaders tend to overestimate what AI can do, often based on exposure to consumer-facing generative AI tools. Around 25% of AI proposals quickly run into data governance issues — and about half of those are shut down entirely. Among the survivors, a deeper issue emerges: a disconnect between business goals and actionable AI strategies.

Many organizations admit they don’t yet understand AI’s practical limits or applications. As one CIO put it, many of the proposals they receive are “invitations to AI fishing trips” — vague initiatives framed around high-level business outcomes (“improve sales,” “reduce costs”) rather than clear implementation paths. What’s really needed first is a discovery phase to map those business ambitions to viable AI use cases — and then define a project to make it happen.

Enterprises point to one recurring theme — the value of “strategic vendors.” But asking every organization to find and nurture one isn’t realistic. Those relationships often take a decade or more to build. The encouraging news, though, is that the functions a strategic vendor performs can actually be replicated in-house.

To keep cloud and AI projects on schedule — and make them successful — enterprises need a continuous, shared understanding of both business objectives and technological possibilities. Strategic vendors help bridge that gap, but they’re not the only way to do it.

Enterprises with the highest success rates in cloud initiatives tend to have something in common: they’ve built dedicated “cloud teams.” These groups maintain ongoing collaboration between IT and business leaders, ensuring both sides understand the technology’s capabilities and its strategic impact.

It’s early, but a similar pattern seems to be emerging for AI projects. Organizations seeing the best outcomes are forming “AI teams” — mixed groups of line-of-business managers and AI specialists. Both sides are educated in how AI actually works, allowing them to choose the right approach, build realistic business cases, and implement solutions together.

Ultimately, AI is teaching enterprises a new language — one that blends business intent with technical execution. The challenge is to ensure it becomes one shared language, not separate dialects spoken by IT and management. And that means investing the time to learn it well, across the organization.

Source: NETWORK WORLD

Cite this article:

Priyadharshini S (2025), Why Cloud and AI Projects Stall — And How to Get Them Back on Track, AnaTechMaz, pp.170