Every few weeks, another report pops up questioning whether AI is really delivering on its promise. Gartner recently found that only 28% of AI initiatives in infrastructure and operations meet ROI expectations, while one in five fails outright. Other studies paint an even bleaker picture, with many reporting that the majority of AI projects fail to deliver meaningful business value.
It's easy to look at these statistics and conclude that AI simply isn't living up to the hype. Some organizations have even started pumping the brakes on AI investments altogether. But before we blame the technology, it's worth asking a different question: Why are these projects actually failing?
The more we look at successful and unsuccessful AI initiatives, the more one thing becomes clear. Most failures we've seen have very little to do with the underlying models or the technology itself. Instead, they stem from the same execution mistakes that have derailed technology projects for decades. AI may be new, but the fundamentals of delivering successful business technology haven't changed.
In this article, we'll look at why AI project failure rates are often misunderstood, what lessons we can learn from previous waves of technology innovation, and why organizations that focus on fundamentals like user experience, data, integration, and change management are the ones seeing the strongest returns from AI today.
Lessons Learned From the Dot-Com Era
To put some of these concepts into perspective, it's helpful to look back at another recent(ish) technology revolution. The rise of the Internet in the late 1990s created a similar level of excitement, attracting enormous investment and the promise to fundamentally change the way businesses operated.
During that timeframe, investors poured billions into Internet startups, venture capital firms raced to fund the next big idea, and entrepreneurs scrambled to stake their claim in what felt like a once-in-a-generation opportunity. For a while, it seemed like simply having an Internet-based business was enough to attract investment and guarantee success.
When the dot-com bubble burst in the early 2000s, it was easy to conclude that the Internet had simply been overhyped. Countless startups disappeared almost overnight, billions of dollars in investment evaporated, and many people questioned whether e-commerce would ever become the transformative force its advocates had promised.
Looking back, it's pretty obvious that these knee-jerk reactions were way off base. The Internet didn't fail. Rather, a lot of businesses built on weak ideas, unrealistic expectations, and poor execution did. That's an important distinction, because people often mistake failed implementations for failed technology.
If you look at the companies that succeeded, they didn't win simply because they were using the Internet. They won because they paired new technology with sound business practices. They understood their customers, designed better experiences, built efficient operations, and executed relentlessly over the long haul. Certainly many of them introduced important innovations of their own, but those innovations were built on a foundation of disciplined execution rather than hype. The Internet amplified good business practices. It didn't replace them.
Amazon is an excellent example of this. While many companies chased website traffic, media attention, and rapid growth at all costs, Amazon focused relentlessly on customer experience, operational efficiency, and continuously improving the fundamentals of how people bought products online. Technology enabled Amazon's success, but disciplined execution is what separated it from its competitors.
Today's AI landscape feels remarkably similar. Headlines about disappointing ROI and failed AI initiatives don't necessarily tell us that the technology has fallen short. More often, they tell us that organizations have underestimated what it takes to successfully adopt any transformational technology. The technology may be different, but success is still built the same way.
The Rules Haven't Changed
Perhaps the biggest misconception surrounding AI is that it's fundamentally different from every other technology innovation that came before it. While it's true that AI has introduced capabilities that did not exist before, the pattern behind it is not new.
Every major technology shift has taken something that was once scarce, expensive, and locked inside a handful of organizations, and made it available on demand to everyone else. Cloud computing did this with infrastructure. The internet did this with information. Now, AI is doing this with intelligence itself.
Organizations no longer need to build and staff every analytical or decision-making capability in-house. They can access reasoning capabilities, analytical assistance, and specialized knowledge the same way they already acquire compute or storage: as a service, priced by consumption, available the moment they need it. That shift, not the technology itself, is the real story.
But while the delivery model has changed dramatically, the work required to turn technology into business value has not. If anything, AI has reinforced the importance of getting the basics right. After all, AI doesn't eliminate the need to understand your business. It increases the value of understanding it well.
Start with People, Not Technology
At the end of the day, every successful technology project is about helping people do their jobs more effectively. That starts with understanding the people you're building for and the work they're actually trying to accomplish. From there come the disciplines that make great solutions possible: thoughtful process design, intuitive user experiences, well-integrated systems, and data that is accurate, trustworthy, and well governed. None of those disciplines disappeared when generative AI burst onto the scene.
In many ways, AI adoption is like a stress test for an organization. An agent is only as effective as the business knowledge it's given. A chatbot can't confidently answer questions if the underlying data is incomplete or inconsistent. A beautifully designed AI experience still falls flat if it doesn't fit naturally into the way people actually work. AI can dramatically accelerate good organizations, but it also has a way of exposing weaknesses that have been there all along. In other words, AI rarely creates organizational problems. It reveals them.
That's why organizations seeing the strongest returns from AI aren't necessarily the ones with the most advanced models. More often than not, they're the ones that invested in understanding their people, organizing their data, connecting their systems, and designing solutions around the way work actually gets done. Those principles have been driving successful technology projects for decades, and they're just as important today.
You Have To Put in the Work
This is probably one of the hardest concepts to communicate because we've all seen AI do things that would have seemed impossible just a few years ago. It can write production-ready code, build working prototypes in minutes, analyze thousands of documents, and solve surprisingly complex problems. There's no question these tools have fundamentally changed how software gets built and dramatically reduced the effort required to turn ideas into working solutions.
The trap is thinking this is a zero-sum game. It's easy to assume that because AI has made implementation faster, it can also take over the hard work of figuring out where and how these technologies should be applied to transform a business. That's a very different challenge. It requires understanding people, business processes, organizational goals, competing priorities, and the countless nuances that make every company unique.
Someone still has to burn the mental calories to understand how the business actually works. Someone has to identify the bottlenecks worth solving, define what success looks like, reconcile competing stakeholder priorities, and determine where AI can genuinely improve the way work gets done. Those decisions can't be delegated to a language model because they require context, judgment, and an understanding of the organization itself.
In many ways, AI has empowered organizations to shift where they invest their time. Instead of spending months writing boilerplate code or building basic user interfaces, teams can spend more time talking to users, validating assumptions, refining business processes, and experimenting with solutions. That's a tremendous advantage, but only if you choose to reinvest those savings into better discovery and design rather than simply trying to deliver projects faster.
The organizations seeing the strongest returns from AI understand this tradeoff. They aren't using AI to skip the work. They're using it to spend more time on the work that matters most.
The Home Run Trap
Another pattern we've noticed is that many organizations approach AI as if they're looking for one massive breakthrough. They want the transformational use case that changes everything overnight. While those opportunities certainly exist, chasing them first often leads to lengthy projects, unrealistic expectations, and disappointing results.
The organizations making the most progress tend to take a different approach. They start with well-defined problems, deliver solutions that make a measurable difference, and use those early successes to build confidence across the business. One AI agent that saves employees an hour every day. A Copilot grounded in trusted company knowledge. An intelligent workflow that eliminates a tedious manual process. On their own, none of these projects transform a company. Together, they begin to change how work gets done.
This isn't about thinking small. It's about building momentum. The organizations creating lasting competitive advantage aren't treating AI as a project with a finish line. They're treating it as a capability they'll continue to refine over time. Every successful implementation teaches the organization something about governance, user adoption, data quality, and where AI creates the most value. Those lessons make the next project faster, better, and less risky. Over time, the organization develops the skills, trust, and institutional knowledge needed to tackle much larger opportunities.
Ironically, that's how many of the biggest AI success stories are likely to be written. Competitive advantage is rarely built on one spectacular swing. More often, it's built by consistently getting on base until those small wins compound into something much bigger.
Good IT Discipline Still Matters
It's easy to get caught up in the excitement surrounding AI and assume that traditional IT disciplines somehow become less important. In reality, the opposite is true. As AI becomes more deeply embedded in everyday business processes, the quality of your architecture, data, integration strategy, security, and governance become even more important than before.
Every AI agent, Copilot, or intelligent workflow ultimately depends on the technology foundation beneath it. If your systems aren't connected, your data isn't trustworthy, or your security model is inconsistent, AI doesn't solve those problems. It inherits them. Indeed, it many cases, it amplifies them.
The good news is that organizations don't have to choose between strengthening their technology foundation and pursuing AI innovation. The two go hand in hand. Every investment in integration, data quality, governance, and modern architecture expands what's possible with AI tomorrow. Likewise, every successful AI initiative helps organizations identify where those foundational investments will have the greatest impact.
The organizations that will realize the greatest long-term value from AI won't necessarily be the ones with access to the most advanced models. They'll be the ones that consistently combine strong technology leadership with a deep understanding of their business and a relentless focus on solving real problems. AI may be changing how software is built, but it hasn't changed what great IT organizations do best.
Closing Thoughts
The headlines about disappointing AI ROI can make it sound like the technology has failed to live up to the hype. I don't think that's what's happening. More often than not, organizations are learning the same lessons that have accompanied every major technology shift over the past several decades. Success isn't determined by the technology alone. It's determined by how thoughtfully it's applied to solve real business problems.
The encouraging news is that none of this requires organizations to wait for the next generation of models or some future breakthrough. The capabilities already exist today. The companies seeing the greatest returns are investing in understanding their business, building strong technology foundations, delivering incremental wins, and continuously refining how AI fits into the way their people work. Those aren't new ideas, but they remain just as relevant in the age of AI as they were during every technology revolution that came before it.
If your organization is wrestling with disappointing AI results or simply trying to determine where to begin, it may be worth asking a different question. Instead of wondering whether AI is ready for your business, ask whether your business is ready to take full advantage of AI. If you'd like to explore what that looks like for your organization, we'd welcome the opportunity to have that conversation.
