Companies are abandoning AI projects at an alarming rate. The technology isn't the problem.
The pattern is showing up everywhere right now, and the numbers are starting to confirm it.
A majority of senior leaders at major professional services firms say they abandoned at least one AI project in the past year. Not because the technology didn't work. Not because the business case wasn't there. Because their people didn't have the skills to deliver it.
Deloitte's most recent State of AI in the Enterprise report, drawing on thousands of business and technology leaders across more than two dozen countries, landed on the same conclusion: insufficient workforce skills are now the single biggest barrier to integrating AI into the business. Not budget. Not technology access. Not executive buy-in. People.
This is the part of the AI conversation that gets the least attention and causes the most damage.
The problem with how companies are responding
When organizations realize they have an AI capability gap, the instinct is usually to hire around it. Bring in contractors with the specific skills you need right now. Recruit a few specialists. Stand up a small internal AI team and let them handle it.
This approach works in the short term and creates problems in the long term. It's expensive to sustain. It creates critical dependencies on a small group of people who, if they leave, take irreplaceable knowledge with them. And it doesn't build anything. The organization is still just as unprepared for the next capability gap as it was before.
The alternative, building genuine AI capability across the organization systematically, feels slower and harder. It is slower and harder. It is also the only approach that actually compounds.
What organizational AI capability actually consists of
Most companies think about AI capability as a single thing. You either have it or you don't. The more useful frame is that it consists of several distinct layers, each of which needs to be developed differently.
The first is technical depth. This is the specialized engineering capability that builds and maintains AI systems at the infrastructure level: machine learning engineering, data pipelines, model evaluation, AI security. Without it, organizations buy and build the wrong things, and they create risks they don't have the expertise to recognize or manage.
The second is domain application. This is where AI strategy meets operational reality. It's the capability to apply AI within a specific business function in ways that create actual value. The people who carry this capability understand both what the technology can do and where it matters in their specific context. This layer is where most AI initiatives succeed or fail, and it's the one that requires the deepest investment to build because it can't be hired from outside in any scalable way.
The third is general workforce fluency. This is the baseline every knowledge worker needs to function effectively in an AI-enabled organization. Not technical expertise, but enough understanding to use AI tools productively, recognize when outputs are unreliable, spot misuse, and contribute meaningfully to conversations about how AI is being deployed in their area. Without this layer, adoption stalls at the edges of the organization regardless of how strong the technical core is.
Each layer requires a different development strategy. Lumping them together into a single "AI training" initiative is one of the most common mistakes organizations make.
What the first 90 days should look like
The companies that build durable AI capability start with an honest assessment rather than a training program. Before deciding what to teach anyone, the first step is understanding exactly where the gaps are, by function, by role, and by layer.
That means mapping what your business actually needs AI to do, identifying which parts of the organization need which kinds of capability, and being honest about the difference between what you have and what you need. The gap analysis is usually more sobering than expected, which is why many organizations skip it in favor of jumping straight to solutions.
Once the gap is understood, the sequencing matters. Building technical depth takes the longest and requires the most selective hiring and development. Building domain application capability requires pairing technical people with deep functional experts and giving them the time and mandate to work through real problems together, not hypothetical ones. Building general workforce fluency is the fastest layer to move and often has the most immediate impact on how broadly AI adoption actually spreads.
The organizations getting ahead of this aren't trying to solve everything at once. They're picking the highest-leverage gaps, moving on them systematically, and building the infrastructure to keep learning as the technology keeps changing. Deloitte's research found that more than half of organizations are focused on raising overall AI fluency across the workforce, but far fewer are doing the harder work of redesigning roles, career paths, and workflows to match the new reality.
Why this matters more than the technology
The AI tools available to companies today are genuinely powerful. The gap between what the technology can do and what most organizations are actually doing with it is enormous, and it's almost never explained by the technology itself.
The companies that close that gap over the next few years won't be the ones that bought the best tools or hired the best consultants. They'll be the ones that built the organizational capacity to actually use what's available, learn from it, and adapt as it evolves.
That kind of capability doesn't come from a vendor. It doesn't come from a one-time training initiative. It comes from treating AI capability as an organizational muscle that needs to be developed continuously, starting now, and building systematically from wherever you actually are.
The organizations that get ahead of this will compound the advantage every quarter. The ones that keep hiring around the gap will keep being surprised by how little their AI investments are delivering.
