Findings from the 2026 ZAI AI Skills Gap Survey. 113 senior leaders on the skills they cannot find, the roles that need training most, and what the next two years will look like for the workforce.
The conversation about AI in the enterprise has matured. A year ago, leaders were asking whether AI was real. Today, they are asking who can do the work. The 2026 ZAI AI Skills Gap Survey captures that shift in sharp relief.
113 senior leaders from across industries told us which AI skills are hardest to find, which roles need training most urgently, how confident they are in their workforce's readiness, and where they see jobs headed over the next two years. The picture that emerges is a workforce in motion, but not in formation.
The headline finding will reframe how organizations think about this problem. The skills gap is real, but it is not the root cause. The root cause is upstream of skills, and most organizations are trying to solve the wrong problem.
When we asked leaders to name their single biggest barrier to closing the AI skills gap, the answer was not budget. It was not workforce resistance. It was not even a lack of credible programs. The biggest barrier, by a wide margin, is that organizations have not yet defined what they are trying to do with AI in the first place.
Strategy and skill identification together account for 60% of the answers. Add workforce resistance, which is itself a downstream symptom of unclear strategy, and you reach 80%. Budget and program availability, the two answers most organizations would have offered three years ago, sit at the bottom.
This reframes the problem. The companies waiting for a credible AI training program to arrive are not actually blocked by training availability. They are blocked by the question of what to train people to do.
It has less to do with AI-trained talent and more to do with a lack of an operational model. People don't know when, where, and how to use AI in real work. — Senior Leader, Q8 Open Response
The skills story is downstream of the strategy story, and the data confirms it. When asked which AI skills are hardest to find or develop in their workforce, leaders pointed first to AI strategy and roadmapping. Not the technical skills. Not even the tooling skills. The strategic ones.
The pattern is unmistakable. The four hardest skills to find are all strategic and human in nature: strategy itself, change management, decision-making with AI in the loop, and ethical governance. The technical skills sit lower. Prompt engineering, the skill that dominated the 2024 conversation, ranks dead last.
This is a market signal that organizations are no longer struggling to teach people which buttons to press. They are struggling to teach people how to think with AI as part of their team.
47% of leaders say their biggest barrier is undefined AI strategy. 51% say strategy is the hardest skill to find. The two findings are the same finding stated from two directions. Organizations cannot define the strategy because they do not have strategists. They cannot develop strategists because they cannot define what to train them on.
Asked which roles most need AI training to be productive over the next two years, leaders did not point first to executives. They did not point first to engineers. They pointed to middle management.
Middle management leads the list at 47%, with frontline workers close behind at 41%. The combined senior leadership and C-suite tiers come in lower, suggesting that leaders see themselves as more ready than the layers immediately below them. That assessment is generous and worth examining, but the directional message is clear: the productivity battle is being fought in the middle and on the front line.
Middle managers are the layer where strategy meets execution. They are the people who translate executive intent into team-level action. If they cannot apply AI to their teams, the organization cannot scale its use. Frontline workers, meanwhile, are where actual work gets done. Investment that skips both layers is investment in theater.
I believe the skills gap is natural, and it is important to provide literacy skills to senior leadership and middle management so that they can train and set expectations for the employees and set a company strategy. — Senior Leader, Q8 Open Response
Only 22% of leaders are very confident their workforce will have the AI skills it needs over the next two years. The dominant answer, at exactly half of all respondents, is "somewhat confident." That phrase is doing a lot of work.
| Confidence Level | Count | % |
|---|---|---|
| Very confident | 25 | 22% |
| Somewhat confident | 56 | 50% |
| Not very confident | 29 | 26% |
| Not at all confident | 3 | 3% |
"Somewhat confident" is the corporate version of "I hope so." It is the answer leaders give when they cannot point to a plan but do not want to admit there is not one. Combined with the 28% who are not confident at all, three out of four leaders are looking at the next two years without conviction that their people will be ready.
The 22% who answered "very confident" are the population worth studying. They are not waiting for a perfect program. They are working a clear plan.
When asked how their organization is investing in upskilling employees on AI, leaders described a landscape dominated by internal effort and individual initiative. Universities are barely on the map.
Two percent. Out of 113 senior leaders, two are partnering with a university to build their workforce's AI capabilities. Internal programs lead the field at 40%, self-directed effort is the second-most-common approach at 27%, and 17% have not started.
This is a striking gap, and it is more interesting because it does not match what leaders say they value. When asked whether university accreditation matters in selecting a provider, half of leaders said it is at least somewhat important. The willingness exists. The participation does not. Universities have not yet shown up in a form leaders can buy.
| Response | Count | % |
|---|---|---|
| Very important | 20 | 18% |
| Somewhat important | 37 | 33% |
| Not important | 38 | 34% |
| No preference | 18 | 16% |
Half the market values university credibility. Two percent buy from universities. The space between those two numbers is the opportunity.
The most personal question on the survey asked leaders to forecast what AI would do to jobs in their industry over the next two years. The answers do not paint a picture of equilibrium.
53% of leaders expect AI to eliminate more jobs than it creates over the next two years. Only 17% expect more creation than elimination. The remaining 30% expect a wash.
This matters because it shapes the emotional environment in which every training program will land. Workforce resistance, our second-most-cited barrier, is not a mystery. It is a rational response to leaders quietly forecasting net job loss. The implication for any organization investing in AI training is clear: psychological safety and clarity about who AI is for must be designed into the program, not bolted on after.
Saying AI literacy is an important skill isn't enough. You must provide incentives for employees to participate and also buffer against the very true fear that AI will steal jobs. — Senior Leader, Q8 Open Response
The 2026 ZAI AI Skills Gap Survey points to a workforce that is moving but not aligned. The gap is real, the urgency is real, and the appetite for change is real. But the bottleneck is upstream of where most organizations are looking. Skills cannot be developed in the absence of a strategy that defines them, and strategy cannot be developed in the absence of leaders trained to think strategically about AI. The two problems collapse into one.
47% of leaders cited undefined strategy as their single biggest barrier. Organizations that invest in skills training without first defining where AI fits in the operating model will spend money on a workforce that does not know what to do with what they learn. Strategy first, skills second.
Middle managers are the leverage point. They are also the most cited training need at 47%. The fastest path to organizational productivity gains is not at the top or the bottom of the org chart. It is in the layer that turns intent into action.
53% of leaders quietly believe AI will eliminate more jobs than it creates. Employees can read that belief in their leaders. Training programs that do not address what AI is for, who benefits, and what happens to the people doing the work today will face resistance no matter how good the curriculum is.
Prompt engineering ranks last on the list of skills leaders cannot find. AI strategy ranks first. Education providers still selling tool training are competing in the smallest part of the market. The biggest part is unbuilt.
Half the market values university accreditation. Two percent of organizations are partnering with a university for AI upskilling. The credibility exists. The product does not. The institutions that close that gap first will own the category.
The biggest training need is middle management, not the C-suite. Programs designed only for executives are leaving the highest-leverage population unserved.
The 2026 ZAI AI Skills Gap Survey was distributed in April 2026 through the ZAI Advisory Network to senior professionals in leadership and management roles across multiple industries. 113 completed responses were collected between April 1, 2026 and May 1, 2026, and analyzed for this report.
The survey covered eight question areas: skills hardest to find, current upskilling investment, roles needing training, workforce confidence, the single biggest barrier, jobs outlook, the value of university accreditation in provider selection, and an open response. All quotations in this report are drawn from the open response field and are reproduced with light edits for clarity.
We partner with university leaders and organizations ready to make AI readiness a core institutional capability.
Partner With Us