The CEO gap, the distance between the chief executive who talks about AI and the one who uses it every week, came up at every Silicon Valley AI gathering I attended this month. According to the IBM Institute for Business Value’s 2026 CEO Study , the CEO gap is stalling AI adoption. Executives average 1.5 hours a week with AI, and only 25% of workers use it regularly.

This same topic surfaced at HTEC’s Executive AI Dinner, Innovation in Action, Dreamforce, the AI Infra Summit, Stanford Board sessions and GetOnBoard Week. Around the HTECH table were Darko Todorovic, CTO of HTEC; Marcus East of Autodesk; Peyton Sherwood, CTO of Condoit; and Matt Lungren, a physician building AI Diagnostics for ophthalmology and imaging.

The research puts numbers on it. An NBER working paper, Firm Data on AI , surveyed nearly 6000 executives and found more than two-thirds use AI regularly, but for an average of only 1.5 hours a week. BCG’s AI Radar 2026 found the CEOs leading AI transformation spend more than eight hours a week building their own skills.

Ilana Golan, founder and CEO of Leap Academy, a career-reinvention company, put it plainly: “A CEO can’t delegate adaptability. If the person at the top isn’t learning the tools in front of the team, everyone below reads that as permission to wait.”

HTEC’s research of more than 1500 executives points the same way. They found that 77% agree that “AI value is constrained more by people and processes than tech.” Their Chief Strategy Officer, Lawrence Whittle told me that “as well as starting to use more AI, the CEO needs to do more on communication, enablement, and building trust throughout their business.”

The Consensus From IBM, BCG and Silicon Valley Is Where The CEO Gap Shows Up

A leader at a large manufacturing company described a chief executive who breaks the machine on purpose. The CEO expects work to run through AI before it reachs the corner office and has pushed deployments through when internal analysis said to wait. The results justified it, confirmed the leader.

That CEO is rare. Seventy two percent of CEOs now say they are the main decision makers on AI, double that of last year’s share. BCG places only about 15% in its Trailblazer group, the ones driving transformation through decisive investment and rapid up-skilling.

Sherwood, CTO of Conduit, said his best customers are CEOs who ask for forward deployed engineers, senior developers placed in the customer’s building to see the work and build for it. He added that AI can double a code base in weeks, so nothing should ship until the controls catch up.

The CEO gap has another version, the chief executive who bans the tools. Amy Webb, CEO of Future Today Strategy Group, wrote on Linkedin about a CEO who banned generative AI , only to have their employees paste company data into AI apps on their phones. She ties it to reactance, psychologist Jack Brehm’s term for wanting a freedom more once it is taken away. "The constraint is rarely what does the lasting damage…it’s the reaction by executives she wrote.

Per IBM, The CEO Gap From The Middle

A leader running a multinational group of acquired companies said reluctant teams engaged only after AI went into KPIs and token usage got tracked. A token is the small chuck of text a model reads or writes, and it is how AI usage gets metered. Some engineers then burned tokens on sloppy code that buried in order to ensure their KPIs were met.

A new white paper from All Tech is Human and the Notre Dame-IBM Technology Ethics lab, Two Views of the Same Transformation , heard the same story from below. Mid-level professionals said checking AI output is a growing, unrecognized share of their work, and some called it “babysitting AI slop.” The glue work of a rollout is missing from performance reviews, and women reported carrying more of it. Executives, meanwhile, described middle management as a layer to cut, then worried about “distributed deskilling.”

The IBM Institute for Business Value’s 2026 CEO Study of more than 2,000 chief executives measures the distance between the two views. CEOs estimate only 25% of their workforce uses AI regularly on the job, even though 86% believe their people have the skills to work with it.

IBM and BCG Provide Insight Into Five Moves To Close the CEO Gap

The CEO has to go first and lead the way. But what else needs to happen.

C-Suite Leaders, including the CEO, need to:

Use the tools themselves every week and show your work. When you did math at school, you didn’t get credit unless you showed your work and the same is true here. Open a staff meeting with something you built. People copy what the CEO does faster than what the memo says.

Put Validation hours on the dashboard. Report the hours teams spend checking the AI output beside the productivity gains. A deployment that saves ten hours and adds eight hours of review really only saved two hours. Autonomous AI is right around the corner.

Write glue work into Performance reviews . Training colleagues and turning strategy into team workflows should be named goals with credit given. And you need to understand who carries that load.

Hold a weekly demo day. Find the employees getting real results and have them share their work as well.

Name the middle managers who carry mentorship. List who trains the new hires and who people call when the AI gets it wrong, and give them a voice in the AI decisions. There have been many companies that cut this layer only later to frantically bring it back when they realize the value that left.

Ask your CEO how many hours a week they spend using AI, then ask your middle managers how many they spend checking it. Read the IBM study as the distance between those two numbers is your CEO gap.