In May 2025, Moderna CEO Stéphane Bancel did something few other chief executives had tried. He merged human resources and IT under one executive, a former HR chief. The reasoning was that an organization running thousands of internal AI agents alongside 5,800 employees can no longer design its workforce and its systems separately.

Time will tell if that exact structure is right. What matters is that rewiring any organization for AI will require cross-functional shifts of this magnitude, and that mandate can only come from one person: the CEO.

Why CDOs and CAIOs cannot transform the organization alone

Organizations have spent a decade trying to delegate this accountability. They first invented the Chief Data Officer (CDO) to initiate data-driven transformation, and now the Chief AI Officer (CAIO). Most large firm now has one, but their average tenure is the shortest in the C-suite, barely three years.

This is not to say the CDO or CAIO is a failed experiment. Far from it. These leaders hold the sharpest diagnostic minds in the building. But for the past decade, many of them have been sent into a structural trap: giving them the responsibility to rewire the organization, but not the badge authority to enforce it.

A CDO asked to execute their role with a limited budget and no seat at the table where tradeoffs get settled is basically being asked to redesign a house from the garden shed.

AI produces winners and losers inside the C-suite. Putting AI to work at scale changes who decides what. Functions that never had to share authority suddenly must. Budgets move. The way the organization learns from its own operations gets rewritten.

Every functional leader therefore has an inherent interest in protecting their territory, and a peer cannot strip authority from a peer. A CAIO who asks a Chief Operating Officer (COO) to cede decisions to a system can hope for a good conversation, but the likely outcome is an organization operating exactly as before.

Boardrooms are beginning to recognize this dynamic. In a recent survey of CEOs and board members, fewer than one in ten thought AI strategy should be led entirely by a CAIO without the CEO’s active engagement.

This stance isn’t purely driven by protectionism, but by an existential reality: organizations that fail to centralize this authority will waste millions building fragmented, incompatible AI pilots that scale nowhere.

But the inverse is just as stark. BCG’s Widening AI Gap research found that leadership teams personally engaged with AI were twelve times more likely to land in the top 5% of organizations creating value with the technology. Twelve times! A gap that size is the difference between an effective strategy and a slide deck.

The organizations winning the AI race haven’t fired their data leaders; they have elevated them. The CAIO must report directly to the CEO, acting as the architect of the transformation. But when the blueprints require redrawing the org chart, they must operate on borrowed authority. The CEO must be the one holding the pen.

CEO ownership does not mean CEO execution

The honest objection here is capacity. A CEO’s calendar is the most overcommitted asset in the building, and telling chief executives to lead AI sounds like advice from someone who has never seen their schedule. But the objection mixes up ownership and execution.

Nobody needs the CEO evaluating models. But two tasks cannot be pushed down: deciding who decides, and settling the fights between functions that AI has forced into the open.

This requires the CEO to step into the room to veto siloed AI budgets or explicitly tie executive compensation to cross-functional adoption. Everything else cascades to the P&L owners. According to recent surveys, trailblazing CEOs have already found the time, over eight hours a week, because they know a chief executive who wades into delivery becomes a bottleneck, while one who shirks governance causes gridlock.

The job takes less labor than judgment about authority, applied in person, whenever the org chart is being redrawn.

The CEO’s engagement signals whether AI is truly a priority

Accountability for AI already sits with the CEO. When BCG published its AI Radar survey in January, 72% of CEOs described themselves as their organization’s primary decision-maker on AI. More than half of the non-CEO executives polled said the CEO or board should resign if their business loses market share to a competitor with a better AI strategy.

These numbers may surprise those who view AI mostly as a technology upgrade to be driven by IT. But they make sense when we consider that the biggest barriers to AI adoption are not technological in nature.

A Sierra Ventures survey of Fortune 2000 executives found 73% naming organizational mindset as the biggest barrier to adoption, well ahead of technology, cost, or security. And there is exactly one person with full authority over the organization’s mindset.

CDOs and CAIOs can only act as instruments of transformation if the CEO backs them up with their own calendar. The workforce is watching too; people read the boss’s calendar as the real strategy memo, and delegation tells them AI is optional.

If the CEO’s AI engagement is a budget approval and a quarterly review, that is delegation dressed up as leadership.

The question in front of the board has changed shape. It is no longer whether the organization has the right CDO or CAIO. It is whether the CEO is willing to personally lead the deepest redesign of decision-making the organization has ever attempted.

You cannot delegate the rewiring of a organization. To make AI work at scale, the CDO and CAIO can draw the blueprints, but the CEO must step into the room with the badge authority to make them real.

About the author:

Sebastian Wernicke, Ph.D., is a Partner at Oxera Consulting, where he leads the Data Science and AI practice, and the author of Data Inspired: Building an Organizational Culture of Inquiry for Lasting Transformation (2026). A leading expert in data and AI strategy, he has over 15 years of experience in helping clients achieve data-driven value creation and transformation. His ability to make complex data topics accessible and engaging has made him a sought-after speaker and facilitator, with his three acclaimed TED Talks reaching over 5 million viewers.