Record CEO turnover, fractured geopolitics, and an AI transition that most companies are failing to profit from require a fundamental shift in the ways we lead and govern. Here is how to excel in this transition.
by Faisal Hoque, Pranay Sanklecha, Paul Scade
Rapid read:
- Leadership is failing to set the ownership, governance, and organizational redesign that convert the acquisition of new technology into a changed business.
- Ownership of this transformation must be distributed across the organization, starting at the top.
- Rewire your organization for continuous change.
The demands on corporate leadership are changing fast, leaving many CEOs and boards ill-equipped to respond to them. Against this backdrop, some of the world’s most accomplished chief executives have concluded they are not the leaders for this moment.
Walmart’s Doug McMillon and Coca-Cola’s James Quincey stepped aside in part because they believed that the next era would demand new energy and a longer runway than they could provide. They were not outliers. Russell Reynolds Associates recorded 234 CEO departures from globally listed companies in 2025 – a second consecutive record year – with exits continuing into 2026 at Workday, PayPal, and The Washington Post.
This is not a story about individual failure. The forces converging on the corner office, the wider C-suite, and the boardroom – geopolitical fragmentation, compressed investor patience, and a technology reshaping strategy, operations, culture, governance, and ethics simultaneously – demand competencies that few leadership teams or boards have ever had to build.
Boards are asking whether the CEO can lead an AI-driven transformation that is unfolding against a landscape in which fragmentation is rewiring supply chains, capital flows, and the rules of digital sovereignty. Similarly, boards themselves are being held to a rising standard of AI competency for the simple reason that you cannot govern what you do not understand. Boards are now expected to be able to read a model risk register, interrogate a business case, and tell a governed deployment from a merely launched one. A board that cannot evaluate an AI strategy on its merits has no sound basis for evaluating the chief executive who brings it.
It is not going to be easy. A recent study of 6,000 senior executives found that while 69% said their companies actively use AI, 90% reported no measurable productivity impact. Meanwhile, Boston Consulting Group finds that only around 5% of companies are generating substantial value from the new technology. Read together, those two findings suggest a failure of leadership rather than technology. The models have been bought and are being used; what is missing is the ownership, governance, and organizational redesign that convert new technology into a changed business.
For organizations to succeed, CEOs and boards must build the capabilities this moment demands by working together. Six imperatives top that agenda.
The gap between AI investment and AI value is not a technology gap
1 – The transformation must be owned – and ownership must be precisely distributed
The gap between AI investment and AI value is not a technology gap. Your competitors can buy exactly the same models you can. What separates the frontier 5% from the laggards resides at the organizational level: the 2026 AI & Data Leadership Executive Benchmark Survey found that 93% of Fortune 1000 data leaders identify culture and change management rather than technology as the primary barriers to AI adoption.
Despite this recognition of where the real challenge lies, most companies still delegate AI to the technology function while the rest of the leadership team monitors progress from a distance. The result is that no one owns the dimensions – culture, workforce readiness, decision rights, identity – that will determine whether an AI initiative succeeds. In most organizations today, nobody is clearly accountable for how AI changes what employees do, how decisions are made, or what the organization comes to believe about itself. Ownership has to be labeled precisely, and it divides three ways. The CEO owns the transformation itself and answers for it personally. The leadership team owns the dimensions that decide the outcome – culture, workforce readiness, decision rights, identity – with a named executive accountable for each. The board owns the standard of evidence it will accept that any of this is working.
When one software company rolled out an AI coding assistant to 28,000 engineers, adoption stalled not because the company’s leadership wasn’t behind the project but because the broader organizational culture viewed AI-assisted coding with suspicion. Research at the firm showed that work labeled as AI-assisted was judged to be 9% less competent than identical work without the label. The result was that workers who adopted the tool risked damage to their careers. That was a cultural failure, not a technical one – and cultural signals are set at the top. CEOs must treat AI as a force that reconfigures how the organization works, owned by every senior leader. Boards must stop accepting deployment metrics as evidence of transformation.
In the C-suite: Give each non-technical dimension of AI – culture, workforce, decision rights, identity – a named owner on the leadership team and review those owners on the same cadence as revenue.
In the boardroom: Stop accepting seats licensed, models deployed, and pilots launched as evidence of progress. Ask instead what work has changed and what the company has stopped doing.
Given that nearly everyone is implementing AI but only 5% are capturing value, the differentiator must be something the tools cannot supply.
2 – Treat governance as a source of competitive advantage, not a tax on it
Given that nearly everyone is implementing AI but only 5% are capturing value, the differentiator must be something the tools cannot supply. That differentiator is the decisions wrapped around every system – where it can operate, how its output is checked, and where you retain the ability to change its course. In a word, governance is the difference between tools and transformation.
This year’s IMD World Competitiveness Ranking showed this at the level of nation-states, where the most competitive economies – Singapore, Hong Kong, and Switzerland – are also among the most carefully governed. The same logic applies at the level of individual companies, because rather than obstructing innovation, good governance accelerates it. Moreover, it steers innovation in the right direction, concentrating investment where the organization can defend the results and catching failures while they are still cheap to fix.
One of us (Faisal Hoque), with colleagues, formalized an approach to this topic in the dual OPEN and CARE frameworks, which provide a systematic methodology for using governance to drive responsible innovation. But whatever approach one uses, it is imperative for boards and CEOs to recognize that rather than being simply a compliance line item, AI governance is the driver of value generation.
In the C-suite: Fund governance from the innovation budget rather than the compliance budget and approve no AI investment whose guardrails are not specified alongside its business case.
In the boardroom: Put AI governance on the strategy agenda, not only the risk committee’s agenda, and require management to show where governance accelerated a decision or killed a weak bet while it was still cheap.
3 – Build a sovereign enterprise
In June, the European Union announced its Technological Sovereignty Package, a coordinated effort to cut Europe’s dependence on non-EU providers across the technology stack. Companies are moving in the same direction: in a recent survey of IT leaders in major economies, 98% called digital sovereignty a priority.
But almost all of that engagement is technical – control over data and infrastructure – and so sovereignty rarely reaches the top of the house: Accenture finds only 15% of organizations treat sovereign AI as a matter for the CEO or board. And that is a dangerous blind spot. A genuinely sovereign enterprise is not defined by the proportion of its tech stack it owns but by its capacity to make, enact, and revise its own choices.
Of course, data and infrastructure sovereignty are the foundation, but two further dimensions determine what that foundation is worth. First is decisional sovereignty, which is the capacity to interrogate and govern the judgments AI systems make and shape. Decisional sovereignty erodes quietly, because heavy reliance on external tools can weaken the very expertise leaders need to tell a sound AI output from a flawed one.
Second is adaptive sovereignty, which is the capacity to change course as models, providers, and the landscape shift; it is lost when a single vendor’s roadmap becomes the company’s strategic horizon. And no organization should assume it can simply demand control when it matters: even the US Department of Defense discovered, in its recent dispute with Anthropic, that access to a provider’s models did not carry the power to set the terms on which they ran.
Boards should start with an inventory of every AI dependency and ask three questions of each: was it consciously chosen or drifted into? Is it actively governed? Could you exit it if you had to? The dependencies that fail any of the three are where the work begins.
In the C-suite: Commission a full inventory of AI dependencies, assign an owner to each, and document a credible exit path for every dependency that touches a core decision.
In the boardroom: Treat sovereignty as a board-level matter rather than an IT one and hold management to the three questions above for each critical dependency.
The board’s role in this transformation is oversight of the whole portfolio, not cheerleading individual projects.
4 – Transform on all fronts, or don’t bother
The productivity gains generated by the adoption of new technology tend to be concentrated in a small number of firms rather than being spread equally across the economy. OECD researchfound the top 5% of “frontier” firms captured productivity gains more than four times larger than the remaining 95% during the digital revolution. The same pattern is repeating with AI. And what distinguishes frontier firms is their willingness to transform comprehensively rather than selectively.
Real enterprise transformation with AI requires six foundations, built in parallel: an innovation pipeline with stage-gated funding; a responsible governance framework woven into operations; an enterprise architecture that supports AI at scale; and the three human dimensions encapsulated by leadership, culture, and workforce capability. There is no single correct starting point, but all six are mandatory; weakness in any one area constrains all the others.
The board’s role in this transformation is oversight of the whole portfolio, not cheerleading individual projects. Lloyds Banking Group offers a working model: a cross-functional control tower that scores and ranks AI initiatives against strategic objectives, enforces stage gates, reallocates resources as conditions change, and abandons projects already underway when better ones emerge. The bank deployed more than fifty generative AI solutions in 2025 and expects £100m in value from these projects in 2026. That is what a repeatable management system – rather than a scattering of pilots – looks like.
In the C-suite: Build all six foundations in parallel under a single portfolio view, with stage gates and the authority to stop funded work when a better use of the money appears.
In the boardroom: Oversee the portfolio, not the projects. Ask which initiatives were stopped this quarter and why – a portfolio with no cancellations is not being managed.
Algorithms can pursue goals, but they cannot themselves determine what is worth pursuing
5 – Make philosophical proficiency a board competency
Sundar Pichai has called the CEO role “one of the easier things” for AI to handle; Sam Altman has said that AI superintelligence will at some point be capable of running a major company better than any executive, including himself. These provocations rest on the assumption that leadership can be reduced to algorithmic operations. That is false. Algorithms can pursue goals, but they cannot themselves determine what is worth pursuing. Defining the identity and purpose of the enterprise – what it is, who it serves, and what it will refuse to do under pressure – involves value judgments that no algorithm can make. CEOs will continue to need to make these value judgments. And it is precisely this kind of judgment that boards exist to oversee.
This means that leaders must develop what we might call philosophical proficiency; specifically, this consists of three practices.
First, treat philosophical literacy as a board competency: boards audit for financial expertise, industry knowledge, and increasingly for technological fluency, but few ask whether anyone at the table can interrogate the foundational assumptions that underpin everything the business does.
Second, require a purpose-and-principles impact assessment for every major AI implementation – what it assumes, what it optimizes for, and whose values shaped its defaults. Boards would never approve a major investment without understanding the financial implications; they should not approve a major AI implementation without understanding the philosophical ones.
Third, conduct an annual alignment review comparing what the organization says it stands for with the assumptions embedded in the tools, partnerships, and processes it adopted over the past year. Where these diverge, the company’s commitments and its operational reality are drifting apart.
In the C-suite: Require a purpose-and-principles assessment alongside the business case for every major AI implementation, and answer for its conclusions personally.
In the boardroom: Audit the board’s own composition for the capacity to interrogate assumptions, and commission the annual alignment review as a standing item rather than a one-off exercise.
Boards demand transformative initiatives from the senior leaders they oversee, but they rarely turn this lens on themselves.
6 – Rewire the board for continuous change
Boards demand transformative initiatives from the senior leaders they oversee, but they rarely turn this lens on themselves. New research from Board Intelligence found that 86% of directors, CEOs, and CFOs surveyed say rigid decision-making frameworks at board level contributed to delayed, rushed, or poor decisions in just the past six months, yet 40% expect little change in how their boards operate over the next five years.
Perhaps the posture was tolerable when the world was stable and transformation was episodic – discrete change programs with well-defined beginnings and endings. But the world is no longer stable, and AI-driven change is continuous: multiple initiatives will be assessed, tested, and deployed in parallel, indefinitely. Governing this transformation requires boards to build their own AI fluency through structured education; to realign CEO evaluation and succession benchmarks against what the AI era actually demands; and to redesign meeting rhythms around forward-looking oversight of a living portfolio.
In the C-suite: Give the board what continuous oversight actually requires: shorter and more forward-looking materials, genuine option sets rather than single recommendations, and access to the people doing the work.
In the boardroom: Rebuild the board’s own operating model – structured AI education, revised CEO evaluation and succession criteria, and meeting rhythms designed around a living portfolio rather than an annual cycle.
The leadership that remains
Underlying all six imperatives is a redefinition of what senior leadership involves. For half a century, the executive has been the decision-maker – synthesizing information, allocating resources, and selecting among options. Machines can now do much of this work faster than we can, and their capabilities are only growing. But what will always remain is the need and ability to determine the terms on which decisions get made: what the organization is, what it stands for, what it treats as true, and what it will not do at any price.
And so the real risk is not that AI replaces the CEO. It is that the capacity to set those terms withers away while everyone is busy marveling at the new horizons the technology opens up.
Original article @ IMD.
