Why Decentralization Disappoints

At Porsche’s 2026 annual press conference, its CEO announced a familiar set of measures: the company would simplify its management structure, remove hierarchical layers and reduce bureaucracy. The intention was straightforward. A leaner organization would accelerate decision-making, reduce costs and allow the company to respond more quickly to changing market conditions. Similar announcements have become common across industries. As markets become more volatile and complex, organizations increasingly attempt to replace hierarchical control with greater autonomy at the team level.

Yet, the results often follow a recognizable pattern. After the first layers of management disappear, decision-making does not become faster. The number of meetings multiplies. Teams spend increasing amounts of time coordinating their contributions across the organization. Instead of greater agility, organizations experience more meetings, increased friction between teams, delays in resolving dependencies, and growing effort spent on alignment. These outcomes are often attributed to cultural resistance or insufficient commitment to new ways of working. A closer examination suggests a different explanation. Hierarchy is not primarily about authority. It is a mechanism for concentrating coordination. The hierarchical layers being removed performed important coordination work across teams. Their removal eliminates authority, but also the mechanisms that performed coordination across the organization. Hierarchy concentrates the coordination tasks required to align teams, resolve dependencies, and translate intent across the organization into defined roles. When hierarchical layers are removed, the coordination functions they carried must be allocated elsewhere. This requires a clearer understanding of what hierarchical structures actually coordinate.

Three Layers of Organizational Coordination

Starting from first principles, organizations exist to coordinate collaboration. Coordination refers to the mechanisms through which contributions are aligned, ranging from direct interaction to structured interfaces and system-level governance. Whenever value creation is sufficiently large or complex to require the involvement of more than one person or team, those contributions must be aligned. In small settings, people communicate directly, adjust their contributions, and resolve issues as they arise. Coordination remains local, informal, and largely invisible. As organizations grow, these conditions change. Once multiple teams contribute to the same outcome, coordination becomes unavoidable, regardless of structure. Relying on individual, ad-hoc communication patterns is still possible, but as the number of participants increases, it becomes excessively time-consuming. Moreover, agreements between individuals do not automatically create cohesion at the team level. Taken together, these bilateral agreements may still contradict one another or work at cross purposes.

Intra-Team Coordination

The first qualitative step in scaling coordination occurs within teams. Moving from individual, ad-hoc alignment to intra-team coordination requires a shared understanding of how work is coordinated and the willingness to accept binding rules within the team. As many self-managed teams experience, establishing this level of coordination can already be challenging, especially when rules are unclear and no one is able to claim authority.

Ideally, intra-team coordination takes place within the team itself. Individuals align their activities and interact directly around shared work. Ownership is clear, and decisions can be made quickly because participants remain close to both the task and to one another. At this scale, coordination relies primarily on social mechanisms: shared understanding, trust, and frequent communication. Teams function as small adaptive systems capable of responding rapidly to local signals. This form of coordination is highly effective, but only within the limited scope of work that a single team can perform. Agile methods originated at this intra-team level, focusing on improving collaboration, adaptability, and delivery within small, cross-functional teams. Approaches such as Scrum and Extreme Programming enabled teams to become highly effective units by emphasizing fast feedback, shared ownership, and continuous improvement. At this stage, agility was largely synonymous with team-level excellence.

Inter-Team Coordination

Whenever multiple teams are required for shared value creation, coordination must extend beyond the boundaries of individual teams. Organizations therefore need mechanisms that enable inter-team collaboration: coordinate how specific teams interact, how their conflicts are resolved, and how their contributions fit together and evolve over time. The coordination task thus expands from local optimization to larger-scale productivity. Inter-team coordination aligns those teams whose contributions depend on one another. Unlike intra-team coordination, which relies primarily on direct interaction, this level increasingly depends on establishing how contributions connect through defined interfaces, agreements, and shared structures. At this point, agile methods begin to show their limitations. While they optimize feedback and alignment within teams, they provide only limited mechanisms for coordinating dependencies across multiple teams. Yet, in larger settings, such coordination is essential, as teams cannot operate independently from one another. In response, organizations introduce additional scaffolding, such as Large-Scale Scrum or the Scaled Agile Framework, to preserve short-term agile value-creation cycles across several teams. These are necessary to coordinate and synchronize the work of several teams. Ironically, it leads to the emergence of yet another caste of managers, albeit with more “agile” sounding titles. Of course, organizations need not adopt these frameworks. They can take inspiration and invent their own way of coordinating large-scale efforts. In non-agile organizations, inter-team coordination often emerges within product divisions through specific operations management and processes. Over time, most organizations develop effective solutions to handle recurring coordination tasks for fixed constellations of teams.

As organizations begin to systematically coordinate the value creation of multiple teams in this manner, the focus shifts from optimizing individual team performance to enabling recurring collaboration between teams. But this is not sufficient in a dynamic environment, where new challenges and opportunities emerge all the time. Such environments call for a more fluid structure of teams: teams need to increasingly work in dynamic constellations, not fixed ones. Coordination therefore has to extend beyond the existing constellations of teams. It must operate at the scale of the organization, establishing the conditions under which teams can connect and collaborate, regardless of prior relationships. This trans-team layer of coordination maintains coherence across the organization while allowing teams to recombine as needed. In doing so, it makes the organization more adaptable and effective.

Trans-Team Coordination

When constellations change and teams need to recombine quickly to respond to new customer needs, then coordination cannot rely on historical precedent or accumulated agreements. Teams then do not have the luxury to establish new collaboration contracts from scratch in a time-consuming process. Nor do they necessarily share the same assumptions about collaboration and alignment mechanisms. Each team may be accustomed to its own set of alignment rules, which do not easily transfer to new constellations. Establishing a new rule set for collaboration then recreates a familiar problem from intra-team coordination: who has the authority to define a new rule set for the new inter-team collaboration? If time is of the essence, leaving this to ad-hoc negotiations risks delaying or undermining the collaboration itself. This is where trans-team coordination becomes necessary. It establishes a shared and accepted basis for collaboration that applies independently of which teams are involved. The basic conditions of collaboration are shared standards that exist across the organization. These standards provide a common basis for teams to quickly connect and collaborate. In large corporations, these standards take the form of HR policies and governance practices. However, they are not designed for a dynamic environment. In a dynamic environment, these standards create barriers to collaboration because they rely on the very things that need to be overcome: assumptions of stability and fixed hierarchies for coordination.

To see how trans-team coordination operates at scale, consider Alibaba’s Singles’ Day, the world’s largest shopping event. Every year, in just 24 hours, Alibaba processes tens of billions of dollars in sales. But the real story is not the volume. It’s the way it happens. Behind the scenes, thousands of independent teams, from logistics, payments, product design, to customer service, are making live adjustments in real time. They’re responding to signals from the market as they happen, coordinating through shared data and platforms for instant cooperation. Singles’ Day has become one of the world’s most dynamic launchpads for new products. In the weeks leading up to the event, brands use Alibaba’s systems to design and test items specifically for that day. During the event itself, successful products can be rolled out and scaled rapidly. No pyramid structure could orchestrate that complexity. No alignment through control could respond that fast. It’s coherence through shared intent and live information. It works because every team knows what matters, sees the same context, and has the autonomy to act on it. They also have a common framework and mindset that facilitates connecting and collaborating in real time across thousands of dynamically forming teams.

That is what an entrepreneurial operating system looks like. One that senses, responds, launches, and learns at the speed of its environment. And that is what modern organizations might aim for when they restructure for the purpose of “decentralization.” The extent to which they can be successful depends primarily on the dynamic needs of their environment, and the interdependencies of structures they try to implement.

The three layers of coordination exist to manage these interdependencies so that the organization as a whole can create value in a dynamic environment, and evolve in sync with new challenges and opportunities. The three layers also differ in their temporal dynamics. Coordination within teams unfolds continuously around the value creation itself, coordination between teams occurs in recurring cycles of alignment, and trans-team coordination operates on longer time horizons, shaping the conditions under which the collaboration potential develops. The question, then, is how to make the three layers of coordination more effective without necessarily resorting to the old hierarchical structures.

The Coordination Challenge at Scale

A closer look at existing solutions to coordination reveals the two main dimensions of the challenge. First, the size and number of teams that are required to co-create value. Second, the stability of their inter-team cooperation needs. The number of teams largely depends on the complexity and size of their shared effort: you cannot build a spacecraft with a two-pizza team. The stability of their cooperation needs depends on the stability of demand for their product: your demand for food remains relatively stable, but your diet and the means of delivering it to you can change rapidly.

This means that for any collaboration challenge, the number of teams required for value creation and the stability of their coordination needs determine the required solution. Different combinations lead to fundamentally different coordination requirements:

When only a few teams are involved, coordination can remain largely local, even under changing conditions. As the number of teams increases, coordination must be structured more explicitly. In stable environments, this structuring can be achieved with traditional management, fixed reporting lines and procedures that align the different contributions in the value stream. In dynamic environments, however, coordination cannot rely on these fixed arrangements to continually create value. When customers or markets demand new solutions, then the different contributions and the ways to create value need to adapt fast. This is where agile methods and self-management frameworks show their strengths and also their limitations. They begin to fail when the number of teams required for value creation passes certain thresholds. It is in this context that hierarchy emerged as a historical solution to large-scale coordination.

When Hierarchy Becomes a Constraint

Hierarchies can effectively structure and scale coordination across large numbers of contributing teams. But they mostly do it through fixed relationships: predefined roles, reporting lines, and escalation paths. In stable environments, this works well. In dynamic environments, it becomes a constraint. The fixed coordinating structures cannot adapt fast enough to changing coordination needs. As this adaptation gap grows, slowly but surely, it increasingly disconnects the governing structure from the operational needs of teams, leading to bad coordination priorities and decision lag. The governing structure then becomes a burden, not the support and enabling function it should be. It is no longer able to effectively balance short- and long-term goals. The organizational structure becomes fragmented, competing for resources and aligning on half-hearted compromises that serve neither productivity nor innovation. Many organizations respond by removing hierarchical layers. But this does not remove the underlying coordination demand. It only removes the structures that previously contained it. The operational dependencies between teams remain. What disappears are the mechanisms that made them manageable. Coordination then shifts from structured roles into the informal space between teams, where it becomes a continuous distraction: negotiating conditions for support, alignment meetings, and the intransparent forming of brittle alliances to win internal, and often personalized, power struggles. Power that was once clearly allocated becomes diffused, and harder to invoke for the long-term viability of the organization. Paradoxically, as the number of “autonomous” teams increases, so do their internal coordination costs. This is the structural challenge at the heart of decentralization.

To summarize, organizations struggle when the autonomy of teams is not matched by equally well-designed coordination mechanisms. Decentralization removes the coordination mechanisms of hierarchy, but until a replacement is found, organizations lose the ability to effectively align the contributions of large numbers of teams.

Reducing Coordination Complexity

One way to solve this challenge without relying on hierarchy is to reduce the need for inter-team coordination altogether by organizing work into self-contained value streams, as exemplified by Buurtzorg, a pioneering Dutch home healthcare organization. Teams are structured so that each unit can deliver value to customers largely independently. Most coordination occurs within the team itself, while inter-team interaction needs remain limited to a small number of topics and occasions. This approach is particularly effective when the number of teams required for value creation is small and their coordination needs are relatively stable. At Buurtzorg, small teams care for their own group of patients. Care planning, scheduling, and treatment decisions are handled within the team itself. Although interaction with central services exists, the delivery of care does not depend on continuous coordination with other teams.

A related logic appears in organizations where value creation requires multiple teams, but the number of interactions between them is deliberately kept small. Here, coordination is not eliminated but contained. W. L. Gore & Associates, for example, deliberately limits the size of its organizational units. The company’s founder, Bill Gore, observed that once groups grew beyond roughly 150 to 200 people, interpersonal communication and shared ownership began to deteriorate. To preserve effective and fast coordination, the company caps the size of individual units and establishes new ones as soon as this threshold is reached. By constraining the number of interactions, the organization limits the growth of coordination complexity while keeping its highly effective social coordination mechanisms intact.

Both approaches reduce coordination complexity by limiting either the number of teams involved or the number of dependencies between them. Both are effective as long as coordination needs remain relatively stable and bounded.

Building Coordination Infrastructure

However, not all forms of value creation can be organized this way. Many products and services require the coordinated contribution of multiple specialized teams, often under dynamically changing conditions. In these cases, coordination mechanisms need to continually evolve to fulfill their purpose, but without overburdening the operational system. This means that the total volume of coordination tasks rises, but overall coordination effort must not. Coordination efficiency needs a large boost. Only a few companies have mastered this challenge. They design their whole organizational system around ultra-efficient coordination infrastructure and highly adaptive value creation. This infrastructure defines how contributions connect: through shared interfaces, transparent commitments, and common standards for collaboration that apply across teams. Specialized tech and interaction protocols make dependencies visible, establish how value flows between teams, and act as the single source of truth for commitments, resources, and results. It allows coordination to scale across large numbers of teams and interactions. In the hyper-dynamic environment of the Chinese market, more such organizations can be found. The Chinese appliance manufacturing giant Haier provides another showcase of coordination infrastructure. Thousands of decentralized micro-enterprises operate as autonomous units responsible for specific customer solutions. A shared digital platform makes capabilities, resources, and contractual relationships visible across the organization, allowing micro-enterprises to discover partners and rapidly form collaborations as needed. Both inter-team and trans-team coordination are fully structured through shared mechanisms. In its recently announced five-year strategy for the coming years, Haier repeatedly calls for more improvements in coordination infrastructure, explicitly prioritizing cross-ecosystem collaboration, shared platforms and common standards as the foundation for autonomy and coherence. Not because Haier is lagging behind in this regard, quite the contrary: they know how much more they can gain through progress in this field.

Decentralization Does Not Eliminate Coordination

Behind the current trend towards decentralization lies the assumption that removing hierarchy and shifting to smaller autonomous teams will make organizations more adaptive or more productive. In doing so, they inadvertently also remove the mechanisms of efficient coordination. But large-scale decentralization does not eliminate the need for coordination. It shifts it. When hierarchical layers are removed, coordination has to move out of fixed roles and reporting lines and into the structures and rules that connect operational teams. The long-term viability of organizations requires that these new coordination mechanisms are made transparent, clearly governed and able to effectively balance short- and long-term needs. Without such coordination infrastructure, decentralization cannot succeed at scale and will likely be reversed. Embedding highly effective coordination routines, training and skill development into the interaction infrastructure of organizations makes large-scale decentralization viable. Without it, decentralization does not scale but collapses back into hierarchy.

Janka Krings-Klebe and Jörg Schreiner are strategy experts, speakers and lecturers specializing in ecosystemic organizational transformation and operating models for complexity. They are co-founders of co-shift GmbH, where they work with organizations to move from hierarchical control structures toward adaptive, ecosystem-based forms of value creation. They are the authors of The Antifragile Organization: From Hierarchies to Ecosystems (2025), for which they were shortlisted for the 2025 Thinkers50 Strategy Award.

As algorithms move from advising organizations to acting through them, leaders need to separate algorithmic powers. No single system should be able to concentrate the power to set the rules, execute them, and judge its own results. Independent counterpower capable of constraining the authority delegated to algorithmic systems is a condition of the legitimate exercise of algorithmic power.

Algorithms Acquiring Power Are Not New

In 2019, Deanna Amato discovered that the Australian government believed she owed it thousands of dollars when it seized her tax refund.

The debt had been generated through what became known as Robodebt, an automated welfare-compliance program that compared tax-office income data with social-security records. Rather than establishing what recipients had actually earned during individual two-week benefit periods, the system could average annual income across those periods and use that figure to calculate an alleged overpayment.

For Amato, the assumption was wrong. The resulting debt was nevertheless treated as real. The Federal Court declared the demand for payment invalid.

Once software can encode rules, operationalize them at scale, and produce conclusions that institutions treat as authoritative, the relevant question is no longer simply whether the algorithm is accurate. It is how much power has accumulated around it and who possesses the power to contradict it.

What’s New Is Algorithmic Power Becoming Normalized Authority

Fast-forward. In May, Senbin Hou walked into an Immigration and Customs Enforcement office near Seattle for a scheduled check-in. He left in custody.

Hou, a Chinese national whose immigration proceedings were still pending, had previously been released and enrolled in ICE’s Alternatives to Detention program. Part of that program required him to use a mobile application to complete biometric check-ins through facial recognition.

According to records maintained by ICE’s contractor, Hou had failed to complete 15 of those check-ins. Hou told a different story. He said that technical problems with the application’s facial-recognition function repeatedly prevented him from completing them, and that each time this happened, he contacted ICE to report the problem.

The system recorded noncompliance. Hou described technology failure. ICE acted on the former.

After what it described as a routine review of his record, the agency revoked Hou’s release and arrested him when he appeared voluntarily for an in-person appointment. For weeks, a database entry had become a deprivation of liberty.

In July 2026, a federal judge ordered Hou’s immediate release. The court found that his detention violated due process and emphasized the substantial risk of erroneous deprivation when the government could act on recorded biometric failures without first allowing the person affected to contest them.

Both Robodebt and the system governing Hou were algorithmic systems, but they represented very different generations of algorithmic technology. Yet while the technological sophistication had changed considerably, the governance structure around how algorithmic authority was exercised had not.

In both cases, a rule defined what counted as compliance; an algorithmic system translated messy real-world circumstances into a standardized administrative determination; that determination could trigger state action; and the individual subject to it lacked a meaningful opportunity to challenge the system-generated premise before serious consequences followed.

In effect, this concentrates algorithmic power across the chain from rule-making to execution and judgment.

Algorithmic Power Is Scaling Faster Than Counterpower

If we think this concentration of algorithmic power can have life-altering consequences only in the public sector, we are wrong. The private sector may demand even greater attention.

A 2026 Grant Thornton survey of 950 senior executives from midmarket and large organizations across ten industries, including banking, construction and real estate, technology and telecommunications, insurance, and energy, all sectors in which algorithmic power can materially affect people’s lives, found that 74% of organizations were already giving agentic AI access to their data and business processes. Five percent of organizations were already allowing AI agents to execute high-stakes decisions without human review. Yet 78% of senior executives lacked strong confidence that their organizations could pass an independent AI-governance audit within 90 days, and only 20% had a tested AI incident-response plan for an AI failure.

What those numbers imply when combined is striking. If 74% of organizations are giving AI agents access to data and business processes, while 80% lack a tested incident-response plan, then even under the most favorable possible distribution, and allowing for rounding in the reported percentages, at least about 53% of respondents must fall into both categories, potentially rising to about 74%. In a sample of 950 senior executives, that represents roughly 504 to 703 respondents whose organizations give AI agents access to data and business processes while lacking a tested failure-response plan.

Apply the same arithmetic to governance readiness: with 74% granting agentic access and 78% lacking strong confidence that they could pass an independent AI-governance audit, and similarly taking the reported percentages conservatively, at least about 51% of respondents must again fall into both categories, potentially rising to about 74%. In a sample of 950 senior executives, that represents roughly 486 to 703 respondents. These are not correlations inferred from the data, but the range of overlap mathematically implied by the reported percentages.

And the 5% permitting AI agents to make high-stakes decisions autonomously should not be dismissed as a small number. Applied to a sample of 950 senior executives, that represents roughly 48 respondents whose organizations allow algorithmic decisions to acquire a significant degree of operational finality. The greater that finality, the greater the risk that algorithmic power becomes difficult, costly, or even impossible to reverse. In the high-stakes contexts represented across these industries, that can mean decisions affecting whether someone receives credit, gains access to housing, retains access to a digital or communications service, qualifies for insurance, or receives or can afford energy.

The picture is clear.

The governance gap lies in the mismatch between the power organizations delegate to algorithmic systems and the governance infrastructure available to constrain that power.

The direction of travel is clear.

The central challenge is to govern the power acquired by algorithmic outputs, which should be inseparable from ensuring that meaningful counterpower exists before those outputs become power exercised over people’s lives.

The Powers of Algorithm We Mistake for Technical Functions

Every consequential algorithmic system sits within three functional forms of power.

The first is algorithmic legislative power: the ability to encode the rules. A threshold, objective function, ranking criterion, or system prompt may look technical. But these choices determine how attention, suspicion, opportunity, workload, capital, and risk are distributed. A corporate rule does not cease to be a rule because it is expressed as a parameter.

The second is algorithmic executive power: the authority to translate rules into action. Whether an algorithm recommends an intervention or is permitted to execute it directly by freezing an account, changing a price, advancing or rejecting a candidate, committing a payment, or triggering another operational process, it is the operational arm of algorithmic power. The critical resource is not computing power. It is permission.

The third is algorithmic judicial power: the authority surrounding the determination that an output, action, or person satisfies the relevant rules. It determines whether a fraud alert was justified, reviews rejected applicants, decides whether a model failure warrants suspension, hears appeals, and ultimately determines whether a prior decision is genuinely reconsidered or rubber-stamped.

Fused in algorithmic systems, these three powers can produce absolute authority over both people and the organizations that deploy them.

At a time where we are delegating even greater authority to algorithmic systems whose capabilities are becoming more sophisticated at a pace measured in weeks, extending the reach of concentrated algorithmic power into an ever-wider range of decisions, governing that power is not an option.

If history has taught us anything, it is that the separation of powers and the establishment of countervailing power are instrumental in preserving rights of appeal, protecting human agency, and ensuring the legitimate exercise of power.

Oversight Should Not Be Confused With Algorithmic Counterpower

Corporate governance already understands the principle of counterpower through separation of power.

Traders do not unilaterally set their own risk limits. Companies do not prepare their financial statements and independently audit them. Credit officers operate within authorities established elsewhere.

Yet, the algorithmic concentration of power often takes root in organizational concentration of power.

This occurs, for example, when a technical team defines an algorithmic system’s purpose, builds it, deploys it, evaluates its performance, interprets its failures, and decides whether its outputs should stand; in effect, cumulating functions analogous to policy-setting, risk management, safety, compliance, validation, incident review, and appeal can accumulate within the same chain of command.

Governance failures do not remain outside the technology; they compound within the algorithmic system itself. The organization concentrates authority over the system, while the system concentrates the algorithmic functions inadvertently delegated to it, through which its power is ultimately exercised.

Separation of algorithmic powers is therefore not simply a matter of inserting a human somewhere in the process or assigning AI oversight to another committee. It requires identifying where power has accumulated, organizationally and technically, ensuring that no actor controls the entire chain from rule-making to execution to judgment, while also subjecting each exercise of authority within that chain to productive friction and independent counterpower.

That means separating goal-setting from model building; data stewardship from model optimization; optimization from the limits within which optimization may occur; system design from deployment authorization; product success from independent monitoring; automated decisions from meaningful appeal; and technical capacity to scale from authority to create enterprise-wide exposure.

And the greater the power delegated to an algorithm, the stronger the counterpower organized around it must be.

The Leader Test For Governing Algorithmic Power

Boards and executive meetings do not need to become AI engineering committees. But they need to acknowledge that algorithmic systems do not simply enter the technology stack; they enter the organization’s authority stack. Leaders must therefore keep sight of how power is redistributed, including the authority granted to the algorithmic systems they hire. They must build their map on how algorithmic power reconfigures authority across the organization.
Three questions help provide one.

• What power is actually being delegated to the algorithmic system, and how far does that authority extend within the governance of the firm or institution?
• How is that power distributed around the system, and what prevents rule-setting, execution, and judgment from collapsing into the same hands?
• What independent counterpower can challenge, constrain, or stop algorithmic power when it exceeds its mandate or causes, or risks causing, harm?

That is the premise of Trias Algorithmica.

Ultimately, governing algorithmic power means acknowledging the social necessity of continually revisiting a simple question: what makes the power it exercises legitimate?

The Lifetime Achievement Award is given to someone who has had a long-term impact on the way people think about and practice management. 

A pioneer in the field of business education, Marshall is the global authority in helping successful leaders achieve positive, lasting change in behaviour: for themselves, their people and their teams.  He is the creator of the Stakeholder-Centered Coaching method and the inspiration behind the 100 Coaches global network and has authored or edited over 60 books including four New York Times and Wall Street Journal bestsellers:

• What Got You Here Won’t Get You There (with Mark Reiter, Grand Central Publishing, 2007)
• MOJO: How To Get It, How To Keep It, How To Get It Back If You Lose It (with Mark Reiter, Grand Central Publishing, 2010)
• Triggers (with Mark Reiter, Crown Currency, 2015)
• The Earned Life (with Mark Reiter, Crown Currency, 2022)

“Hyperactive and relentlessly positive, Marshall Goldsmith’s reputation is based on results. His contribution to leadership excellence is second-to-none.”– Des Dearlove, Thinkers50 co-founder.

“Marshall’s impact can be seen in the hundreds of leaders he has mentored, and the thousands they in turn lead. At the heart of his approach is a deep understanding of the human side of leadership. His own humility and openness encourage others to recognise both their extraordinary strengths as well as their vulnerabilities – and, importantly, to ask for help.” – Sanyin Siang, Executive Director of the Coach K Center on Leadership & Ethics (COLE) at Duke’s Fuqua School of Business; Thinkers50 Coaching & Mentoring Award 2019, Coaching Legends 2023, Thinkers50 Ranking.

Marshall featured in the Thinkers50 Ranking between 2007 and 2017 and was inducted into the Thinkers50 Hall of Fame in 2018, upon which the Marshall Goldsmith Distinguished Achievement Award for Coaching & Mentoring was created. He will be presented with the Thinkers50 Lifetime Achievement Award at an exclusive reception at London’s Guildhall on Sunday 1 November, the evening prior to the Thinkers50 London Summit & Leaders50 Awards Gala. Previous recipients of the Lifetime Achievement Award include Lynda Gratton, Rosabeth Moss Kanter, Tom Peters, and Charles Handy.

Celebrating 25 years of Thinkers50, this year’s London Summit brings together the world’s foremost management thinkers and business leaders to challenge assumptions, spark fresh ideas, and shape the future of leadership. Discussions and keynotes will follow three key themes at the heart of management and leadership: purpose, technology, and talent. Networking breaks will facilitate the all-important encounters and conversations over coffee, lunch, or a glass of wine. 

Speakers at this year’s London Summit include Liz Wiseman, author of Multipliers; Jesper Brodin, former CEO of IKEA; Rahaf Harfoush, author of Hustle & Float; Erin Meyer, author of The Culture Map; Ndidi Okonkwo Nwuneli, executive chair, Sahel Group and co-founder, Wealth4Impact; and Kevin Nolan, President/CEO of GE Appliances. VIEW FULL LINEUP

We hope you can join us in London!

Tickets for the London Summit are upon application, via the link below. 

APPLY FOR LONDON SUMMIT TICKETS

London Summit & Leaders50 Awards Gala tickets (including Gala Dinner and presentation of the Leaders50 Distinguished Achievement Awards) may be requested by email to events@thinkers50.com – note availability is limited.

Notable Marshall Goldsmith Content

	What Got You Here Won’t Get You There (with Mark Reiter, 2007)
	Triggers (with Mark Reiter, 2015)
	How Women Rise (with Sally Helgesen, 2018)

LONDON 14 SEPTEMBER 2026 – Thinkers50 today unveils its Leaders50 2026, a biennial list of 50 exceptional leaders from around the world.

What makes Leaders50 unique is who chooses it. Nominations come exclusively from Thinkers50 Ranked Thinkers and recipients of the Thinkers50 Distinguished Achievement Awards – the experts, scholars and authors who study leadership for a living and help define contemporary management thinking. Rather than measuring company size, revenue or media profile, they nominate leaders on the evidence of their impact, contribution, purpose, and commitment to developing others.

That collective perspective gives Leaders50 a depth that conventional rankings cannot replicate. It connects scholarship and practice, allowing nominators to recognise not only visible performance, but also the ideas, behaviours, and purposeful choices that are changing how organisations and teams are led – including work that may not yet command broad public attention.

This expert-led approach enables Leaders50 to recognise outstanding leadership wherever it occurs. The 2026 list includes internationally recognised figures, such as Satya Nadella, CEO of Microsoft, Jensen Huang, CEO of NVIDIA, and Leena Nair, CEO of Chanel, but it also brings important, less familiar stories to a global audience. Among these hidden gems is Jure Knez, founder of DEWESoft and Acrobat City, an entrepreneurial hub in Slovenia; former refugee Mursal Hedayat, founder of the UK-based social enterprise Chatterbox; and Sharon Price John, former CEO of Build-a-Bear and now CEO of children’s apparel company Carter’s. Their inclusion exemplifies the list’s ability to look beyond the most visible and obvious names and identify leaders that the world’s foremost management thinkers believe deserve wider attention.

“The challenges facing the world, organisations, and individuals are enormous. We do not know all the elements needed to tackle these challenges, but we do know that it will require exceptional leadership. Leaders50 shines a light on remarkable leaders who are making a real impact in the world, so that others may learn from them, be inspired by them, and take on the mantle of purposeful leadership.” – Des Dearlove, co-founder, Thinkers50

Those named to the Leaders50 will be honoured at the Thinkers50 London Summit & Leaders50 Awards Gala on 2 November 2026. The inaugural Leaders50 Distinguished Achievement Awards will also be announced on the night. Apply for accreditation.

2026 Leaders50

The full Leaders50 listing (in alphabetical order) is:

Tarang Amin
CEO, e.l.f. Beauty

Bill Anderson
CEO, Bayer

Ana Paula Assis
Senior VP and Chair for IBM Europe, Middle East, North Africa and Asia Pacific

Mary Barra
CEO, General Motors

Subha Barry
President, Seramount

Ed Bastian
CEO, Delta Airlines

Cécile Béliot
CEO, Bel Group

Olivier Blum
CEO, Schneider Electric

Jesper Brodin
Senior advisor to the IKEA Foundation; chair of The Earthshot Prize

Niels Christiansen
CEO, LEGO Group

Ken Denman
Managing Director, Sway Ventures

Saori Dubourg
CEO, Greiner AG

Aicha Evans
CEO, Zoox

Margo Georgiadis
CEO-partner, Flagship Pioneering

Mursal Hedayat
CEO and founder, Chatterbox

Jensen Huang
Co-founder, President and CEO, Nvidia

Merete Hverven
CEO, Visma

Andrea Illy
Chair, illycaffè

Sharon Price John
President and CEO, Carters

Francine Katsoudas
Executive Vice President and Chief People, Policy & Purpose Officer, Cisco

Ken Keller
Chair and CEO, Daiichi Sankyo, Inc.

Jure Knez
President and co-founder, DEWESoft

Li Bin
Founder, Chairman, and CEO, NIO

Mike Mahoney
Chair and CEO, Boston Scientific

Raja Al Mazrouei
CEO, Etihad Credit Insurance

Frédéric Mazzella
Founder of BlaBlaCar and Dift

Kenneth McGrath
CEO, Lidl

Doug McMillon
Former President and CEO, Walmart

Florent Menegaux
CEO, Michelin Group

Michael Miebach
CEO, Mastercard

Francesca Moriani
CEO, Var Group

Satya Nadella
CEO, Microsoft

Leena Nair
CEO, Chanel

Vas Narasimhan
CEO, Novartis

Kevin Nolan
President and CEO, GE Appliances 

Jacqueline Novogratz
Founder and CEO, Acumen

Ndidi Okonkwo Nwuneli
Executive Chair, Sahel Group; co-founder, Wealth4Impact

Miranda Qu
President and co-founder, Xiaohongshu

Cindy Rose
CEO, WPP

Leigh-Ann Russell
CTO and global head of engineering, BNY

Etienne Salborn
Founder, Social Innovation Academy

Dilhan Pillay Sandrasegara
CEO, Temasek

Ted Sarandos
Co-CEO, Netflix

Jim Hagemann Snabe
Chairman, Siemens AG

John Stankey
Chair and CEO, AT&T

Nick Studer
President and CEO, Marsh Risk

Lisa Su
Chair and CEO, AMD

Jessica Tan
Executive Vice-President and President, Sun Life Canada

Warner Thomas
CEO, Sutter Health

Suresh Venkatarayalu
SVP and CTO, Honeywell; President of Honeywell Connected Enterprise

VIEW THE 2026 LEADERS50

ABOUT THINKERS50

Founded in 2001 by business journalists Des Dearlove and Stuart Crainer, Thinkers50 began as the first global Ranking of Management Thinkers. Over the past quarter century, it has grown into a worldwide platform and a vibrant Community. Today, Thinkers50 encompasses the Thinkers50 Distinguished Achievement Awards alongside the Hall of Fame, Radar, Coaching Legends, Leaders50, Coaches50, and curated Booklists. The Thinkers50 London Summit & Awards Gala has been described by the Financial Times as the “Oscars of management thinking.”

MEDIA CONTACT

Visit Thinkers50 Press Room

To get in touch please email: enquiries@thinkers50.com
thinkers50.com

For decades, boards have treated experience as a competitive advantage. The underlying assumption has been simple: the more seasoned the directors, the better the board’s judgement. But what if that assumption no longer holds? In today’s business environment, knowledge has a much shorter shelf life than it once did. Artificial intelligence, geopolitical shifts, new regulation and rapidly evolving business models mean that expertise acquired over a successful career can become outdated surprisingly quickly. Experience still matters, but unlike financial capital, its value is no longer steadily compounding. In many areas it is quietly depreciating, as the world changes faster than even the most accomplished directors can keep pace with. Rod Adkins, board chair at Avnet and board member at United Parcel Service (UPS), PayPal, and W. W. Grainger summarised this idea:

“Going forward, boards must think about a dynamic versus a static model. In the past, board members may have been on the board for as much as twenty or thirty years… High‑performance boards will always have to focus on the types of skills, backgrounds, and experiences of the directors needed at any time. It must be consistent with the pace of change in the marketplace.”

We call this the half-life of board expertise: the time it takes for the strategic relevance of a board’s collective knowledge to diminish. The implication for how boards should be staffed are significant. Adapting to a shorter half-life of expertise does not simply mean refreshing boards more often. It means rethinking the role that different directors play. Some provide continuity and institutional memory. Others contribute specialist expertise that is particularly valuable during specific phases of a company’s development.

Balancing stability with renewal

The challenge is therefore not to choose between stability and renewal, but to combine them deliberately. Boards still need directors who understand the company’s history, culture and strategic context. That perspective can only be built over time and remains essential for effective oversight. At the same time, boards increasingly need directors who bring fresh perspectives and deep expertise in emerging technologies, new business models or geopolitical developments. The future belongs neither to static boards nor to constantly changing ones, but to boards designed to balance institutional memory with intellectual renewal. The balance between stability and change does not mean regular replacement of all the board’s members. Instead, boards should maintain a stable core deliberately creating room for directors whose contribution is tied to a particular strategic challenge. Not every board seat needs to serve the same purpose. Some directors provide continuity. Others contribute expertise that is valuable precisely because it is time-bound. Christian Trümpler of Partners Group described how this works in practice:

“In one of our investments, we knew that we had to reposition the brand within the first two years because although it was successful in Europe, it was not known in the US. So, rebranding was important. We needed an American with brand experience on the board to support this project because we had to reposition it globally. But from the outset, we also knew that the best person to help us with rebranding would not be the same person who would continuously drive our go‑to‑market efforts in the US. We communicated this to the board members right from the start: we need one person on this board to help us rebrand, then in two years’ time, we will need somebody else to get us into Walmart and Target. That was the plan from the beginning, and it’s now being executed one‑to‑one without any bad blood because it was made clear to everyone from the start.”

Curiosity as a governance capability

The priority is no longer simply to recruit accomplished directors. It is to ensure that the board’s collective expertise evolves as quickly as the environment it is expected to govern. This changes the questions boards should ask themselves. Rather than focusing only on whether they have the right mix of skills and experience, they should also ask how quickly they update their assumptions, expose themselves to new ideas and recognise when long-held beliefs no longer fit reality. As Debbie Hewitt, chair of England’s Football Association put it: “For board development to work, you need that the board openly accepts it does not know everything, that it is curious about finding out different ways of learning, and that it is open to acknowledging where the weaknesses might be.” In a world where the half-life of expertise is shrinking, curiosity becomes as important a governance capability as experience itself. The boards that remain effective will not be those that assume they already have the answers, but those that deliberately seek out new perspectives and continually renew their collective understanding of the world around them.

The most effective boards of the future may not be those with the deepest reservoir of experience, but those that recognise when experience is no longer enough. As the half-life of board expertise continues to shrink, the real competitive advantage will lie in a board’s ability to renew its collective knowledge while preserving the judgement that comes from understanding the business. The next frontier in governance is therefore not simply board refreshment, but designing boards that evolve as deliberately as the companies they oversee.

In 2015 Thinkers50 presented an award for putting ideas into practice to Zhang Ruimin, then the CEO of the Chinese white goods company, the Haier Group. The work of Haier first came to our attention in 2005 when it deconstructed itself into a series of self-managing business units. As part of this, Haier established the inverted-pyramid organisation, a way of putting users first and reorganising the structure of the company.

Then in 2012 Haier effectively eliminated 12,000 middle manager roles and transformed the company’s original pyramid into a flat structure. The headlines screamed “Debureaucratisation! The smashing of middle management!” And for once they were entirely accurate. From Haier’s 80,000 employees, those with entrepreneurial willingness and capability were screened to form more than 4,000 autonomously operating micro-enterprises (MEs). The former middle managers had two choices: either seek opportunities elsewhere or become entrepreneurs within the Haier system, becoming ME owners or core members, serving as resource providers for the MEs.

Over the years since then we have witnessed at first hand Mr Zhang’s relentless quest to create an organisation truly fit for the twenty-first century and his similarly relentless curiosity.

Haier came from near bankruptcy in a semiderelict factory to become a modern international corporation. In 2025 it achieved a record high 426.8 billion RMB ($63 billion) in global revenue and 32.2 billion RMB ($4.75 billion) in total profit. But the issues that Haier has encountered and approached in its unique way are truly universal. Inefficient and indifferent organisations, and unhappy people within them, exist throughout the world. And at the same time, wonderfully imaginative, progressive, and successful organisations flourish. 

The Haier philosophy was first articulated in 2005 and is called RenDanHeYi. This Mandarin word is a compelling distillation of what is important in organisations. In RenDanHeYi Ren refers to the value of employees, Dan refers to the value of users, and HeYi means that the value of employees is realized through the value they create for users. The fusing of the creativity of employees and the needs of users lies at the heart of RenDanHeYi. In the language of Haier, the objective must be to continually insist on Zero Distance between employees and users, to continuously iterate and upgrade that relationship.

The spirit of continual improvement and iteration lies at the heart of Zhang Ruimin’s worldview. He is now the company’s chairman emeritus and remains intent on continually pushing the ideas and practice forward. We have spent the last few months working on a book together which combines the insights and experience of Haier with other inspiring examples of progressive management in organisations worldwide.

Working together we have sought to truly understand how ideas and best practices fit together. The result is Theory R: Rethinking How People and Organizations Work which will be published by Amplify later in 2026. The concept of Theory R takes the ideas and practices of RenDanHeYi as its foundation—as well as its first letter. Theory R is a distillation and translation of the original philosophy, intended to facilitate understanding and the application of the ideas in practice globally. It reflects the encouraging reality that organisations worldwide are taking these ideas and using them to reconfigure their own contexts.

Theory R aims to enable organisations to move toward a state of self-organisation, with autonomous individuals and lifelong users in an infinite loop of connection and collaboration to fuel constant change, learning, refinement, and performance. More pithily, Theory R is focused on how we can maximise human value in organisations.

And while we label it a theory and think of it as a philosophy, Theory R is practical. It is working today throughout the world. 

The primary distinction between Theory R and classical management models lies in their ultimate purpose. The classical model is material centric, reflecting what may be called a product-based economic paradigm. This paradigm pursues the appreciation of capital and wealth, benefiting only shareholders. Because capital belongs to shareholders, the model is inherently oriented toward shareholder value. The flaw is clear: shareholders can only share value, not create it. This explains why the classical model lacks an intrinsic engine for sustainable growth.

Though both are business models, Theory R and classical approaches differ radically in their development philosophies. We cannot comprehend Theory R through the lens of classical thinking, just as we cannot understand mobile phones through the logic of landlines. They represent two fundamentally distinct concepts.

Theory R is people centric, seeking the appreciation of human value—and that necessarily depends on human awakening or, more precisely, on self-enlightenment. By dismantling bureaucracy, Theory R enables every individual to move from being a mere executor to being a self-driven entrepreneur—an act of self-enlightenment. 

Stuart Crainer is cofounder of Thinkers50.

For two decades, Chinese white goods giant Haier has been rethinking the fundamentals of how organisations work – dismantling bureaucracy, replacing hierarchy with thousands of autonomous micro-enterprises, and putting employees directly in touch with the users they serve.

Now Haier chairman emeritus Zhang Ruimin and Thinkers50 co-founder Stuart Crainer are taking those ideas a step further. Their forthcoming book, Theory R: Rethinking How People and Organizations Work, published by Amplify later this year, distils the philosophy and practice of Haier’s RenDanHeYi model into a new framework for organisations everywhere – one centred not on maximising shareholder value, but human value.

Announced every two years, the Coaches50 list identifies and celebrates 50 executive coaches who are making a significant contribution to leadership, management, and organisations across the world.

“We see the coaching community as playing a vital part in our ecosystem and having a big impact on thought leadership,” explains Thinkers50 co-founder, Des Dearlove.  

“As well as being originators of ideas, coaches are the honeybees of best practice: cross pollinating the world of practitioners with the freshest and best ideas wherever they come from.”

“Coaches50 celebrates some of the people doing the most thoughtful and consequential work in coaching today,” adds Charlotte Jordan, CEO of Coaching.com.

“They are not simply contributing ideas; they are challenging the profession to think differently, raising the quality of practice and expanding what coaching can make possible. We’re incredibly proud to support Thinkers50 in recognising their contribution.”

2026 Coaches50

In alphabetical order, the 2026 Coaches50 are:

1. Jennifer Garvey Berger
2. Ann Betz & Ursula Pottinga
3. Robert Biswas-Diener
4. Peter Bregman
5. Valorie Burton
6. Ron Carucci
7. Shirzad Chamine
8. Peter Chee
9. Nihar Chhaya
10. Alisa Cohn
11. Gena Cox
12. Bruna De Palo
13. David Drake
14. Maja Djikic
15. Scott Eblin
16. Chester Elton
17. Tasha Eurich
18. Hortense le Gentil
19. David Goldsmith
20. Adrian Gostick
21. Ruth Gotian & Andy Lopata
22. Janet M. Harvey
23. Peter Hawkins
24. Jesse Hsu
25. Jeffrey Hull & Margaret Moore 
26. Amy Jen Su
27. Shweta Jhajharia
28. Atchara Juicharern 
29. Ben Koh
30. David Lancefield
31. Jacquelyn Lane & Scott Osman
32. John Mattone
33. Sharon Melnick
34. Julia Milner
35. Haesun Moon
36. Reeta Nathwani
37. David Noble
38. Lisa Orbé-Austin
39. Divya Parekh
40. Jonathan Reitz
41. Marcia Reynolds
42. Neha Sangwan 
43. Liz Simpson
44. Tracy Sinclair
45. Nicky Terblanche
46. Bonita Thompson
47. Brian O. Underhill
48. Hitendra Wadhwa
49. Caroline Webb
50. Andrew White

Members of the Coaches50 are selected according to three key levels of criteria:

• experience and professional reputation
• influence and impact
• activity and engagement

Within these three levels are considered key factors such as client impact, peer recognition, media presence, published content, contribution to the coaching profession, and engagement within the global coaching community. The coach’s qualifications, depth of experience, professional development, and expertise in a specific area of coaching are also taken into consideration.

MORE ABOUT COACHES50

ABOUT THINKERS50

Founded in 2001 by business journalists Des Dearlove and Stuart Crainer, Thinkers50 began as the first global Ranking of Management Thinkers. Over the past quarter century, it has grown into a worldwide platform and a vibrant Community. Today, Thinkers50 encompasses the Thinkers50 Distinguished Achievement Awards alongside the Hall of Fame, Radar, Coaching Legends, Leaders50, Coaches50, and curated Booklists. The Thinkers50 London Summit & Awards Gala has been described by the Financial Times as the “Oscars of management thinking.”  

MEDIA CONTACT

Visit Thinkers50 Press Room
To get in touch please email: enquiries@thinkers50.com
thinkers50.com

APPLY FOR ACCREDITATION

2 November 2026
Guildhall, London 

IMD Professor Mike Wade argues that the biggest challenge facing organisations is not adopting AI, but leading the cultural and organisational change that comes with it. In this interview he explains why so many transformation initiatives fail, how leaders can use AI to improve their own decision-making, and why organisations must focus on value creation rather than technology for technology’s sake. He also explores the ethical challenges surrounding AI and why resilience may prove just as important as agility in an increasingly volatile world.

Mike is Professor of Strategy and Digital at IMD and Director of the Global Center for Digital and AI Transformation. He is also the author of Digital Vortex, Hacking Digital, Alien Thinking, Gain: Demystifying GenAI for office and home, and Twin Transformation.

Watch the full interview or read the transcript below. 

This interview took place at IMD’s signature Orchestrating Winning Performance program in Lausanne, 2026.

Transcript

Des Dearlove:

Hello, I’m Des Dearlove, co-founder of Thinkers50. My guest today is Mike Wade. Mike is professor of strategy and digital at IMD and director of the Global Center for Digital and AI Transformation. A leading thinker on digital disruption and transformation, his books include Digital Vortex, an award-winning bestseller on disruptive competition, Hacking Digital, a practical guide to accelerating business transformation, and Alien Thinking on turning breakthrough ideas into reality. In 2025, Mike published Gain: Demystifying GenAI for office and home, exploring generative AI’s impact on personal and professional life. And in the same year, Twin Transformation, examining how AI and sustainability can be combined to create long-term competitive advantage. Mike, welcome.

Mike Wade:

Thank you. It’s a pleasure to be here, Des.

Des Dearlove:

You’ve been writing about digital disruption and transformation for many years.

Mike Wade:

Yes.

Des Dearlove:

What has changed most in the way leaders understand disruption and what have they still not really got to grips with?

Mike Wade:

I think when we wrote Digital Vortex 10 years ago now, disruption was still something that most executives thought happened to other people. And so they were still coming to terms with the idea of disruption, so we unpacked it in that book. I think today that’s no longer the case. I think disruption has become very much a factor of everyday life for executives so they’re more used to it. What hasn’t changed, I think, as we’re focused very much on the digital side and now AI, of course, is that still today, most executives and most organisations treat this as a technology challenge rather than what it really is, which is a change management challenge and often a cultural transformation as well.

Des Dearlove:

So I think you’ve said 87% of digital transformations actually fail, so why are the statistics so scary and so bad?

Mike Wade:

Yeah, we keep track of those statistics as many organisations do. We’ve given up trying to focus on a precise number, but it’s over 80% and it’s really hard. These transformations are really tough. The technology part is difficult, but even if you get the technology part right, you can still fail. Part of the issue is with expectations being unrealistically high. Part of it is that the organisational change doesn’t go as well as organisations like. Part of it is because in the rush to adopt the latest technology like AI today, organisations end up with a really fragmented mess of different things going on rather than what they should be doing, which is prioritising based on value creation.

Des Dearlove:

Now, I’ve been sort of analysing this sort of research for a long time. I mean, 80% of change management… We used to hear 80% of change management transformations didn’t work or failed. The number doesn’t seem… If anything, it’s getting worse, you’re putting it into the high 80s. Does the digital thing make it more precarious or is it… I mean, why aren’t we moving the needle on this?

Mike Wade:

Why are we not learning from past failures? And it is surprising to me constantly that organisations don’t learn. They do make the mistakes over and over again. I think what technology does is it accelerates the process. So the change, maybe you had a little bit more time in the past to adapt to a changing world and do what you need to do. That’s been very much compressed. And I think the stakes are higher because the opportunities are larger, but the threats are also larger, so the whole game is getting harder to play. And even if organisations are getting better at playing the game, because it’s getting harder to play, the success rate doesn’t necessarily improve.

Des Dearlove:

Obviously, everyone’s talking about AI and gen AI and what do leaders need to understand about it beyond the kind of the hype? What is it that they really need to get their heads around?

Mike Wade:

I think they need to get their heads around the fact that unlike other waves of technology disruption, this one is a multipurpose disruptor. It’s a general purpose. It has the ability to disrupt by itself, but also to disrupt many other areas, unlike let’s say blockchain, which can be disruptive within very specific niche areas, or 3D printing, or augmented reality. They can be disruptive but I think AI and particularly generative AI has this broad scale ability to shift and change the way things work and how organisations operate. And that’s new and it’s really important for leaders to get their heads around that.

Des Dearlove:

I think the other thing is that if you’re the CEO of an organisation, perhaps you understand that everybody else needs to change the way they work because this gen AI is going to make a big difference, but how does it actually… I mean, what are CEOs themselves doing? Because it can be like everybody else should be taking the medicine except me, I carry on doing what I was doing before, but how do the best CEOs that you’ve seen use AI in their everyday leadership role?

Mike Wade:

Yeah, I think you’re right. I think that the best CEOs, the best leaders are practitioners. They’re using it. They’re engaging with it. They’re not outsourcing it or delegating it to others. In fact, this week we’re running a stream here focused exactly on that. So what can executives get personally from these technologies? How can they use these technologies to become more efficient, to become more productive, to make better decisions, to get more time back into their week? There’s so many practical ways that generative AI and related technologies can help these executives do what they do better. And the ones that leverage that and explore that, I think are the ones that see the potential that then can be leveraged more in a bigger way throughout the organisation.

Des Dearlove:

So what sort of things are you talking about? Can you give me an example of how a CEO might deploy AI in their own life, in their own world, to perhaps give themselves more time, perhaps make better strategic decisions or be better at analysing data?

Mike Wade:

I think that the forward-looking leaders are using AI as a kind of a virtual chief of staff. So a chief of staff, there’s an organisational part of that. So helping to simplify the calendars, to prepare better for meetings, to organise the inboxes, things like that, but also to stress test decisions. So you can create agents now with certain personas and certain roles. What would a competitor think of this challenge or this decision I’m thinking of making? What would a customer feel? What would somebody who’s a finance executive or technology executive or somebody who’s an extreme in this area or that area? So I think these executives are essentially placing AI between them and all the things that they need to do and getting a little bit of additional insight to prepare them better for all the things they need to do on a daily basis.

Des Dearlove:

Can you see a situation where someone would have an AI board member?

Mike Wade:

There are some examples today of boards that are experimenting with AI board members. I think the time when you have an independent board member with fiduciary responsibility who’s an AI, that’s probably still a little bit into the future. But I do think that AI can help board members to be more effective by anticipating questions to ask management or that management may ask them, to summarise documents, to stress test decisions. Again, as I mentioned before, to play the role of different personas that could improve the discussion that boards have. So I think AI has a very, very strong role to play for boards. And yet, Des, it’s interesting, we’re doing some research now into this question, how boards are using AI. And there’s quite a strong sense within many boards today that if you use AI, you’re cheating, that boards are paying a lot for board members and for their brains and they shouldn’t be using AI because that’s not what they’re there for, which to me is a missed opportunity, not to replace a board member, but to augment what the board members do for boards and for organisations.

Des Dearlove:

I mean, this is bringing us nicely to another one of your sort of topics, which is AI and ethics, but I’m going to come to that. But I mean, I think you’re right. There is this sense almost that either you have AI in your life and you’re forward-looking or that there’s a kind of a pushback, that there’s an element of you’re cheating if you’re using AI. We seem to be hung up on this a little bit. How do we get through that particular kind of forest and then we’ll talk about the ethics of AI and how we make sure there’s some guardrails.

Mike Wade:

Yeah. I think there is a real thing, this AI shaming.

Des Dearlove:

AI shaming. Yeah.

Mike Wade:

We see it in organisations where people will say, “I don’t think it’s right that if an email is sent with my name on it that is written by AI, it’s just not ethically right.” Or they’ll say, “My kids are not allowed to use it in school. Why is it okay for me to use it at work? Don’t these things hallucinate all the time?” And I think a lot of that comes from just a lack of understanding of what the technology is, the strengths, the weaknesses, how it works. I think a lot of that can be addressed through training, helping people, executives or frontline workers, whoever, really understand what the technology is, how it works, how it’s different than traditional AI or other technologies, what are the benefits, what are the risks? And I think armed with that knowledge, they’re much better able to go out and use the technology in the most effective way.

Des Dearlove:

So let’s talk about guardrails, let’s talk about ethics. Where are we in your view in that whole debate? How are we going to ensure some of the nightmare scenarios that obviously are being rolled out all the time? How do we protect ourselves and how real are the risks, I guess?

Mike Wade:

How real are the risks is something that’s mostly down to a matter of opinions and when it comes to opinions, you have them all over the place. So it’s very, very difficult to separate the signal from the noise and there’s a lot of noise around this. This is actually something we looked at two years ago. We launched something we call the AI Safety Clock, which is based on the Doomsday Clock, which as you all know, was something set up after the Second World War to measure the likelihood of nuclear war. And so we did something similar for AI safety and for us, midnight, so it goes from 11:00 to midnight, is uncontrolled AGI. So AI that is as intelligent as we are generally, which is out of our control. And we tried to take a very, very fact-based approach. So collecting a lot of data, looking through the legislation, looking through the advances in the technology, monitoring things like agentic AI, which basically gives AI more autonomy, physical AI and all the rest of it.

And when we launched the clock a little over two years ago, we placed it at 31 minutes after 11:00, so 29 minutes to midnight. So going from kind of medium to getting into that higher risk, since then we’ve moved it 11 minutes closer to midnight. So we actually see the risk going up and unfortunately the checks and balances that you would normally have with a very powerful technology are not very advanced. In fact, most countries, when they’re enacting legislation around AI, many times they’re taking away checks and balances as safeguards because they want to compete.

Des Dearlove:

Yeah. No, I think this is one of the great challenges, that technology moves much faster than the speed of legislation.

Mike Wade:

Yes.

Des Dearlove:

And particularly with AI, there is this sense that if government gets in the way, if regulations get in the way, we’re going to fall behind to China or whatever. So it’s a real conundrum because on the one hand, we’re racing as fast as we can go in order to be the ones out front, but at the same time, we don’t really know quite what we’re heading towards.

Mike Wade:

Yeah, that’s right. And I think the danger, as you point out, is that we’re advancing this technology faster than our ability to control it. And at some point we can’t put the horse back in the barn and so we should be, I think, thinking a little bit more than we are about the risks and our ability to control those risks. But as you say, it’s a geopolitical race and we’ve got the two big powers when it comes to AI, the US and China, taking different approaches. And then we have Europe and the approach of Europe, which often is the case, let’s put the protections in place first and then let’s see what we can do with the technology. The US saying, “Let’s just go. We’ll worry about the future later. We don’t want to constrain innovation.” And then China taking a somewhat kind of in the middle approach, let it happen and then when they’re ready, come in with legislation. So the results are that the US and China are far, far ahead today when it comes to AI.

Des Dearlove:

That’s the technology side. Let’s talk a bit about the human side. I mean, I think for some reason, AI has this different sort of persona in terms of I think we understood that technology is a tool and enabler, but something about AI has spooked us or some parts of humanity and there’s a fear that it’s going to push out what makes us human. How can leaders… Well, presumably it’s desirable to keep the humanity, let’s debate that first. But then secondly, how does one do that in a way that you’ve, one, that the company, the organisation remains competitive?

Mike Wade:

I think a lot of it comes down to communication and messaging. As we spoke earlier, digital transformation is very much about change management and AI driven transformation is the same, but with any change management, there’s resistance, always resistance. Many people don’t like to change. But the thing when it comes to AI, there’s-

Des Dearlove:

Sorry, many people are frightened of change.

Mike Wade:

They’re frightened of it, of course.

Des Dearlove:

I mean, that’s what… Because what does it mean for me? Often it’s the case of how do you create some reassurance or a sense that there is a future?

Mike Wade:

They’re worried about the future that they can’t control, they don’t understand. But when it comes to AI, there’s this extra element, this kind of existential, it’s not just resistance because they don’t like change. It’s resistance because they may not have a job. And this is kind of an extra layer of resistance that organisations have to overcome and how do they overcome it? They overcome it in the same way, we’ve always done good change management with solid communications, reinforcing of messages, making sure that people understand that it’s not there to replace them, it’s there to help them do better work than they’re doing today. So it’s not just about replacement, it’s about augmentation.

I think that’s a really, really strong message that organisations need to push and help train them because most people, they’re just really scratching the surface when it comes to AI. For most people today, it’s still a chatbot. That’s it. It’s just a chat window. And when they realise there’s so much more you can do than just type in a chat window, possibilities open up and they realise that they can do their jobs not only faster, which is fine, but often that’s taken by the company anyway, but doing it better, doing their jobs better. And then that helps to overcome some of that resistance.

Des Dearlove:

Where do you see the pockets of best practice? I mean, I know this is a very fast moving picture, but I mean, obviously the technology companies are, that’s what they do, but some of the more, perhaps the more traditional companies, do you see examples where they’re managing… Because there’s a tension. Do you deploy AI everywhere and invest squillions of dollars? Which is one way of doing it. Do you wait and see a little bit, but you don’t want to wait too long because the competitor will have leapt over you and be disappearing into the distance. So there’s a tension between, do we go now? Do we wait? But who’s doing this well? Who’s managing this process well?

Mike Wade:

So I think there are organisations that are doing this well. I think the best organisations are taking upskilling AI literacy and training seriously. So they’re equipping their people with the skills and knowledge to leverage the technology. I think the best organisations are avoiding exactly what you said, the temptation to just do a bunch of things, just run ahead all kinds of projects that individually may make sense, but as a portfolio make no sense at all and focusing instead on the value that can be created and being rigorous about that and then working backwards. Because whatever the value is created, maybe they want to increase the top line, maybe it’s about efficiency, maybe it’s about market share, maybe it’s about patient care, whatever it is, but being rigorous about that and then working back. And what they may find is that the best way to get there doesn’t have to be generative AI or agentic AI.

It could be something else. And what we see a lot of companies today, it’s like they have the gen AI or the agentic AI hammer in their hands and they’re looking around. Everything looks like a nail, so they’re using it in all kinds of places where it’s not optimal. They could probably do it cheaper, more accurately, with less impact on the environment using some other type of tool. It’s very important, AI, they have to use it, but moving, shifting the focus toward the value that can be created rather than just focusing on the technology itself. I mean, I think that’s what the forward-thinking organisations are doing very well today.

Des Dearlove:

Do you anticipate fallout? I mean, it feels like we’re in the AI bubble where people are… Organisations are throwing money at it. Do you see any signs that some air’s going to come out of the bubble at least?

Mike Wade:

I mean, it has to, right? It has to. There’s so much buzz around it. I mean, today as we’re recording this, we’re seeing a massive selloff in AI stocks now. Will that continue? It’s hard to say. I think when we look at the financial markets, the stock markets, there’s a disconnect between the underlying performance of these companies and the valuation of the companies. So I think there’s going to be probably a retreat there, but where there’s a lot of smoke, there is fire. So these technologies do have the ability to create value in so many different ways. So organisations that can look past the buzz and the hype and really zero in on the concrete use cases where they can show improvement, I think those are the ones, the rigorous prioritisation, and prioritisation is less about what you do, more about what you don’t do. They rigorously prioritise, I think they have the best chance of capturing value from AI today.

Des Dearlove:

And we have to be aware of our context. We’re sitting here in Switzerland, right in the centre of Europe. And as you indicated earlier, the US is charging ahead, the Chinese, charging ahead. Is there any hope for Europe catching up?

Mike Wade:

Europe has all the raw materials it needs. Europe has a smart people, has access to capital, has great universities, it has great companies, it has everything it needs. It’s just got to get its act together. And I think what we need is a really coordinated effort within Europe, which as we know, Des, has been difficult historically to achieve, but there are some examples, right? CERN in Geneva is a good example where for that type of fundamental research, Europe got together and created a place where best scientists come and all of this stuff, like a Silicon Valley.

After the first Gulf War when the US decided to restrict the GPS, Global Positioning System, I think a lot of European countries and companies thought, “Wow, we’re building our business on this technology.” And so Europe developed its own version of the GPS system. So I think there’s hope, but I mean, we need to move much more quickly, much more aggressively than we have done and make some tough choices, like there’s going to have to be a physical location somewhere in Europe that aggregates all this talent. Where’s it going to be? This is one of the-

Des Dearlove:

Our own cluster, our own Silicon Valley or equivalent.

Mike Wade:

We need that. Yes.

Des Dearlove:

Yeah. Okay. Final question. We – the Thinkers50 London Summit & Leaders50 Awards Gala – we’re launching the new Leaders50 list this year. So, 2nd of November, we’re going to have 700 of the leading management thinkers of the world mingling with the most inspiring business leaders, particularly to the leaders in the room, what would be your advice? What do they need to be aware of and what should they do when they go back to the office on the 3rd of November? What would your advice be to them? What do they need to do?

Mike Wade:

I think they need to be laser-focused on where the value is going to be created in the future and how they can capture that value with AI but not just with AI. I think they also, there’s a lot of talk about agility, rightly so. I mean, organisations need to be fast, but I think today there’s almost too much focus on speed and not enough focus on robustness because I don’t think the changes that happened, the disruptions, the threats we talked about earlier are going to slow down. There’s going to be more and more of those. And agile companies that can move quickly won’t be able to avoid them all. They’re going to take damage. They’re going to take hits so I think organisations that blend and match agility and speed with strength and robustness, I think are the ones that are going to survive and thrive in an increasingly unpredictable environment going forward.

Des Dearlove:

Mike Wade, thank you very much.

Mike Wade:

It’s my pleasure.

IMD professor Yoshi Fujikawa argues that leaders need to move beyond traditional value chains and embrace a world of interconnected ecosystems, where customers, employees, partners, and AI all play an active role in value creation. 

He explains why service is no longer simply an industry sector but a new logic for business, how AI is transforming marketing from human-to-human to human-to-AI interactions, and why competitive advantage increasingly comes from orchestrating value constellations, rather than simply selling products. 

Yoshi Fujikawa is Professor of Strategy and Marketing at IMD.

Watch the full interview or read the transcript below. 

This interview took place at IMD’s signature Orchestrating Winning Performance program in Lausanne, 2026.

Transcript

Des Dearlove:

Hello, I’m Des Dearlove, co-founder of Thinkers50. My guest today is Yoshi Fujikawa, professor of strategy and marketing at IMD. His work sits at the intersection of strategy, marketing, service management, digital transformation, and executive learning. Yoshi reframes service not as a sector, but as a fundamental logic of value creation. One that increasingly depends on customers, partners, employees, platforms, and ecosystems, co-creating value together. Yoshi, welcome.

Yoshi Fujikawa:

Thank you, Des.

Des Dearlove:

I said just now, you argue that service is not simply an industry sector, but a fundamental logic of value creation, and we’re going to talk a lot about co-creation, but what are the implications for today’s leaders?

Yoshi Fujikawa:

Yes. I’m calling it a shift in lenses and there’s two ways of looking at the services. The first one, as you mentioned, the service as a sector. That’s a very traditional way of looking at the service. Look at the world saying that everything is basically physical at the core. The physical products and manufacturing, that’s always the core. But look outside and there’s also something non-goods. We have been calling it the services, like the first sector, physical goods, second sector, and whatever is leftover is the services.

That’s the lens one approach to look at the service. But if you look around the world now that everything is becoming a service, everything has a service. The manufacturing companies are becoming more and more service-based. Service companies also start handling the physical products and hardware such as Amazon Echo, Google Nest, and all of these. The tech companies are also holding the physical products for the purpose of getting connected and stay connected with the customers, see what the customers are doing and that creates the opportunities for the value co-creation.

So the implications for the leaders is that these shifting lenses opens up the opportunities for them to find, detect, and then also seize those opportunities for the value co-creation.

Des Dearlove:

So CK Prahalad, who is a great inspiration for Thinkers50 – yeah, absolutely – so whenever I hear the word co-creation, I always think of CK. But I mean, he was talking about it 10, 15 years ago, but some of the technologies we have now really make it much more feasible and much more possible. So how would you describe that sort of journey we’ve been on from what CK was describing, which at the time seemed off in the distance?

Yoshi Fujikawa:

Yes. So the co-creation is actually nothing new as you have just mentioned, but this technology and the current business environment actually opens up opportunities for all the businesses and all the companies and all the industries. That’s basically the phenomena called the “everything as a service”. And so now that you could actually reach out to the customers and technology help is always there so that you can stay connected beyond which that you have already provided the product to the customers, or you already sold the service, which the customer pays for.

That’s a traditional ending point of the value chain, traditional value chains, beyond which there has been open space, the white space right there. But now the technology actually will help you to go beyond the traditional boundary of the value chain, stay engaged with your customers and see what the customers are doing. And then also you can bring in other partners and stakeholders outside of organisations and so you can take the ecosystem approach to it. And now not just the digital technologies, now the AI.

And so now there’s what we have been calling service encounters, where providers and receivers meet and interact, that’s we call it service encounters, used to be human to human and now human with AI. And now the companies are employing AI to do their service operations, but now we as the consumers and our customers, whether that’s B2C or B2B, now we are sending our agents to plan my next trip to Singapore, for example.

And it is the AI who is actually visiting Booking.com. It is the AI visiting the airline website. It is my agentic AI that is actually comparing and then finding what’s possible and then comparing and then also deciding and suggesting what to do. And so far so good, but this also kind of challenges leaders how to think about value co-creation going forward.

Des Dearlove:

So you’re a professor of strategy and marketing. So let’s take the marketing part. And I know those two things are increasingly intertwined. It’s harder to separate, but taking the marketing, what you’re describing is a world where we’re no longer necessarily marketing to human beings. You’re sort of marketing to AI.

Yoshi Fujikawa:

Exactly.

Des Dearlove:

Yeah.

Yoshi Fujikawa:

You need to help AI to find you. You need to help the AI to evaluate you. You need to find AI to trust you. And so the way that you do the marketing is no longer one way from company to humans. It’s company and human and then AI. So you need to design that – who you want to bring in as the actors in your ecosystem and then how you’re going to orchestrate it.

And so the marketing is no longer done by the company; it has been co-creation, but now the co-creation is no longer between humans and humans, but humans and AI and then AI to AI. And so you need to design and then implement it. So that’s the opportunity right now that is emerging.

Des Dearlove:

So can you give me an example of how that sort of changes the marketing approach? I mean, what do we need to do differently to… I mean, it’s interesting language: win the trust and be discovered by AI rather than trying to appeal to human beings and trying to get their attention.

Yoshi Fujikawa:

Yeah. There’s a classic notion of customer centricity that is always there, where we have been using this concept of Jobs To Be Done, JTBD, Clay Christensen, there’s Thinkers50 terminology, I believe. And so whatever you do is basically at the end of the day is for the purpose of the jobs to be done as a north star.

And now going back to our discussion earlier, now technologies help and then everything was going on with AI. Now that you could actually get the jobs to be done, actually get done. So Jobs To Be Done as a concept from the marketing perspective is up to the point of purchase, or up to the point of sales, which is the traditional ending point of the value chain.

Des Dearlove:

Well, to some extent, that was job done!

Yoshi Fujikawa:

It’s kind of promised, right?

Des Dearlove:

Yeah, exactly.

Yoshi Fujikawa:

But when exactly the job gets done is actually through the usage stage… so that is beyond the traditional value chain. We call it “value constellation space”. And here that you need to make sure that a customer is actually using your product and that customers are using your service and that usage opens up the opportunities for you to co-create because usage generates the data with the technology and then that data help you to run the analytics and then get back to your customers and other stakeholders who may be helping the customer’s job actually get done in the value in use… so we call it a “value in exchange” and a “value in use”. Value in exchange is when the customer pays for it. That’s when the value is to be maximised through the market exchange. Again, that brings us back to the traditional ending point of the value chain. Value in use is always to be co-created through the usage stage of their customers. And so this is the opportunity now – that the value co-creation is actually taking a new shape.

Des Dearlove:

So if I’m understanding you correctly, I mean the job to be done as far as the customer is concerned could evolve over time. And if you’re tracking their changing needs at the point of purchase, they may have had one job to be done, but during the lifetime experience of using the service or the product, it could be that that changes. And if you’re tracking them and if you’ve got the analytics, you will see and be able to respond to the new job to be done.

Yoshi Fujikawa:

Right. So now the jobs to be done is kind of like dynamically evolving on both sides. I mean, the jobs to be done for the customers, the jobs to be done actually get done in the process of your operation. And so now the job to be done is not just a kind of static promise at the point of sales and a point of purchase. Now the job to be done is a kind of dynamically changing picture, kind of like a moving target of it.

But that moving target is co-created by customers and your actors – whoever that you’re bringing into your ecosystem – and then yourself. So the challenge for the leaders is now you need to orchestrate this much bigger picture of no longer just the value chain of your single entity and now you need to orchestrate the value constellation, just like, look at the night sky: connecting styles and stars, making meaning out of it. That’s a constellation. And so that’s what you are now actually tasked to do, but that’s actually a great task, an exciting task to do as a leader. And so that’s how I see it from the service innovation perspective.

Des Dearlove:

And CK would have been so excited by the possibilities now. I could just see his face lighting up. Okay. We talked a little bit about the marketing side. What does this mean for strategy? What does it mean for the nature of competitive advantage if you like? If you’re the CEO or you’re trying to direct the future of an organiaation, how does this change strategy?

Yoshi Fujikawa:

Right. So the strategy is always for long-term sustainable superior profit. And so strategy is not just for pursuing profit for the short term, it’s actually for making it sustainable, not for the environmental sense, but they are sustainable in terms of the competitive sense and then also superior, which is above the average of the industry structure. And so how this approach would help the strategy to actually make a difference is now the strategy you need to incorporate and there’s the value co-creation approach, and it’s no longer that you’re competing head-to-head through for the value in exchange, but the value co-creation in the form of the value in use is now possible.

And that’s where the differentiation is coming in. And because the AI is going to become pretty much commoditized as a tool, but the AI as an actor in an ecosystem helps you to build the trust and helps you to build the relationship with your customers and other stakeholders. That’s what’s going to stay as the source of your competitive advantage.

And so combining the service management approach in terms of the service innovation integrated into service, the strategy, will help you to make your strategy even more sustaining and your long-term profit is even more sustaining and then also superior to the industry average, which is competitors.

Des Dearlove:

I mean, obviously for a number of years now, we’ve witnessed how organisations like Apple encourage us to join the Apple ecosystem.

Yoshi Fujikawa:

Yes.

Des Dearlove:

You buy one Apple product and then there’s kind of an inbuilt logic, “Well, I might as well go all Apple, because I know that they’re going to talk to each other nicely and play nicely together.” So we’ve had this kind of ecosystem strategy for a while. We used to see it a little bit with… well, a lot with the software companies. How different is this world that you’re describing, the sort of AI enhanced and enabled work? How will that change what we as customers experience in terms of which products and ecosystems we choose to be part of?

Yoshi Fujikawa:

Yeah. So it’s an interesting world, right? Now the companies are competing… ecosystems competing with other ecosystems, and there’s some of the industries where customers are easily switching across each other, whether you’re using Uber today, or like Lyft, or like other platforms where the companies are building an ecosystem. And then none of us wants to be just confined into just one ecosystem. And because we always fear how much we are actually getting exploited, and that’s basically a trust issue of the ecosystem.

And so now the ecosystem used to be only kind of possible for like a tech firm, who operates in terms of the platform as a means to build the ecosystem. But now though, you don’t have to be the platformer to become the ecosystem builder. So for example, like Phillips, Sonicare, the toothbrush which I’ve just used this morning, and now the Sonicare toothbrush comes with the app and this app kind of evaluates in real time now, how I’m actually brushing my teeth, and most of the act of brushing the teeth is unconscious and that makes it kind of like a conscious part of what I do and then every day… and then also that stores the data and then that runs analytics and then that gives a lot of different kinds of advice.

And then this data can be shared with the service providers such as the dentists and then other product providers so that Phillips can build the ecosystem around the toothbrush. And so who thought about, like, 20, 30 years ago, like a toothbrush can become an ecosystem? 

And now Nestlé just down the street, they are doing the pet food business called Purina, right? Initially Purina. Now dog food and pet food is just a product selling marketing business and B2C, and probably sometimes in a B2B, but now this dog food and a pet food can become the ecosystem as well because they’re going back to this job to be done. The reason that you are selecting Purina versus the other competitor’s product is that Purina comes with this app that basically helped me to become a better pet owner, become a better father and then a mother of the pet that I have. And because this helped me to manage what food that I’m providing to my pets, and then also that comes with a lot of different kinds of suggestions, and then the sharing knowledge and then how to become the… so there’s other pet owners also sharing their data as well.

And the same goes to Nike Run Club and many of these things are now open up for… anyone out there can become the ecosystem builder, which is very different from 20, 30 years ago where the world was only limited to those people who could actually build their tech platform such as Google and Apple and such.

Des Dearlove:

It puts me in mind of a few years ago when we were in China and we went to the factory of Haier. You know the white goods manufacturing firm?

Yoshi Fujikawa:

Yes.

Des Dearlove:

And they were very clear that they were building ecosystems around the refrigerator in the kitchen.

Yoshi Fujikawa:

Yes.

Des Dearlove:

So the refrigerator was going to be able to tell you what meals it was capable of-

Yoshi Fujikawa:

What you have in there, what’s possible.

Des Dearlove:

… and we order whatever you need there and all that. So it sounds like some of that is coming to pass. It seemed very futuristic at the time, but obviously things catch up very quickly.

Yoshi Fujikawa:

Yeah.

Des Dearlove:

There is a sense of an arms race with AI very much, the tech companies in particular, but all companies at the moment seem to be absolutely throwing money. It almost feels like, as I say, an arms race, who’s going to be first to get to the moon? And do you think that this is a window that’s going to close at some point? Because what tends to happen with ecosystems, as you say, we all like to think that we’re free to move between different ecosystems, but it feels like there’s a lot of competing AI tools at the moment, but it will consolidate.

We saw it with the motorcar back in the beginning of the 20th century. There were lots and lots of car companies and then gradually it winnows down to a few giants. Is this window, particularly with AI, is that going to stay permanently open or will we see this… are we in a kind of a special moment where it’s winner takes all?

Yoshi Fujikawa:

For the user side of the technology?

Des Dearlove:

From a strategy point of view, I mean, there does seem to be a great deal of almost panic in the marketplace in terms of the investment in AI as if a more measured approach isn’t going to serve. So we’ve got to do it now and we’ve got to be first.

Yoshi Fujikawa:

So I think in terms of the companies who are employing AI for their strategy – not the AI companies propagating the AI solutions to their clients –  in terms of the companies who are employing AI for their strategy, I think the more important thing is not just to rush into the AI investment for the sake of the AI investment. You need to design, going back to what we discussed earlier, you need to decide what role that AI plays as an actor in the ecosystem, what role that they’re actually going to play, and then how you’re going to do the resource integration as we call it.

The AI may then have access to the resource so that they can find what’s available out there, whereas the humans can also have access to their resources and then our time, our knowledge, our skills… and then also other stakeholders that you invite in an ecosystem have their access to their resources.

And so it’s basically value consolidation, it’s resource integration across different actors, including you. And so when it comes to your question, going back to your question, that what role that you want to assign the AI to play in this ecosystem is far more important to comparing rushing into the AI investment for the sake of just expanding an investment and for that. So I think what we need to really worry about is more about the ecosystem design too, from my perspective, rather than just how to invest AI for the future.

Des Dearlove:

This is fascinating stuff and I could happily sit here and talk to you for hours, but we are running out of time. And I do want to ask you a final question. The 2026 Thinkers50 London Summit is taking place, I mean, as we’ve sort of hinted at, obviously at a moment of technological disruption and institutional uncertainty, what kind of leadership mindset is needed now and what would be your advice to…

We’re bringing together some of the world’s leading management thinkers, but also our new Leaders50 list is coming and they’re going to convene together for the first time at the 2026 Thinkers50 event. So there’ll be leaders in the room and there’ll be thinkers in the room. What would your advice be to the leaders and how should they approach this brave new world? What sort of mindset is needed to really take advantage of these opportunities?

Yoshi Fujikawa:

I will start with a very, very classic mindset, which is customer centricity again. So the customers’ jobs to be done is always north star, that is not what you want to be changing. And that’s from my perspective, that’s how the world actually operates. And so the customer job to be done, although it could be dynamically changing as we discussed earlier, is always a north star.

And now how to get there, by redesigning orchestration of the value co-creation in a multiple actor ecosystem. But in order to do that, you need to make sure by deciding or reconfiguring three, four things. And one is who are the actors that you want to invite and what role that you want to assign them to play and which resources that you want them to have then access to. And then how you want to change the rules of resource integration.

And so that’s the kind of like, the mindset that we should all have in order for us to kind of detect and grasp those opportunities that are actually emerging right now. And so that is the thinking underneath what’s actually happening on the surface. And so that’s how I see it.

Des Dearlove:

And finally, if you were talking to a leader today, should they be optimistic? Is this a time of opportunity or a time of threat?

Yoshi Fujikawa:

I’m super optimistic about it and this is actually the most exciting time in terms of value co-creation. And so the world is actually helping every single one of us to see the world in this way, which used to be limited to only those organisations and individuals who have that access to… and that, but now it is open up for everybody to see that.

Des Dearlove:

And you use AI all the time?

Yoshi Fujikawa:

I do.

Des Dearlove:

What do you use it for?

Yoshi Fujikawa:

I do. I just brushed my teeth right there with the Sonicare, just right before I rushed into this interview!

Des Dearlove:

Yoshi, thank you very much.

Yoshi Fujikawa:

Thank you.

By Dan Toma

Artificial intelligence has fundamentally changed the economics of product development. According to Cursor’s latest Developer Habits Report, developers are coding significantly faster than they were a year ago, while Nvidia recently reported that AI has enabled its engineering teams to produce roughly three times more code than before. Whether the exact productivity gain is twofold or threefold is almost beside the point. Building software has become dramatically cheaper and faster than it was only a few years ago.

Naturally, this raises an interesting question: what happens to the Innovation Accounting indicator, Cost of Failure, when the cost of building collapses?

Recently, I’ve heard arguments that Cost of Failure is becoming a less relevant Innovation Accounting KPI in the age of AI. If the cost of building continues to fall, then surely the cost of failing falls with it.

The opposite may be true.

AI hasn’t made Cost of Failure less relevant. It has simply changed where the cost comes from – and, as a result, what the metric tells us.

Cost of Failure was never just about building

The idea that Cost of Failure only measures the cost of building something is the root of the misconception.

Cost of Failure has never been solely about development costs. It is a compound indicator made up of three distinct costs that every organization incurs while bringing new products to market. The first is the Cost to Build the product or service. The second is the Cost to Learn whether customers actually value it by generating sufficient evidence. The third is the Cost to Decide what to do with that evidence – whether to stop, pivot, persevere or scale.

Cost of Failure = Cost to Build + Cost to Learn + Cost to Decide

AI hasn’t changed the equation. It has changed which part of the equation dominates.

Historically, the Cost to Build was by far the largest component of the equation. Software development was expensive, engineering teams were large and development cycles lasted months or even years. It was therefore natural to interpret a high Cost of Failure primarily as an engineering problem.

AI has changed what Cost of Failure measures

There is little debate that AI has dramatically reduced the Cost to Build. Code generation, automated testing, AI-assisted design and low-code platforms have compressed development timelines and reduced the effort required to create digital products. As organizations successfully adopt these capabilities, the Cost of Failure should logically decline.

The mistake is assuming that every component of the equation declines at the same rate.

Learning still requires customers. AI can summarize interviews, synthesize research and generate hypotheses, but it cannot tell you whether customers will actually buy your product. That still requires real-world evidence. While the Cost to Learn may be lower today because prototypes can be built and tested much faster, generating meaningful evidence and allowing experiments to run their course still takes time.

The same is true – perhaps even more so – for the Cost to Decide.

The Cost to Decide isn’t the cost of the meeting. It’s the cost of waiting for the meeting.

Once a team has generated sufficient evidence to justify a decision, every additional day before that decision is made continues to consume resources. Teams remain allocated, budgets remain committed and opportunities to redeploy people to higher-potential initiatives are delayed. In practice, the Cost to Decide is the fully loaded cost of keeping a team and its initiative alive while the organization decides what to do next.

A simple example illustrates the point. If a team costs €8,000 per day to run and the decision to stop or continue the initiative is delayed by 45 days while waiting for the next portfolio review, the organization incurs an additional €360,000 in Cost to Decide. That money isn’t spent building. It isn’t spent learning. It is simply the cost of waiting.

With or without AI, most organizations still review innovation portfolios quarterly, allocate funding annually and require multiple approval layers before teams receive permission to continue or stop an initiative. AI has compressed execution, but many organizations still govern innovation at the same pace they did before AI.

Ironically, AI may amplify this problem. As the Cost to Build falls, organizations can pursue far more ideas in parallel. Instead of reviewing ten initiatives, leadership may suddenly find itself reviewing fifty. The engineering bottleneck shrinks. The decision-making bottleneck grows. AI increases an organization’s capacity to build faster than its capacity to make decisions.

As the Cost to Build shrinks, the Cost to Learn and the Cost to Decide account for an increasingly larger share of the total Cost of Failure.

A high Cost of Failure no longer tells us primarily that engineering is expensive. Increasingly, it tells us that organizations are taking too long to learn – or taking too long to act on what they have already learned.

Twenty years ago, a high Cost of Failure often reflected the economics of software development. Today, it increasingly reflects the economics of organizational decision-making.

That is precisely why AI hasn’t made Cost of Failure less important. It has made it a far better diagnostic tool.

The new bottleneck is organizational

Most executives already know their organizations are slower than they would like. They know investment decisions take too long. They know projects remain alive long after the evidence suggests they should be stopped.

What they rarely know is what that bureaucracy actually costs.

Organizations that continue to govern innovation through quarterly portfolio reviews, annual funding cycles and rigid approval processes will simply build unsuccessful products faster than ever before. AI won’t eliminate waste – it will amplify the consequences of slow governance.

The difference has nothing to do with technology. It has everything to do with governance.

The greatest opportunity AI creates isn’t faster coding. It’s faster decision-making.

Otherwise, organizations simply replace one bottleneck with another. Instead of waiting for engineering teams to build products, they wait for leadership teams to make decisions about products that have already been built and validated.

The companies that benefit most from AI won’t necessarily be those with the best models or the fastest developers. They’ll be the ones that make better decisions, faster.

AI didn’t kill Cost of Failure. It made it a better diagnostic tool, putting a dollar value on slow and inefficient governance.

Dan Toma is CEO and Partner at OUTCOME and a leading voice on how established companies build sustainable top-line growth. His work focuses on the systems that enable growth – from governance and funding to portfolio management and execution – helping organizations move beyond “innovation theatre” to measurable business results. He is the award-winning co-author of The Corporate Startup, Innovation Accounting, and Open Innovation Works, and has advised executive teams across the financial services, energy, pharmaceutical, healthcare, and engineering sectors.

Dan was named to the Thinkers50 Radar in 2020.