The Reconfiguration Advantage

Scott Brinker captured an important tension more than a decade ago in what he called Martec's Law.

The exact mathematics matter less than the intuition. Technology can move very quickly; organisations cannot. People need to learn, processes need to change, systems need to connect, governance and incentives need to adapt, and customers need to behave differently.

Speed has always mattered. What may be changing is the penalty for organisational lag.

Brinker's conclusion was never that companies should absorb every new technology. That would be impossible. Strategy is partly the act of deciding which changes matter enough to absorb. If the technology curve steepens while organisational change remains difficult, the gap between technological possibility and operating reality can become a competitive fault line.

Further reading: Scott Brinker — Martec's Law

Management theory has been wrestling with versions of this problem for decades. The ambidextrous organisation asks companies to do two apparently contradictory things at once:

Charles O'Reilly and Michael Tushman framed this as the challenge of competing simultaneously in mature businesses and emerging ones. That remains powerful, but I think the AI era makes another capability more important.

It is not enough to explore the future while exploiting the present. The organisation also has to be capable of absorbing what it discovers and reconfiguring itself around it fast enough for the discovery to matter.

Further reading: O'Reilly & Tushman — The Ambidextrous Organization

The Innovator's Dilemma gave us one of the most useful explanations for why successful companies can lose to newcomers. Incumbents could make rational decisions based on existing customers, economics and business models — and still create space for entrants whose initially less attractive offerings eventually became strong enough to reshape a market.

AI does not invalidate that problem. It may intensify it because AI is also changing the economics of being a newcomer. Agentic AI is compressing the time, capital and headcount required to create, test and scale new businesses; small teams can increasingly coordinate AI capabilities to perform work that once required much larger organisations.

For decades, incumbents competed with startups from a position of resource superiority: capital, people, distribution, data, infrastructure and brand. AI does not eliminate those advantages, but it can reduce the cost of competing with them.

The emerging question is not simply "Can the incumbent adopt the same technology?" Of course it can: buy the licence, run the pilot, hire consultants, create an AI team, invest in a startup or acquire one.

Can it redesign itself around what the technology now makes possible?

A newcomer can design workflows, roles and economics around AI from day one. The incumbent often starts by inserting AI into a system built around very different assumptions. The result can be a faster version of the old operating model rather than a fundamentally different one.

Perhaps this is becoming an Organisational Reconfiguration Dilemma: the incumbent can see the future, and can even buy access to the future, while still being unable to become it.

Further reading: Christensen, Raynor & McDonald — What Is Disruptive Innovation? · HBR — How Agentic AI Supercharges Startups and Threatens Incumbents

We usually treat incumbents and startups as opposing categories: one protects and optimises an existing model; the other attacks it. But an established company does not necessarily have to behave permanently like an incumbent.

Its operating model can deliberately create and connect multiple forms of entrepreneurial activity — internal ventures; spin-outs and venture studios; CVC and minority investments; open innovation and venture clienting; startup partnerships and joint ventures; acquisitions with retained autonomy.

Companies have been doing all of these things for years. The interesting question is not whether the mechanisms exist, but whether they form a coherent adaptive system: one through which the organisation can access entrepreneurial capability wherever it originates, learn from it, absorb what matters and reconfigure the core when necessary.

That changes the relationship between incumbents and newcomers. The startup is not simply a threat, nor merely a resource for an incumbent to use. The two can become parts of a wider innovation and adaptation ecosystem, with a large company containing newcomer logic inside itself while remaining connected to newcomers outside its boundaries.

Perhaps the most resilient incumbent is one capable of continuously creating, fostering, partnering with and absorbing newcomers — without forcing all of them immediately into the logic of the core.

Corporate venture capital provides a useful example. Recent research by Ezra Carlson, Mehdi Safavi and Nicolas Sauvage followed more than 100 CVC leaders and participants over seven years and found that successful CVCs do not eliminate the tensions between corporate and startup worlds; they develop repeatable ways of working across them.

The stronger CVC units become something closer to an organisational interface between the corporate core and the startup ecosystem. That matters because access to innovation is not the same as the ability to benefit from innovation.

A corporation can have a venture fund, a world-class scouting function and a portfolio of brilliant companies — and still fail to change what the organisation can actually do.

This is where absorptive capacity becomes important. Cohen and Levinthal described it decades ago as the ability of a firm to recognise the value of external knowledge, assimilate it and apply it commercially. AI may make that capability considerably more strategic — not because external innovation is new, but because the volume of potentially relevant innovation is growing while the time between "interesting" and "competitively important" may be shrinking.

Can we recognise what matters, access it, test it, learn from it, absorb it, change the organisation around it and scale it?

Further reading: Carlson, Safavi & Sauvage — What Successful Corporate Venture Capital Funds Do Differently · Cohen & Levinthal — Absorptive Capacity

This is where the different ideas begin to converge. Ambidexterity tells us to exploit and explore. Martec's Law reminds us that technology can move faster than organisations naturally change. Absorptive capacity tells us that access to external knowledge is not enough. Dynamic capabilities tell us that competitive survival depends partly on the ability to sense, seize and reconfigure. Corporate venturing and open innovation provide mechanisms for connecting the firm to entrepreneurial capability beyond the traditional core. AI increases the pressure on the whole system.

Put those things together and I think we get a different view of the competitive organisation: one capable of repeatedly moving through a reconfiguration loop while continuing to exploit today's business and explore tomorrow's.

That is what I mean by the Reconfiguration Advantage.

Further reading: David Teece — Dynamic Capabilities

A company cannot follow the technology curve, nor should it try. Most new technologies will not matter equally: some will disappear, some will matter to another industry but not yours, some will be useful but not differentiating, and some will require more organisational change than the value they create justifies.

Competitive advantage therefore cannot come from maximum adoption. It has to come from selective reconfiguration: choosing which external changes matter, understanding them early enough, accessing the right capability and changing the organisation sufficiently to capture the value.

Not pilots, demos or announcements. Operating capability. Something that changes how the company creates value.

Two companies may have access to exactly the same AI model. The competitive difference is not the model; it is how long each takes to turn that capability into a lower cost structure, a better customer experience, a redesigned workflow, a faster decision cycle, a different product or an entirely new business.

The Reconfiguration Advantage is not a universal organisational blueprint. A bank, an industrial manufacturer, a healthcare company and a software scale-up should not converge on the same operating model; their constraints, regulatory environments, cultures, starting points and ambitions are different.

So the destination is a capability, not a structure. Each organisation needs to understand its own starting point: what already works and needs protecting; where organisational friction is highest; which technologies or market changes are strategically relevant; which capabilities are missing; which innovation mechanisms already exist; and where the gaps sit between exploration and operating reality.

The framework is not the answer. It is a way of defining the capability the organisation may need to build; the path to get there remains specific to the company.

For a long time, transformation was something companies did periodically: an ERP programme, a restructuring, a digital transformation, a post-merger integration, a new operating model. There was a beginning, a programme and, theoretically at least, an end.

I am increasingly unsure that this fits the environment we are entering. If technology continues to move faster than organisations naturally absorb it, transformation cannot remain only an occasional intervention. But neither can organisations exist in permanent disruption; people cannot continually reorganise and processes cannot be reinvented every six months.

So the answer is probably not continuous transformation. It is continuous capability to reconfigure when necessary.

The distinction matters: the goal is not constant change, but being structurally capable of changing when the strategic value of doing so exceeds the cost.

This is also where my recent thinking about the multi-language human connects.

A reconfigurable organisation still needs deep specialists — perhaps more than ever — but it also needs people who can operate between domains: people who understand enough business, technology, operations, data and people to recognise the real problem, connect different forms of expertise and help move ideas into implementation. AI increases their leverage.

The organisation needs at a macro level something similar to what the multi-language human provides at a micro level: fewer walls between capabilities. People capable of crossing interfaces, inside organisations capable of crossing boundaries.

This brings me back to Anthropic. Damodaran's $1 trillion revenue scenario may prove wildly optimistic or conservative — we do not know. But for one AI company to generate anything close to that level of revenue, the economic landscape around it must change dramatically.

The discussion is often framed around jobs: how many will AI replace? That is an important question, but for business competitiveness there is another:

What happens when one company can produce with 30 people what another still requires 100 to produce?

That does not mean every company immediately eliminates 70 jobs. It means the economics of competition have already changed. The company with the lower cost structure can reduce prices, invest more, move faster, enter markets previously too small to serve, experiment more often or simply operate at margins the incumbent cannot match.

AI-driven productivity does not need to eliminate an entire profession to create enormous competitive pressure. And increasingly, that more productive competitor may be a newcomer designed from day one around a fundamentally different operating model.

For much of business history, competitive advantage has been described in terms of things a company possesses: scale, capital, technology, distribution, brand, IP, talent and data. Those things still matter.

But in an environment where technology becomes more accessible and entrepreneurship becomes cheaper, advantages based purely on possession may decay faster. The enduring advantage may increasingly be the organisation's capacity to renew the configuration of those assets: to sense what matters, access what it does not have, absorb what it learns, reorganise around it and scale the result before competitors do — without destroying the business that still works.

That is why I think the competitive organisation of the AI era may need to be more than innovative, efficient or ambidextrous. It may need to be reconfigurable.

Not because change is inherently good, but because the cost of being unable to change may be rising very quickly.

The future may not belong to the company that predicts it best.

It may belong to the one that can absorb it fastest — and turn it into operating reality without breaking what already works.