The McLean-Walton Divergence

The shipping container was supposed to make Malcolm McLean rich. Instead, it made Sam Walton a fortune. This paradox shows us who will profit from AI, and it is not always the people who build the technology.

In 1956 McLean standardised the shipping container. Transport costs fell by 90 per cent and the modern supply chain was born. Yet McLean’s Sea-Land never saw matching profits. It needed constant reinvestment, struggled with margins, was sold by its founder, and in time its international arm was absorbed by Maersk. The winners were retailers like Walmart, who could source goods globally, and East Asian manufacturers, who could now reach those markets.

The inventors built the technology. The money went to the businesses that organised around it. AI is repeating the same story.

When foundation models settle into standards, the advantage shifts. It no longer comes from physical assets or proprietary datasets but from supporting assets that strengthen with use. These include feedback loops that improve with activity, communities built on trust, and ways of working that embed judgment over decades.

How Businesses Move from Static Data to Continuous Learning

Many assume proprietary data creates a defensive advantage. That view is fading as intelligence becomes a commodity. The real advantage lies in how operations generate fresh insights from each user interaction, producing a feedback loop that strengthens with every use.

Take GitHub’s Copilot. Its original training data was public code, available to anyone. Once deployed, it began learning from millions of developers accepting, rejecting, and editing its suggestions in real time. Each of those choices teaches the system about code, but also about how people work, how organisations structure projects, and what best practice looks like in practice. That feedback loop cannot be bought.

A capable model can be built by anyone. What cannot be replicated is millions of people training it every day through lived use.

The lesson is clear. Stop thinking of data as something to stockpile. Start building systems where every interaction generates proprietary intelligence that improves tomorrow’s service. Zoom refines its platform through millions of live calls. Tesla’s autonomous driving improves with every mile driven. Both companies redesigned their operations around this principle.

Why Established Relationships Outperform the Best Algorithms

As synthetic media blurs with reality, trust becomes scarce. A trusted brand or community is a supporting asset that AI-native startups cannot engineer, no matter how sophisticated their models. Consider financial services. A Silicon Valley firm could launch a technically stronger robo-advisor. Without trust, it would struggle to gain clients. A regional wealth manager with three generations of relationships could use a simpler tool and still achieve better results. Clients extend existing trust into new domains.

The same holds in media. Anyone can offer an AI summary for free. Readers still pay for The Economist because they trust its editorial judgment, even when filtered through machines. Stronger technology without trust loses to weaker technology in trusted hands. The relationship, not the algorithm, carries the weight.

How Organisations Can Turn Experience into Lasting Systems of Judgment

Many organisations say they value human judgment, but few preserve it in a way that lasts. The ones that do create structured systems that AI can extend but never replace.

Toyota’s “Five Whys” is an example. It looks simple, but it took decades to refine. An AI can ask questions, but it cannot copy the culture that allows honesty, the intuition that knows when to keep probing, or the recognition built from thousands of investigations. Processes shaped over decades resist imitation. What looks straightforward on paper is inseparable from the culture that sustains it.

Bridgewater Associates offers another case. Its practice of “radical transparency” is not a quirk, but a system for stress-testing investment ideas. AI can analyse market data, but it cannot reproduce the exact way Bridgewater surfaces dissent, confronts assumptions, and weighs conflicting views. Their system converts generic insights into differentiated decisions.

The Hard Choices About What to Build, What to Codify, and What to Let Go

The hard task is to identify where everyday operations generate insights that competitors cannot buy. This is not about digital skills. It is about spotting where behaviour produces patterns that, once captured, compound over time.

Trust raises another difficulty. The organisations with the deepest reserves of trust—banks, medical providers, long-standing media—often move slowest with new technology. Digital natives move fast but lack the trust to make adoption profitable. Each side is disadvantaged.

Wisdom presents the hardest challenge of all. In many organisations it lives with senior staff close to retirement. Codifying it risks flattening what makes it useful: context, culture, the lessons drawn through practice. But letting it walk out the door is just as damaging. Each departure thins the organisation while rivals may succeed in building systems that keep learning.

Executives who endure will be those who can sit with these tensions. They will not chase neat answers. They will accept that models are commodities, and that the real edge lies in what cannot be copied: feedback loops that strengthen with use, trust built over years, and ways of working that preserve hard-won judgment. Building these takes imagination, patience, and judgment of a different kind.

Closing Reflection

The shipping container made global trade frictionless, but the companies that hauled the boxes rarely made the real money. As AI makes intelligence frictionless, the critical question is not whether you own the model but whether you have built the supporting assets that others cannot copy. Can you develop feedback loops before rivals do? Can decades of trust withstand disruption? Can wisdom survive codification without losing its force?

These choices decide whether you become the Sea-Land or the Walmart of the algorithmic age. Unlike McLean, you can already see which fate lies ahead for those who perfect the technology while others perfect the business.