In 1956, Malcolm McLean changed global commerce forever. His standardized shipping container slashed transport costs by 90% and made modern supply chains possible. Yet McLean’s company, Sea-Land, never captured the wealth his innovation created. It demanded constant reinvestment, struggled with thin margins, changed hands multiple times, and eventually saw its international operations absorbed by Maersk.
The real winners? Walmart and other retailers who could suddenly source products from anywhere on earth. East Asian manufacturers who gained instant access to global markets. The companies that built around the innovation, not the ones who built the innovation.
This pattern is repeating with AI. And most organizations are on the wrong side of it.
The Commodity Trap
When foundation models become as standardized as shipping containers—and they will—competitive advantage doesn’t just shift. It transforms entirely. The static assets that defined industrial-era success (prime real estate, proprietary datasets, regulatory moats) lose their protective power.
What replaces them? Dynamic, living assets that strengthen through use:
- Feedback loops that compound with every interaction
- Trust networks built over years, not quarters
- Judgment systems refined through thousands of decisions
The companies that master these will capture AI’s value. Everyone else will rent commoditized intelligence at razor-thin margins.
Why Your Data Isn’t the Moat You Think It Is
The conventional wisdom says proprietary data creates defensible AI advantages. This fundamentally misunderstands how value accrues when intelligence becomes cheap.
Your historical dataset matters far less than your mechanism for generating new insights from user behavior.
GitHub Copilot illustrates this perfectly. Its initial training data was public—anyone could access the same code repositories. But once deployed, Copilot began learning from millions of developers accepting, rejecting, and modifying its suggestions in real time. Each interaction taught the system about human coding preferences, organizational patterns, and emerging best practices.
A competitor can build an equally capable model. What they cannot buy is ten million developers actively training the system through daily use.
This is the new moat: systems that learn continuously from operations, not static data stores.
Zoom refines its platform based on millions of live calls. Tesla’s autonomous driving improves with every mile driven across its fleet. Both have rebuilt their entire operations around feedback loops that generate proprietary intelligence competitors cannot replicate.
The strategic question isn’t “What data do we own?” It’s “Where in our operations can customer interactions generate insights that compound over time?”
Trust: The Infrastructure AI Can’t Engineer
When synthetic media becomes indistinguishable from reality—and we’re nearly there—trust becomes the scarcest resource in business.
A deeply engaged community or established brand reputation is an asset AI-native startups cannot engineer, regardless of model sophistication. This isn’t about marketing. It’s about deployment infrastructure.
Consider financial advisory. A Silicon Valley startup might build a superior robo-advisor, but without trust, customer acquisition becomes prohibitively expensive. Meanwhile, a regional wealth manager with three generations of client relationships can deploy inferior AI technology yet earn superior returns.
Why? Their clients aren’t evaluating algorithms. They’re extending existing trust relationships into new technological domains.
This explains why The Economist can charge for AI-summarized content while generic AI summaries remain free. The technology isn’t better—the trust is. Readers trust The Economist’s editorial judgment, even when mediated through machines. Trust becomes the distribution channel that makes commodity AI profitable.
Stronger technology without trust loses to weaker technology in trusted hands.
The paradox: organizations with the strongest trust (century-old banks, established medical practices, legacy media) are often slowest to adopt new technology. Meanwhile, digital natives who can rapidly deploy AI lack generational trust that makes deployment profitable.
Which disadvantage is easier to overcome—technological debt or trust deficit? The answer determines who wins.
Judgment: The Asset Hiding in Plain Sight
Every organization claims to value human judgment. Few have codified it into defensible systems.
Toyota’s “Five Whys” methodology exemplifies this. It’s not just a problem-solving technique—it’s codified wisdom that took decades to perfect. An AI could simulate the questions, but it cannot replicate:
- The organizational culture that makes brutal honesty possible
- The trained intuition that knows when to dig deeper
- The accumulated pattern recognition from thousands of investigations
The process itself becomes an asset that evolves through use and resists reverse-engineering.
Similarly, Bridgewater Associates’ “radical transparency” isn’t cultural theater—it’s a structured system for stress-testing investment ideas. AI can analyze market data, but it cannot replicate the specific chemistry of their idea meritocracy: how they surface dissent, challenge assumptions, and synthesize conflicting viewpoints.
What looks simple on paper is inseparable from the culture and judgment that sustain it.
The hardest challenge? This judgment typically sits with senior employees nearing retirement. Codifying it into manuals risks flattening it—context gets lost, culture gets lost, the slow lessons from trial and error disappear. Yet if nothing is done, the knowledge walks out the door with each retirement.
Organizations must find ways to preserve judgment without hollowing it out. Those that succeed turn experience into systems that keep learning and improving. Those that fail grow thinner while rivals strengthen.
The Strategic Choices That Matter
Three questions separate future winners from future casualties:
1. Where can you build feedback loops before competitors do?
Inventory your operations. Where does human behavior reveal patterns that, once captured, create compounding returns? This requires creative thinking, not technological sophistication. The organizations that fail to find these opportunities will discover their expensively licensed AI produces identical insights to competitors’.
2. Can trust built over decades survive technological disruption?
In a world of synthetic content, how do you prove your AI-enhanced services deserve human confidence? The answer isn’t in the technology—it’s in the relationships you’ve cultivated and the reputation you’ve earned. These cannot be purchased or engineered on startup timelines.
3. Which judgment can you codify without damage?
Not all wisdom survives translation into process. Some must remain embedded in practice, evolving through use. The skill is knowing which is which—and having the patience to let trust and judgment develop when quarterly pressures demand immediate results.
The Real Question
The shipping container made global trade frictionless. The companies that moved the boxes rarely made real money.
As AI makes intelligence frictionless, history is rhyming. The question isn’t whether you own the model—it’s whether you’ve built the assets around it that cannot be copied.
Can you develop feedback loops that strengthen with every customer interaction? Can trust built over decades translate into AI-era advantage? Can hard-won judgment be preserved and amplified without being flattened?
These choices will determine whether you become the Sea-Land or the Walmart of the algorithmic age.
Unlike McLean, you can already see which fate awaits those who perfect the technology while others perfect the business around it.
The divergence is happening now. Which side are you on?