For the past two decades, the technology industry has been obsessed with building moats.

Network effects. Proprietary data. Switching costs. Regulation. Lock-in.

The idea was simple: build something competitors cannot copy. Grow fast. Monetize later. The customer was not the point. The customer was the resource.

It started with social media. Grab users now. Figure out the business model later. Treat attention as inventory to be harvested and sold to the highest bidder. The customers were never really customers at all. They were the product.

And it worked. For a while.

Then came the reckoning.

The Extractive Model Is Breaking Down

The numbers tell a story the industry does not want to hear.

Trust in technology companies has collapsed. According to the Edelman Trust Barometer, technology as a sector has seen consistent erosion in public trust over the past five years. Only 23% of consumers say they trust social media platforms. 61% question the authenticity of content they encounter online.

Engagement is fragmenting. The platforms that once commanded billions of hours of daily attention are losing ground to smaller, more intimate communities. Users are not leaving because a better platform came along. They are leaving because they no longer feel the relationship is worth it.

And now GenAI has arrived, funded by many of the same investors who built social media, with the same playbook. Grow fast. Capture users. Monetize the data. Build the moat through scale, not trust.

Enterprise adoption tells an even starker story. Despite hundreds of billions in investment, enterprise AI adoption has barely moved — declining from 14% to 12% in recent surveys. RAND Corporation research puts AI pilot failure rates at 80–95%. Companies are spending enormous sums on technology that is not delivering.

The extractive model, it turns out, does not scale into the age of AI.

Why the Traditional Moats Are Failing

Let us examine each traditional moat in the new environment.

Network effects assumed that more users made the product more valuable. But when users do not trust the platform, scale becomes a liability. A billion disengaged users is not a moat. It is a slow-motion churn problem.

Proprietary data assumed that data was the scarce resource. But AI is democratizing data analysis at a speed nobody anticipated. The advantage of having more data is eroding as models improve and open-source alternatives proliferate. DeepSeek demonstrated in early 2025 that a fraction of the compute investment could produce comparable results. The data moat is not what it was.

Switching costs assumed that friction kept customers loyal. But friction is not loyalty. It is resentment waiting for an alternative. The moment a credible option appears, customers built on switching costs will leave. And in the age of AI, credible alternatives appear overnight.

Regulatory capture assumed that incumbents could use regulation as a shield. But regulation is a two-edged sword — it protects incumbents and attracts scrutiny simultaneously.

Lock-in assumed that customers had no choice. But customers always have a choice. They may not exercise it today. They will exercise it when the cost of leaving becomes less than the cost of staying.

None of these moats have disappeared entirely. But none of them are sufficient in a world where AI commoditizes technology faster than any previous force in the history of the industry.

The One Moat That Remains

There is a moat that AI cannot commoditize. That regulation cannot manufacture. That competitors cannot copy overnight.

The devoted customer.

Not the casual user. Not the free account. Not the one who churns the moment a cheaper option appears. The customer who believes in what you are building. Who tells others about you unprompted. Who stays not because they are locked in but because they want to be there.

The data on devoted customers is striking.

Bain & Company research consistently shows that increasing customer retention by just 5% increases profits by 25–95%. The Net Promoter Score research from Fred Reichheld at Harvard Business School demonstrated that the single most predictive metric of business growth is the answer to one question: would you recommend this company to a friend?

And in an age when customer acquisition costs are rising everywhere, the company whose customers arrive through devotion rather than advertising has a structural cost advantage that compounds over time.

Zero customer acquisition cost is not a metric. It is a philosophy made visible.

The City-State Model of Business: The Sovereign Brand

The great companies of the next decade may look less like platforms and more like medieval city-states.

This sounds like a metaphor. It is also a business model.

Inside the walls of a medieval city-state were workers, merchants, craftspeople, and citizens. They were bound together not by contract but by covenant. They shared the same values, the same risks, the same rewards. The wall was not primarily a military structure. It was a statement of identity. We are this. Not that.

Workers who believe in the mission take lower salaries in the early days because they are building something that matters to them. Customers who trust the company become investors because they want a stake in what they are helping to build. Investors who share the values provide patient capital instead of growth-at-all-costs pressure.

Each group reinforces the others. The workers build a better product because they believe in it. The customers promote the product because they love it. The investors support the company through the difficult periods because they trust the founders.

This is not idealism. It is a flywheel.

The Data on Community-Powered Growth

The evidence for community-powered growth is accumulating rapidly.

Duolingo built 500 million users with a fraction of the marketing spend of traditional education companies by creating genuine engagement rather than extracting attention. Their daily active user rate is among the highest of any consumer application in the world.

Notion grew to a $10 billion valuation with a sales team that was, for years, essentially nonexistent. Their customers were their sales team.

Figma, acquired by Adobe for $20 billion, built its community of designers through genuine product love and user-generated tutorials. The community was the moat. Adobe understood that.

In each case, the pattern is the same. A product that genuinely solves a real problem. A community that forms around that solution. A flywheel that makes traditional customer acquisition largely irrelevant.

What This Means for AI Companies Specifically

The GenAI industry is making a category error.

It is treating AI as a product feature when the real opportunity is AI as a relationship.

The tools that will win are not the ones that produce the most content the fastest. They are the ones that make people feel more like themselves, not less. More capable. More authentic. More able to do the thing they want to do.

There is a profound difference between a tool that replaces human thinking and a tool that augments it. One produces dependency. The other produces devotion.

Steve Jobs called the computer a bicycle for the mind. Generative AI, in its current dominant form, built a wheelchair.

The companies that recognize this distinction — and build for augmentation rather than replacement — will find that their customers become their most powerful competitive advantage. Not because they are locked in. Because they genuinely do not want to leave.

The Metrics That Matter

If trust is the moat, then the metrics that matter are different from the ones Silicon Valley has traditionally tracked.

Not monthly active users. Net Promoter Score.

Not total revenue. Revenue per devoted customer.

Not growth rate. Retention rate.

Not customer acquisition cost. Organic referral rate.

Not total funding raised. Months of runway at current burn.

Not valuation. Unit economics.

The company with 1,000 devoted customers who pay, refer, and stay is building something more durable than the company with 1,000,000 casual users who tolerate the product until something better arrives.

The last moat is trust.

Stephen Klein is Founder & CEO of Curiouser.AI, the only AI designed to augment human intelligence. He also teaches at UC Berkeley. Alice 2.0 waitlist is now open — the first complete AI thought-leadership system designed to amplify individuality, not replace it. Join the waitlist at curiouser.ai. Curiouser is community-funded on WeFunder.