I have a somewhat one-sided view: In the AI era, there are actually no real moats anymore. If there is one, it’s only one—your user data.

1. Why do I say that in the AI era, all moats have become ineffective?
What did we used to talk about when we discussed barriers?
Technological patents, capital, supply chains, big brands, or an extremely talented R&D team.
But in the AI era, all of these have been flattened:
• Models are open source, becoming cheaper and capabilities increasingly similar.
• Core algorithms? Everyone is tuning similar APIs.
• Product features? The features you stay up all night developing can be copied pixel-for-pixel by others in just a few days.
Now, it’s no longer about “whether you can build it.” Because anyone can create a similar AI product. Technology is no longer a scarce resource, and even talent is converging.
2. Why is “user data” the only true moat?
Because in this stage where everything can be copied, data is the only private asset that cannot be replicated, plagiarized, or purchased externally.
The value of user data lies in its ability to run a terrifyingly closed loop:
• You have real users → then you have real interaction behavior data
• The more data you have → the better your AI is fed, and the more it understands users
• The better the product works → higher retention, more new users, and more data
Once this positive feedback loop starts, latecomers can pour in more money but still can’t catch up.
Competitors can copy your UI, your features, your prompts, even your business model. But they can never copy the accumulated, real, live user behavior data you’ve collected over time.
In the AI era: models determine your lower limit,
and exclusive data determines your upper limit.
Whoever can first run the “more users → better data → smarter AI → better experience” positive flywheel will hold the golden ticket to survival.
3. My conclusion
In the AI era, all “abilities” can be replaced.
Large models are becoming infrastructure (like water, electricity, and gas),
and the only barrier at the application layer is proprietary data.
“User relationships” and “user data” are the ultimate barriers.
Algorithms are public, tools are shared, but data is private.
Whoever holds continuous, authentic, fresh, exclusive user data truly gains the pricing power and survival rights in the AI era.
AI has flattened everything, leaving only this one moat.