One of my least favorite words in startup circles right now is “moat.” It’s everywhere in AI conversations:
“So what’s your moat? What if [insert big company] builds the same thing?”
“Data is the new moat!”
“Actually, evals are the new moat!”
It's a word that makes you sound sophisticated. But I've noticed something: the people who talk about moats tend to make a living through talking, not building. They are investors, commentators, advisors.
Real builders rarely talk about moats.
Consider the question: “What if [insert big company] builds the same thing?”
Perplexity was asked: “What if Google builds the same thing?”
Cursor was asked: “What if Microsoft builds the same thing?”
OpusClip was asked: “What if CapCut builds the same thing?”
None of these companies started by designing a moat. They started by building a product that solved a problem.
Your job as a startup is not to build a moat. It’s to build something people actually want. Users don’t care about moats; investors do.
Ideas are cheap. “AI can help you code.” Duh. Everyone knows that. The difference is turning that into a tangible product. And in the AI era, when prototypes that once took weeks now take hours, you don’t win by out-strategizing; you win by out-executing.
The other day I posted a demo video about Granola where I praised its innovative UX, and it went semi-viral on Chinese internet. Then I got dozens of comments saying "But they have no moat. This UX can be easily copied."
Here’s what those people miss: the UX is just the surface. Why was this team able to build it before anyone else?
They deeply understand their target users.
They have superior product taste, built from years of hands-on experience and hard-won intuition.
They iterated relentlessly with 100 beta users over a year, shipping fast, listening closely, sharpening their judgment.
Today that shows up as AI-enhanced notepads. Tomorrow it might be AI note-taking for phone calls.
You can copy features. You can’t copy taste, soul, or operating rhythm. It’s not about what they built; it’s about how they think.
So in an age when every AI launch video looks the same, what’s the “real moat”?
Perhaps that’s the wrong question to ask. While you’re strategizing about moats, the real builder has already shipped, tested with dozens of users, and iterated five times.
The truth: AI is moving so fast that nobody—not even Sam Altman—can predict the future.
The only way to learn what works is to ship, test, get feedback, iterate, rinse and repeat.
Whoever executes that loop with the most speed and quality has the best odds of winning, simply because they’ll have more data points.
More good ideas have been killed by overthinking than by under-thinking.
So stop talking about moats. Start shipping.
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.