Go on YouTube right now and search “vibe coding.” I’ll wait. You’re going to find a panoply of guys — and it’s always guys — telling you how they “made a $30,000 app,” or a “$100,000-a-month app,” or, if they’re feeling especially generous with the truth, a “million-dollar-a-month app.” Peel back the layers on half of these videos and you’ll find out something funny: they’re not actually running these apps. They’re not fielding support tickets. They’re not getting yelled at in one-star reviews at 11 p.m. because the payment flow broke. What they’re actually showing you is how easy it now is to build something that resembles a real business, and how easy it is to copy the homework of somebody who already did the hard part first.
It’s cringe. Frankly, it’s some of the most cringe content on the entire platform, and that’s a genuinely competitive category. It’s also, annoyingly, telling you something true, and the truth is bigger than any one grifty thumbnail with a screenshot of a Stripe dashboard on it: production capacity — the thing that used to separate the people who could build software from the people who couldn’t — is no longer scarce. And when scarcity is no longer relevant, the moat that it built doesn’t get weaker: it stops existing.
Think of it like an obstacle course built entirely around the assumption that gravity exists, and then somebody quietly switches gravity off halfway through the race. Every wall you had to climb, every gap you had to clear — all of it was calibrated around a resistance that just isn’t there anymore. The people still running the old course don’t realize yet that they’re not being tested on the same thing as everyone floating past them.
Here’s the thing — this isn’t actually a story about coding. It’s a story about what happens to competition once the barrier to entry falls out of the bottom of the market entirely.
The Inherited Moat Fallacy
For as long as software has been a business, competition got shaped by a small, fairly boring set of factors. How unique was your idea. How much talented labor could you get your hands on, and the more talented, the better. How much money did you have to keep that labor fed, retained, and pointed at the right problems, while also handling the unglamorous stuff — marketing, outreach, the operational plumbing that keeps a company alive between good ideas. Stack those three together and you get what I’d call a business’s inherent production capacity: its raw, simple ability to make the thing at all.
What almost nobody says out loud, because it’s a little uncomfortable, is that inherent production capacity was never just a business’s ability to make things. It was simultaneously the moat that kept everyone else out. Those aren’t two separate facts sitting next to each other. They’re the same fact, wearing two different outfits.
I’ve seen this from the inside more than once, as a designer. Plenty of companies are out there running a product that’s an ugly, cobbled-together mess — chaotic process, zero innovation, held together with duct tape and momentum — and it keeps printing money anyway. Not because the product is good. Because nobody else could afford to build a competitor and take a real swing at them. These are businesses that should have been run into the fucking ground years ago on merit alone—and weren’t, because merit was never the test. Capital was. Call it the Inherited Moat Fallacy: mistaking “nobody can afford to compete with us” for “we’re good,” when it was always one and never the other.
It’s the business equivalent of a nightclub with a bouncer who doesn’t actually check anything about you except the size of your wallet. Doesn’t matter if you’re charming, doesn’t matter if you’d be fun to have inside, doesn’t matter if you’d actually improve the room. Can you pay the cover? Fine, in you go. For a long time, “can you pay the cover” was functionally the entire admissions test for competing in software. The line outside just got a lot shorter to walk.
AI didn’t invent this problem. It just deleted the wall that was hiding it.
What actually got disrupted
The disruption here isn’t merely that everyday people can make software now. Plenty of people said that part out loud already and moved on too fast, mostly because it’s the flashy, easy-to-screenshot part of the story. What actually matters is what that capability symbolizes to everyone downstream of it: production capacity no longer protects a mediocre business or a mediocre product from getting completely annihilated in the marketplace of ideas. That protection is just gone. It’s not weakened, it’s not “under pressure,” it’s gone, the way a levee is gone the moment the water’s already on the other side of it.
And the flip side of that is worse news for lazy operators everywhere: a mediocre idea isn’t enough either, not anymore. Somebody can now look at what you built, copy it, change 10% of it — the logo, the onboarding flow, the pricing tier names — and start eating your market before your Q3 planning doc is even finalized. That’s not a hypothetical scenario I’m constructing to make a point. That’s Tuesday. That’s an actual Tuesday for an actual founder somewhere right now, and they probably don’t even know it yet. The timeline for “somebody clones your product” used to be measured in years, sometimes never, because cloning it required the same capital and talent wall you’d already climbed. Now it’s measured in however long it takes someone motivated to sit down with a prompt window and a weekend.
So if production capacity isn’t the moat anymore, what is?
The New Way
Three things end up mattering, and only one of them is genuinely new. The other two were always in the room — businesses just had the luxury of treating them as optional for a while, and that luxury is over.
Conceptual integrity never left the building. How unique is the idea, particularly when that uniqueness is in service of a genuinely common need — not uniqueness for its own sake, which is just a different flavor of mediocrity with better branding. The harder something is to conceptually design in the first place, the harder it is for some guy with a laptop and a caffeine addiction to knock off in a weekend. This is the one piece of the old model that survives completely intact, because you can’t vibe-code your way into an idea nobody else thought of. You can only vibe-code your way into executing somebody else’s, slightly worse, six weeks later.
User experience is the one businesses used to get to skip, and can’t anymore. The gap between something that feels like a real product and something that feels like expensive AI slop was never a features list. It’s whether anyone thought about the actual human being who has to sit there and use the thing. Walk into a grocery store where nothing is where it’s supposed to be — the milk’s in four different places, the checkout lanes have no signage, half the aisles dead-end into a wall — and you don’t think “well, they have every product I need, so this is a great store.” You think “I hate it here,” and you start driving to the other grocery store, the one that put a little thought into where your feet would actually go. Software works the same way, and here’s the part that should genuinely embarrass an entire industry: the irony is that faster production should be good for this, not bad. If building the thing takes a fraction of the time it used to, that’s more runway for research, for testing, for actually talking to the people you’re building for, not less. What I’m watching happen instead is that most people are taking that freed-up runway and using it to ship faster garbage. That’s a choice. It’s not a law of physics, and it’s not an inevitable consequence of the tools getting better. It’s just what happens when speed becomes the only thing anyone’s optimizing for.
Marketing is the big one, and it’s the one nobody wants to hear, because admitting it means admitting the coding was never the hard part to begin with — it just felt like the hard part because it used to take the longest. In a world where production capacity is functionally unlimited, the bottleneck isn’t whether you can build something. It’s whether you can get eyeballs on it once you have, and whether those eyeballs believe what you’re showing them. Picture a farmers market where every single stall is selling the exact same tomato. Same size, same color, same price, grown from the same seed stock two rows over. The stall that sells out first isn’t the one with the objectively superior tomato — there isn’t one. It’s the stall with the guy who actually talks to people, who makes eye contact, who gives you a reason to stop walking. That’s the whole game now. I mean accurate eyeballs, specifically, and that part matters just as much as getting them in the first place. If your marketing is showing screens that don’t exist yet, congratulations, you’ve built vaporware with extra steps, and people can smell that from a mile away now in a way they genuinely couldn’t ten years ago. Nobody wants to be sold a lie twice, and in a market this saturated, you generally only get the one shot.
The part that hits closest to home
I’ve watched this play out in my own AI consulting work, and it’s not subtle. There was a stretch of time where knowing how to wire up n8n, or Zapier, or Make, functioned as a kind of professional moat — you knew the plumbing, your client didn’t, and that knowledge gap was worth real money on an invoice. That gap is basically gone. I sat down recently to sketch out a speed-to-lead automation for a client conversation, and Claude spat out a working version in about five minutes. Five minutes. The specialized knowledge that used to be worth a retainer is now worth a well-written prompt and about as much patience as it takes to finish your coffee. The tool used to be the differentiator. Now the tool is table stakes, and the actual differentiator is whether you understand the business problem well enough to point the tool at it correctly in the first place — which, it turns out, was always the harder skill. It just never had to prove itself before, because the tooling was scarce enough to hide behind.
Which brings me to the distinction that actually matters going forward, and it’s not “can you code” versus “can’t you code.” It’s the ability to make a thing versus the ability to produce one. I work alongside people who can build software completely from scratch — real engineers, the kind who can get down into every line and construct something out of nothing, and I respect the hell out of that. But knowing how to engineer was never the same skill as knowing how to build a product. Swap “engineer” for “vibe coder” and the sentence doesn’t change one bit. Anybody can prompt an app into existence at this point. A much smaller number of people understand why you’d bother with user personas, user journeys, information architecture, or a wireframe before you ever touch a finished screen. The average person wants to skip straight to the pretty part. So, embarrassingly, do a lot of engineers. The people who actually win from here forward are the ones who understand that a product starts as an idea with real clarity, long before it starts as a line of code — or a prompt.
The moat isn’t gone. It just moved. It used to live in who could afford to build. Now it lives in who actually knows what’s worth building, who it’s for, and whether anyone will ever hear about it. Everything else — the “I built a $30K app” flex included — is just noise standing where the wall used to be.