The thing Naval was wrong about is code. He called it leverage -- the kind that works while you sleep, that lets one person do the work of a thousand. He wasn’t wrong then. But then is over.
We are entering a post-software world. And I don’t mean that loosely. I mean that building software is no longer a barrier to anything. You have an idea for an app -- great. Prompt it to a coding agent. It gets written, tested, deployed. The whole thing. By tomorrow morning. By someone who has never written a line of code in their life.
Software, the thing that made engineers the most valuable people in any room for the last two decades, is becoming a commodity.
What This Actually Means
For years, the startup game had a clear shape. You had an idea. You needed someone who could build it. If that someone was you, you had an enormous advantage -- speed, control, no co-founder dependency. If it wasn’t you, your first call was to a technical co-founder, and finding one was half the battle.
That shape is dissolving.
Someone with no technical background can ship an app in a weekend today. And it cost them nothing but a few good prompts. The technical barrier -- the one that separated “I have an idea” from “I have a product” -- is essentially gone.
Which raises an uncomfortable question for anyone building in tech right now: if anyone can ship, what actually matters?
The Arguments That Won’t Save You
The first instinct is to say taste will matter. Product thinking. Design sensibility. The ability to know what to build, not just how. And I think that’s true -- for now. The best products right now do reflect genuine human judgment about what people need.
But I’d be careful about how much weight you put on taste as a long-term moat. We’ve already watched AI get surprisingly good at creative work -- writing, image generation, music. The probability that it develops genuine product taste is not zero, and I suspect it’s higher than most people are comfortable admitting. Taste might buy you a few years. It probably won’t buy you a decade.
The second instinct is to say ideas will matter. And sure -- a brilliant idea is still a brilliant idea. But ideas without distribution, without lock-in, without network effects are just ideas. In a world where execution is free, ideas become more abundant, not more valuable.
What Will Actually Survive
Here’s what I keep coming back to: in a post-software world, the rules of startups stop being special. They start looking like the rules of every other industry that existed before software had its exceptional moment. And in every other industry, competitive advantage has always come from roughly the same three or four places.
Network effects. This is when the product gets more valuable as more people use it. Uber becomes more useful as more drivers join. Airbnb becomes more useful as more hosts list. The product and the network are inseparable, and the network took years to build. You cannot prompt your way to a two-sided marketplace with millions of participants. This is why I think companies like Uber, DoorDash, and Airbnb are genuinely durable in a post-software world. Not because their apps are irreplaceable (a better app could exist tomorrow) but because their networks are.
Switching costs. This is different from network effects, and worth separating. Spotify is a good example. Is Spotify the best music app imaginable? Probably not forever. Could a competitor build something technically superior? Easily, in the post-software world. But Spotify has eight years of my listening history. It knows my taste better than most people in my life do. It has playlists I spent hours building. The psychological and practical cost of leaving is real, even if a competing product is objectively better. B2B software operates the same way, but more so: when a company has deeply integrated a tool into its workflows, the cost of switching isn’t just inconvenience, it’s months of disruption and retraining. Even the Department of Defense reportedly has a hard time replacing Claude since it’s so deeply integrated with its workflows. Deep roots like these are durable.
Brand and emotional loyalty. This one is harder to build but real. Apple is the obvious example -- there are people who will buy an Apple product over a comparable competitor simply because of how they feel about Apple as a company. That feeling was built over decades of consistent design, storytelling, and identity. It is not easily replicated. In a world where the functional differences between products narrow because software is cheap and AI-generated, brand becomes one of the few remaining ways to differentiate. People will still choose based on who they trust and who they feel connected to.
Distribution. This is the one that doesn’t get talked about enough. If anyone can build the app, the person who wins is the one who can reach people. Audience, trust, community -- these become the scarce resource. This is actually why creators and media personalities launching products have been doing so well in recent years. The software was always secondary. The audience was always the asset. In a post-software world, that becomes impossible to ignore. If you have a hundred thousand people who trust you, you can launch almost anything. If you don’t, the best app in the world sits in silence.
A Counter-Argument Worth Taking Seriously
There’s a version of this argument that goes the other way. If AI makes it trivially easy to build, couldn’t a well-funded competitor also use that leverage to unseat incumbents faster? Network effects have been broken before. MySpace lost to Facebook. BlackBerry lost to iPhone. Each of those transitions felt impossible until it happened quickly.
Maybe cheap software makes those transitions faster -- a competitor can rebuild the app layer in a weekend and then pour all resources into attacking the demand side of the network. I don’t think this kills the network effect argument, but it does suggest that even companies with strong networks can’t get complacent. The moat is real but it still requires tending.
What This Means For You
If you’re a developer, a founder, or just someone who has built a career around the ability to ship software -- this is probably a strange moment to be sitting with. The skill you spent years developing is being automated in real time.
I don’t think that means it’s worthless. Understanding how software works will still matter for a long time. But it does mean that if your primary advantage is “I can build this and you can’t,” that advantage has an expiration date and that date is closer than you think.
The question worth asking now, before the expiration, is which of the four things above you’re building toward. A network. Deep switching costs. Genuine brand loyalty. Or an audience that trusts you.
Those are your moats in a post-software world. You can count them on one hand. And that’s actually clarifying, if you let it be.