# Your repo isn't worth protecting
Earlier this year, Sahil Lavingia open-sourced the entire Gumroad codebase. Not just a developer tool or a small library. The complete website that processes millions in creator revenue every year.
This wasn't (just) a publicity stunt. It was a signal for those paying attention that the basic economics of software are changing.
We're entering a world where AI can clone any low / medium complexity website in days for a few hundred dollars in API credits. The bottleneck in software is shifting completely. It's no longer about how fast you can type code. It's about knowing what to build in the first place.
## The great execution liberation
For decades, software development was bottlenecked by execution speed. But with the marginal cost of building new software approaching zero, the ability to write code faster stops being an advantage. Instead of a moat, code becomes a commodity.
Creating software has always been filled with friction. Whether it's debugging CSS layouts, setting up infrastructure, wrestling with imports for the hundredth time, or rewriting the same JavaScript sort function - these bottlenecks have defined the pace of development for a generation.
AI tools are well on their way to eliminating these bottlenecks entirely.
Today, I can describe a complex component and watch Claude generate it in seconds. CSS that would have taken a skilled developer hours (or me, embarrassingly, days) appears instantly. I don't have to remember syntax, Tailwind classes, or API details. Entire features get generated faster than I can type.
The sea change in software development technique is like calculators arriving in a world of long division. The old skill won't disappear overnight, but suddenly the competitive advantage shifts entirely.
Developers are changing from writers to editors. Instead of coding everything from scratch, they can now prompt, review, refine, and integrate code - all written by a model. We're becoming operators who happen to use code as a tool, rather than artisans who occasionally get to think about strategy.
This will not only change the day-to-day for developers, but the dynamics of the industry as a whole.
## What actually becomes defensible
Free things are never a competitive advantage. Even if something was valuable in the past, as soon as it becomes universally accessible, the competitive advantage shifts to whatever remains scarce.
We've already seen this pattern in other creative fields. When desktop publishing killed typesetting as a craft, the value didn't disappear - it moved. Suddenly everyone could set type, so the premium shifted to content, distribution, and brand. When execution becomes commoditized, the only thing with scarcity is taste.
Software is following the same pattern, just compressed into months instead of decades.
The new defensible assets aren't technical. They're strategic, human, and relational:
- →Product decisions. Knowing which features to build and which to skip. Understanding how to sequence development for maximum impact. Getting positioning and presentation right.
- →Deep user understanding. Direct relationships with customers. Insights into how people actually work, not how you think they work. The ability to spot problems users don't even know they have yet.
- →Brand and trust. Reputation for solving real problems consistently. Credibility that comes from understanding a domain deeply enough to make good predictions about where it's heading.
These assets compound over time through actual human interaction. No model can build trust by reading documentation.
## Why every codebase will become open source
When code has no defensive value, keeping it secret becomes a liability.
Open source used to be a philosophical choice. You gave away code to build community, establish standards, or scratch an ideological itch. Closed source was how you captured value.
That equation is backwards now. Closed source signals that you don't understand where value actually lives.
The smartest companies are already acting on this realization. Microsoft open-sourced the entire VS Code codebase (and allowed Cursor to disrupt them through having better taste). Supabase gives away everything except the operational expertise to run it at scale. Even Gumroad - a for-profit company processing real money - realized their code wasn't the asset, their distribution and reputation was.
These companies aren't being generous. They're being realistic.
## The hidden implications nobody wants to discuss
How many companies' entire value proposition is "we built it so you don't have to"? There's nothing novel about the product or execution, just a moat of existing code.
When building becomes trivial, these businesses evaporate. Not evolve. Evaporate.
The timeline is brutal in its simplicity because AI coding isn't scaling linearly, it's scaling exponentially. Within 18 months, AI will be able to handle most well-defined features better than humans. The complex stuff - low-level systems, novel algorithms, legacy spaghetti - gets maybe an extra year or two, but it's only a matter of time.
This creates a stark choice for everyone building software.
## The paths forward
- →Path one - expansion. Embrace strategic thinking, user focus, and product ownership. Learn to ask "Should we build this?" not just "How do we build this?" Understand user workflows. Pay attention to business metrics. Make decisions about what problems actually need solving.
- →Path two - the deep specialization bet. Double down on complexity - embedded systems, kernel development, real-time trading algorithms. But even within these domains, AI capabilities compound exponentially. Today's "too complex for AI" is tomorrow's tutorial example. You're not wrong that these fields need expertise. You're just betting the timeline is longer than it actually is.
- →Path three - obsolescence. Keep optimizing for pure implementation speed and technical perfection. Stay focused on execution while strategic thinking happens elsewhere. Become a coding machine (and watch as AI becomes a better one).
The first path has the highest upside. The second might buy you time but leaves you vulnerable. The third is just a slow death.
The developers who thrive will be those who can think like product creators. They'll understand users, make feature tradeoffs, and use AI as a powerful implementation tool rather than a threat to their relevance.
This change is already happening. Great developers are pushing for more involvement in product decisions. Companies are rebuilding around smaller, more strategic teams. Job descriptions and hiring processes emphasize product thinking alongside technical skills.
What AI enables isn't new thinking - it's the democratization of what only elite companies could previously afford. GitHub, Stripe, and Linear succeeded by treating engineers as product thinkers. But that only worked because they could be extremely selective in hiring.
Now AI eliminates the tradeoff. Any company can have engineers who think strategically because the execution bottleneck is gone. The practices that made these companies exceptional are becoming table stakes - small teams shipping at the speed of thought, not the speed of typing. Engineers owning entire features from conception to metrics. Product velocity limited chiefly by decision-making, not implementation.
The elite company playbook is becoming everyone's playbook. What separated the best from the rest was their ability to execute on product vision quickly.
## What this means for you
If you're a developer, the question isn't whether this shift will happen. It's how you'll adapt to it.
Start asking questions beyond your immediate tickets. Look at user analytics and feedback. Understand how your company makes money and what drives growth. Learn to think about the problems your code is supposed to solve, not just how to build a beautiful system.
The future belongs to people who understand why Sahil open-sourced Gumroad.
He didn't give away something valuable. He acknowledged what was already true - in a world where AI can clone your features in a weekend, hoarding code is like hoarding arithmetic.
The developers who thrive won't be the ones who write the best code. They'll be the ones who know which code should never be written at all. They'll be the ones who can use tactics like open sourcing their codebase to build the things that actually matter - community, trust, and reputation.
Your repository was never your moat. It was just friction.
The real asset was always your judgment. AI just makes that impossible to ignore.