Moats set the ceiling on value capture. Execution determines how close you actually get. They are fully independent variables.
Source: Don’t Call It a Moat by Yoni Rechtman, 99% Derisible (Mar 13, 2026)
The Core Argument
Code was always execution, not a moat. AI did not kill software moats. It relieved the bottleneck that made it hard for competitors to execute as well as you did.
The companies in trouble are the ones that confused an execution bottleneck for a moat.
Why AI Code Gen Is Different from Outsourcing
Outsourcing was a low-tech, high-friction attempt to commodify code. Huge barriers to entry, high management cost, bad results. It never actually relieved the bottleneck.
AI code generation differs in both kind and degree:
- Degree: So much cheaper and faster that it massively expands the market. Roughly at parity with outsourced teams.
- Kind: Outsourcing was useless to (even a burden on) great engineers. AI is a multiplier on their ability to move fast.
Code becomes free at the low end (non-technical operators solving problems with code) and amplified at the high end (great engineers doing more than ever). This compresses code’s value as a differentiator.
Moats and Execution Are Independent Variables
There are only a few real moats: data, network effects, brand, economies of scale. Execution is not one of them.
Google Search illustrates this perfectly. Probably the best business of all time because of multiple strong moats. But the execution is notoriously bad: no product direction, low velocity, bloated, kills products on a whim.
- Moats determine the ceiling on how much value you can capture.
- Execution determines how close to that ceiling you actually get.
Software development was a chokepoint on execution that looked like a moat because a limited number of people could do it well. That talent constraint created the illusion of a barrier to entry.
Everything Shifts Down One Tier
Every company used to solve two problems: the software problem and their actual hard problem. AI knocks out the software layer, so all effort goes to the hard part.
| Before AI | After AI |
|---|---|
| Pure software companies | Look like services businesses (easy to start, hard to differentiate) |
| Infrastructure companies | Look like software companies |
| Hard tech companies | Look like infrastructure companies |
For pure application software companies, there is no harder problem underneath. Software was the whole thing. When you knock out that layer, there is nothing left to create barriers to entry.
For everyone else, this is a gift. Resources go entirely toward the problems that actually differentiate you.
The Four Remaining Tech Roles
Rechtman argues tech companies converge on four archetypes:
- Product eng / vibe coder / PM – High velocity, high tool use generalists. Anyone can be commercial and product minded.
- Security / SRE / infra – The people stitching everything together, making it stable, secure, and robust as output volume explodes.
- Hot people – Sales, CX, people ops. Those who present an easy UX to the world and are pleasant to work with.
- Grown ups – The governor on an accelerating organization. Legal, finance, and non-technical equivalents of #2.
These latent traits have always cut across job titles and orgs.
Key Takeaways
- If your only differentiator was “we write good software,” you never had a moat. You had an execution advantage at a historically scarce chokepoint.
- AI relieved the chokepoint. Classical moats (data, network effects, brand, scale) remain unchanged.
- The right response is to focus resources on the hard problem underneath the software layer.
- Application-only software businesses are converging toward services economics: easy to start, hard to differentiate, lots of competition.
