Startup Moats Before and After AI

I love asking LLMs to plot subjective things like book and movie ratings. I can't believe we have a tool (AI!) that has infinite latent variables representing the collective opinion of humanity.

After a decade as a software engineer in Big Tech, I hung up my corporate hat when I realized the only career I could really justify in the post-AI era was becoming a founder. Today (2026) we are in a Cambrian explosion of building. But as AI has lowered the barrier to entry for builders, it has also weakened many of the traditional moats, the hard-to-copy advantages that protect a startup once everyone else sees the opportunity.

The idea for this chart began when I started questioning my core assumption that data (especially proprietary, real, and accurate data) was still a strong moat.

If AI models have already vaccuummed up so much of the internet, and can now generate synthetic datasets, what advantage is left for startups that go into the world and collect data no AI model has ever seen before?

I asked my current favorite model ChatGPT 5.4 Thinking to create a list of the top startup moats and their relative frequencies in online founder, startup, and VC discourse. Specifically, I asked ChatGPT to scan "accelerator and VC essays, founder interviews and podcast pages, plus public startup discourse on X and Hacker News."

Then I had the model assign a percentage of importance for each moat before and after the launch of ChatGPT. The result is a map of which moats fell victim to AI (data, having a superior product) and which moats became more valuable in the era of AI (accurate data, personal founder brand).