Google connecting its world-building AI work to years of Street View imagery is a useful reminder that the model is only part of the product.
A historical dataset gathered across the real world is difficult for a new competitor to reproduce. When a new capability needs geographic context, continuity over time, or a grounded representation of real places, that archive becomes infrastructure.
This is what a meaningful data moat looks like. It is not simply having more records. It is owning a unique, compounding history that makes a new product possible or materially better.
The product lesson is to look beyond model access. Models will keep improving and becoming more widely available. Proprietary data can remain differentiated when it is collected through a product loop, grows more useful over time, and is difficult to recreate from public sources.
I had not tested Genie when I first wrote about it because access was tied to Google’s $200-per-month plan at the time. The strategic signal was already clear: old data can become the foundation for an entirely new product category.