I started a project similar to this early this year [1] when I got frustrated with the lack of decent free resources documenting data centre build out. The plan was to focus on AI build out specifically, the ones costing billions and the recipients of all the Nvidia chips rather than the boring 'normal' datacenters.
I also wanted to add useful and accurate tools on things like local noise, water or grid impact. In addition to actually monitoring progress via satellite imagery and building a basic graph model for filling in missing attributes.
The reason why I stopped was that I significantly underestimated the effort required to complete such a project to the standards that I wanted to. As you can probably tell the site itself is AI assisted but all of the information was collected by hand which just takes time that I no longer had (~30 mins per site). The only way this would have made sense was as research project or something which sadly didn't line up.