Prometheus is designed to be deployed in a distributed way. In large environments, Prometheus itself just becomes one small piece in a larger overall design.
You will likely want to deploy one or more Prometheus instances in every environment. Typically you follow your failure domain or logical isolation lines.
Then you will want to deploy an overlay layer to allow you a "single pane of glass" view.
This can be done with things like Thanos[0] or Mimir[1]. I've use Thanos for several of these kinds of deployments. I feel it has a better overall design, and doesn't suffer from the single cluster SPoF issues that Mimir does.