Foundation models are the frontier of EO AI precisely because they need far less labelled data than training from scratch — but understanding why requires understanding how they were pre-trained in the first place. This course goes deep on two specific models, Prithvi-EO-2.0 and DINOv3, covering their pre-training objectives, their task heads, and — as a matter of policy throughout this curriculum — an honest comparison of what you get for free (zero-shot) versus what local fine-tuning buys you.