Foundation Models (FMs) are large, general artificial intelligence (AI) models that can be applied to a range of AI and machine learning (ML) tasks. NASA's Office of the Chief Science Data Officer is creating an FM for each division in NASA's Science Mission Directorate. In this talk, I will summarize ongoing work by the Planetary Science Division to create an FM for the Moon, an exciting and timely target for such an endeavor. Recent missions have gathered a large and diverse set of multimodal datasets that inform our understanding of the Moon's interior structure, surface geology, and processes operating from the surface to the core. This initial lunar FM will serve as a basis for downstream community-developed ML models that enable tools and approaches advancing NASA's long-term lunar exploration and discovery goals.

Mars orbital archives now contain enough imagery that finding morphologically similar features is bottlenecked by search, not data. We present a planetary-scale similarity search system built on foundation-model embeddings over the full CTX Murray global mosaic (~26.9M indexed locations). A Vision Transformer pretrained via self-supervised learning on millions of CTX patches produces embeddings that capture surface texture and landform semantics without any labels. Deployed as a quantized vector index on a single server, the system supports sub-second instance-level retrieval ("find terrains like this"), geo-filtered search within regions of interest, and interactive relevance feedback for iterative refinement. The system is publicly accessible at findmars.space.