Satellite imagery has become a cornerstone of modern geospatial intelligence, enabling scalable, repeatable, and globally consistent mapping capabilities. As the demand for up-to-date geospatial data intensifies - driven by defense, security, and civil applications - satellite-based cartography is increasingly positioned as a complementary, and in some cases alternative, solution to airborne and drone-based data acquisition. Compared to aerial or UAV platforms, satellites offer unique advantages in terms of global accessibility, rapid revisit, and operational independence. They enable mapping in denied or contested environments, where deploying aircraft or drones may be impractical or politically sensitive. This makes them particularly relevant for defense applications such as MGCP (Multinational Geospatial Co-production Program) and MUVD (Multinational Urban Vector Data), where standardized, interoperable, and frequently updated geospatial datasets are critical. From a technical standpoint, recent advances in very high-resolution optical satellites now support both 2D and 3D mapping workflows, including digital surface models and building reconstruction. While aerial data still provides superior resolution and flexibility for certain local use cases, satellite imagery has reached a level of maturity that allows it to address a growing share of mapping needs, especially when coverage, timeliness, and scale are prioritized over absolute precision. Satellite-based mapping also enables rapid mapping workflows, including crisis response, obstacle detection for air and ground navigation, and trafficability assessment. In such contexts, the ability to quickly task satellites and process large volumes of data becomes a decisive advantage. Artificial Intelligence is playing an increasingly important role across the cartographic production chain. In particular, deep learning techniques significantly enhance geometry extraction, enabling automated detection and vectorization of features such as buildings, roads and land cover. This contributes to improved productivity and reduced turnaround times. However, AI should not be seen as a silver bullet, especially for complex, specification-driven products like MGCP or MUVD. These datasets require high levels of semantic richness, consistency, and compliance with strict data models. While progress is being made, automated attribute extraction remains a key challenge. Emerging approaches leveraging Large Language Models (LLMs) and Vision-Language Models (VLMs) offer promising perspectives for semantic interpretation and attribute population, but their operational maturity is still evolving. As a result, human expertise remains essential, particularly for validation, enrichment, and handling edge cases. It is likely that, in the foreseeable future, high-quality vector databases will continue to rely on a hybrid approach combining AI-assisted extraction and manual refinement. Similarly, the transformation from structured geospatial databases to finalized cartographic products, respecting symbology, generalization rules, and user-specific requirements, remains largely manual today. Here again, generative AI and FMs (Foundation Models) could play a transformative role, by learning cartographic rules and automating map generation. Yet significant work remains to ensure reliability, explainability, and standard compliance. In conclusion, satellite-based mapping is entering a new phase of maturity, driven by both sensor capabilities and AI advancements. While it does not replace aerial or UAV solutions, it offers a powerful, scalable complement, particularly for defense and large-area applications. The integration of AI across the production chain holds great promise, but must be approached with realism: the future of cartography will be hybrid, combining automation and human expertise to deliver trusted geospatial intelligence at scale.