Cuatro Ciénegas, located in Coahuila, Mexico, is an ancient aquatic system with unique geological and biological characteristics, harboring diverse microbial communities adapted to extreme oligotrophic conditions. This region’s pools and channels, remnants of prehistoric marine ecosystems, provide an exceptional opportunity to study microbial evolution and adaptation. Among the most striking formations are stromatolites, lithified microbial structures that persist over geological timescales. To explore the viral and microbial diversity in these ecosystems, we conducted a metagenomic analysis of a stromatolite and six microbial mat layers from Poza Roja. DNA was extracted and subjected to Illumina deep sequencing, followed by taxonomic classification using BLAST and Kraken2. This approach revealed a complex virome, with a notable presence of giant DNA viruses, including Bodo saltans virus, Tupanvirus deep ocean, Tupanvirus soda lake, Powai lake megavirus, Moumouvirus goulette, and Cotonvirus japonicus. Additionally, members of the Megavirus, Mimivirus, and Moumouvirus genera, as well as representatives of the Mimiviridae family, were detected in both microbial mats and stromatolites, indicating their potential role in microbial community structure. Comparative analyses revealed distinct viral distributions between microbial mats and stromatolites. While some viruses were shared across both environments, others exhibited strong habitat specificity. Several phages, including Escherichia phage vB_EcoM_Bp10, Emaravirus syringae, and Lakivirus lakamhaense, were found exclusively in microbial mats, suggesting a role in bacterial regulation and microbial turnover in these dynamic environments. In contrast, stromatolite-exclusive viruses, such as Ocetevirus paratemnopterygis, Cyclovirus kisikisi, and Alphacoronavirus HKU33, suggest viral adaptations to long-term microbial interactions within stable lithified structures. The presence of Acanthamoeba polyphaga mimivirus and Bodo saltans virus further suggests that giant viruses influence protist populations, potentially playing a regulatory role in microbial trophic networks. To investigate virus-host interactions, we constructed a viral co-occurrence network based on graph theory, where edges represent virus-host associations. This network analysis revealed strong interactions between giant viruses and microbial taxa such as Aeromonas, Gordonia, Acanthamoeba, Vibrio, and Serratia. The presence of these associations suggests that giant viruses contribute to microbial turnover, influence biogeochemical cycles, and modulate ecosystem stability in extreme oligotrophic conditions. Viral predation may regulate microbial populations, contributing to nutrient remineralization and microbial succession within these ecosystems. These findings underscore the ecological and evolutionary importance of giant DNA viruses in ancient microbial communities. Their structured distribution across microbial mats and stromatolites highlights their role in shaping microbial diversity and influencing metabolic interactions. The combination of high-throughput sequencing, computational taxonomic profiling, and network-based ecological inference provides a robust framework for studying virus-host dynamics in extreme environments, offering insights into microbial resilience and ecosystem evolution.