Glacier-fed streams are the cold, ultra-oligotrophic, and unstable streams that are fed by glacial meltwater. Despite these extreme conditions, they harbour a diverse and abundant microbial diversity that develops into biofilms, covering the boulders and sediments that form the streambed. These biofilms play key roles in ecosystem processes and exert a direct influence on downstream biogeochemistry. Here we aim to define the genomic potential of glacier-fed stream microbial communities using metagenomic analyses. As a reference point, we first present a global inventory of cryospheric microbiomes, and find shared taxonomic, functional and phylogenetic features that shape the bacterial communities globally. However, we also denote how these ecosystems remain understudied, and thus further work is required to characterise fully the microbiome of cryospheric ecosystems. Using the dataset of metagenomes generated by the Vanishing glaciers project, we then unravel strategies that microbes developed to thrive in the harsh environmental conditions of glacier-fed streams, including the importance of biofilm formation and cross-domain interactions. Moreover, using metagenome-assembled genomes, we find a unique phylogenomic diversity that harbours distinct genomic features. Limited knowledge exists on how glacier influence shapes bacterial communities in glacier-fed streams. However, improving our understanding is crucial to better forecast how climate change will affect this extreme, yet endangered ecosystem. Using the global dataset of metagenomes and environmental parameters collected by the Vanishing glaciers project, we shed new light on the future of the glacier-fed stream microbiome. We first project environmental parameters onto future scenarios of climate change using predicted changes in glaciology and climate. These predictions corroborate conceptual models that forecast the "greening" of glacier-fed streams, and we further link this process with glacier size. Moreover, using a modelling approach and environmental, glaciological, and climatic variables as covariates, we forecast the changes in abundance of 2333 strains at a global scale. These models predict ecological shifts associated with the phylogenetic structure of the microbiome. Additionally, we find an association between these forecasted changes and the functional potential of these genomes, but also their genomic bulk features. This altered microbiome is expected to play a more important role in future glacier-fed streams, particularly in carbon cycling. To better understand how glacier-fed stream microbial genomes are shaped by glacier influence, we create a new method that identifies phylogenetic clades that drive this relationship. This approach allows us to identify genomic optimisation patterns along the gradient of glacier influence, highlighting the importance of Gammaproteobacteria in shaping the genomic landscape of glacier-fed streams. Overall, this work serves as a reference resource for climate change microbiology, by providing a global dataset of cryospheric microbiomes, a modelling framework that allows to forecast the abundance of bacterial strains, and other methods to analyse microbiomes in a changing environment. Owing to human-induced climate change, the cryosphere is rapidly shrinking. Thus, targeted efforts are still required to unravel the threatened biodiversity of cryospheric ecosystems, and anticipate potential changes in ecosystem functioning.