Strong AI demand for the RTX 5090's 32 GB GDDR7 and Blackwell tensor performance continues to constrain consumer supply, pushing secondary-market purchase prices well above the $1,999 MSRP and supporting elevated rental rates on marketplaces. Current on-demand medians hover near $0.64–$0.70 per GPU-hour across providers such as Vast.ai, RunPod, and Salad, with community and spot tiers often lower. Persistent purchases by AI builders for inference and smaller training workloads, alongside limited new retail inventory, create upward pressure that could keep November averages in the $0.55–$0.70 band. Key swing factors include any production ramp or shift toward professional SKUs, seasonal demand changes, and broader shifts in large-language-model deployment that favor or bypass consumer cards. Traders see roughly balanced odds because supply tightness and alternative data-center options remain closely contested.