Recent A100 rental rates cluster near $0.89–$1.00 per GPU-hour on leading neocloud and marketplace platforms, with spot capacity dipping to $0.68 and hyperscaler on-demand listings higher. This positioning stems from ample legacy Ampere supply entering secondary markets as operators shift workloads to H100, H200, and B200 GPUs for superior throughput on frontier training and inference. Persistent but price-sensitive demand for A100 in fine-tuning, smaller models, and cost-sensitive inference creates a narrow band around $1.00, while ongoing capacity releases and generational substitution exert mild downward pressure. Traders see the tight $0.75–$1.25 contest as reflecting stable near-term fundamentals absent major supply disruptions or sudden demand spikes before October resolution.