**Trader consensus on A100 rental prices at end-November 2026 centers on the $0.75–$1.25 per GPU-hour range, reflecting a tight balance between expanding supply of older Ampere GPUs and steady demand from inference and fine-tuning workloads.** Newer Blackwell B200 and Hopper H100/H200 clusters continue absorbing high-end large language model training, freeing A100 capacity on marketplaces like Vast.ai and RunPod and exerting downward pressure on spot and on-demand rates. At the same time, established pipelines favor A100’s cost-performance for production inference and smaller models, supporting a price floor above $0.50. Recent weeks show median on-demand listings clustered near $1.45–$1.82 with cheapest verified offers dipping to $0.40, while spot availability adds volatility. Key near-term catalysts include additional hyperscaler and neocloud deployments plus any shifts in AI capital expenditure that could tip the narrow band outcome.