**Trader consensus clusters around the $0.75–$1.25 per GPU-hour band for A100 rentals at end-November 2026, reflecting balanced supply and substitution dynamics.** Current marketplace data show on-demand medians near $1.40–$1.80 across providers, with spot and community platforms like Vast.ai frequently below $0.90 and indices settling around $0.92. Steady inflows of older Ampere hardware from hyperscaler fleet refreshes add capacity, while AI workloads continue shifting toward higher-throughput H100/H200 and Blackwell GPUs for frontier training. A100s retain demand for cost-sensitive inference and fine-tuning, supporting a durable floor, yet persistent downward pressure from newer silicon and secondary-market liquidity prevents sharp rebounds. With roughly eight weeks until resolution, any surge in mid-tier inference demand or delays in Blackwell deployments could tilt outcomes within the narrow leading buckets.