Current rental medians for the RTX 5090 sit near $0.67–$0.74 per GPU-hour across marketplaces like Vast.ai and RunPod, with spot and community tiers as low as $0.40–$0.50 while on-demand listings reach $1.19. Sustained AI inference and smaller training workloads continue to support demand for its 32 GB GDDR7 and high tensor throughput, yet expanding availability—over 1,000 units reported on some networks with ~53% utilization—has produced modest recent softening. Tight retail supply and die allocation toward professional Blackwell SKUs limit new consumer inflows, creating a narrow window for price movement by month-end. Competitive pressure from decentralized platforms and interruptible capacity versus secure providers keeps outcomes closely balanced around the $0.65–$0.75 bands, with any acceleration in capacity or drop in utilization likely to cap upside.