137 AI Inference Costs – Serving Models = 99.7% Lower Over Two Years, per Stanford HAI Source: Nestor Maslej et al., ‘The AI Index 2025 Annual Report,’ AI Index Steering Committee, Stanford HAI (4/25) AI Inference Price for Customers (per 1 Million Tokens) – 11/22-12/24, per Stanford HAI Note: Axis is logarithmic; every axis tick represents a 10x price change AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising