Inference Cost Floor

Inference Cost Floor is the minimum per-unit price an agency can charge for an AI-powered deliverable before compute costs erase the margin. It moves for reasons outside the agency's control: energy, water, and data center capacity constraints push API pricing upward, and a provider's own losses can force repricing on short notice. The framework asks one question before quoting any AI-inclusive retainer: what does this deliverable cost to run at current token rates, and what happens to that number if rates rise 30 percent mid-contract? Forrester's 2027 predictions flag AI expansion colliding with real energy and infrastructure limits, which translates into price increases for API-dependent agency tools and compressed margins on AI-inclusive retainers. Agencies that price from a documented cost floor, rather than from competitor rates, can absorb a provider repricing without renegotiating the client contract. Those that do not end up funding a client's AI usage out of their own retainer.