What Is AI Unit Economics?

Question

How do I calculate AI unit economics for pricing and break-even decisions?

Quick answer

Formula: gross_margin_pct = (price_per_user_month - cost_per_user_month) / price_per_user_month * 100

• Assumption: cost includes generation, retrieval, reranking, embeddings, vector, cache, and infra.
• Assumption: pricing and costs are measured on the same per-user monthly basis.
• Assumption: model and usage assumptions are explicit and versioned.

Example: if price is $49 and cost is $9.80, gross margin is 80.0%.

Start with cost per user, then compare that against your current seat price or usage-based price. If the margin is thin, your next decision is usually to reduce the biggest cost driver or raise price before scaling.

The Core Questions

• What does one user cost per month?
• What does one request cost?
• At your price, is the margin healthy?
• Which assumption drives cost the most?

Why It Matters

• You can grow usage and still lose money if unit cost is too high.
• Small changes in tokens, cache, or retrieval can move margins a lot.
• Clear unit economics helps with pricing and roadmap decisions.

Fastest Working Method

1. Estimate cost per user/month with explicit usage and model assumptions.
2. Calculate break-even price so you know the minimum viable price floor.
3. Check which component moves margin the most before you cut quality or add complexity.

For most teams, that means using a deterministic calculator first, then moving into infra recommendations only after the scenario math looks viable.