Variable cost is the AI-SaaS trap
Traditional SaaS has near-zero marginal cost; AI features don't. A heavy user can erase your margin. Cap usage, tier pricing, or meter it. Model the underlying cost on the AI app cost estimator.
Price your AI feature with a healthy margin over its API cost.
Traditional SaaS has near-zero marginal cost; AI features don't. A heavy user can erase your margin. Cap usage, tier pricing, or meter it. Model the underlying cost on the AI app cost estimator.
The AI SaaS Margin Calculator works out the price you should charge per user each month so that your product hits a target gross margin. You give it three numbers: your AI cost per user (the model or inference spend one active user drives in a month), your other cost per user (everything else variable, such as third-party APIs, storage, bandwidth, or support), and the target gross margin you want to keep. It adds the two per-user costs into a total cost of serving one user, then divides that by one minus your margin to produce the required price. The main drivers are therefore your combined per-user cost and how aggressive a margin you set: raise either the AI cost or the margin target and the price the tool recommends climbs.
The key trade-off to watch is that a high target margin protects profitability only if your cost per user is accurate and stable. AI inference costs often move with usage, so a power user can consume far more than the average you entered, quietly eroding the margin the price was built for. A practical tip is to model your heaviest users rather than the average, or run the calculator twice—once for typical usage and once for a worst case—so the price you set still holds when consumption spikes. Also revisit the inputs whenever model pricing or feature usage changes, since a price that delivered a healthy margin last quarter can slip below it as costs drift.
Price = total cost per user ÷ (1 − target margin). Include the AI API cost plus your other per-user costs, then mark up to your desired gross margin. Usage-based AI cost makes this trickier than fixed-cost SaaS.
Because your cost per user is variable — a power user can burn far more tokens than a light one. Flat pricing on variable AI cost can turn your best customers into loss-makers unless you cap or tier.