Stripe billing update targets AI startup pricing challenges

On Monday, Stripe released a preview of a new billing feature that could help artificial-intelligence startups pass through the underlying costs of AI model usage to their customers. The feature allows startups to charge an automatic markup percentage on token usage—the expense incurred by using Large Language Models based on token consumption. Under this system, a startup can select the AI models it uses, and the billing tool will track the API prices of those models, record customer token usage, and apply the profit-margin markup automatically. For example, a startup can automatically charge a 30% margin over the raw LLM token costs that it pays to model providers. In describing the feature, Stripe stated: “Say you’re building an AI app: you want a consistent 30% margin over raw LLM token costs across providers. Billing automates the process.”

This automation addresses a critical pricing challenge for agentic startups—startups building AI agents that perform tasks autonomously. The more customers use these autonomous agents, the more tokens they consume from underlying model providers like OpenAI, Google Gemini, and Anthropic. Without usage caps, this unpredictable consumption can quickly force a startup to operate in the red. While some companies, such as the AI-powered code editor Cursor, have previously adjusted their pricing models from unlimited use to rate-limited usage with extra fees for consumption, Stripe’s new tool tracks API prices, records customer token usage, and applies the profit-margin markup automatically.

The feature works alongside Stripe’s own AI gateway, as well as existing third-party AI gateways—tools that provide access to multiple AI models—such as Vercel and OpenRouter. Startups looking to manage these costs can contrast the options:

• Stripe’s billing feature: Allows startups to set their own custom profit-margin markup (such as 30%) across various model providers, automatically tracking API prices and customer usage.
• OpenRouter: An existing gateway that grants access to over 300 models, which charges a flat 5.5% markup over token fees on its first-tier plan and includes built-in budget controls.

The new billing feature is currently in waitlist mode, and Stripe has not announced when it will be generally available. According to Stripe product manager Miles Matthias, Stripe is not currently charging its own markup on the gateway.

Why it matters

Stripe’s new billing feature allows AI startups to automatically apply profit-margin markups to underlying LLM token costs. This automation potentially solves a critical pricing and margin-management challenge for agentic startups, which face highly unpredictable operational costs from underlying model providers.