Stripe Wants to Turn AI Costs Into a Profit Center

This new billing feature, currently in a preview accessible via a waitlist, allows AI companies to automatically add a customizable markup to the token usage fees they incur from large language model providers. This means a startup can set a consistent margin, for instance 30%, on top of the raw token costs from providers like OpenAI, Google, and Anthropic. The system is designed to track API pricing across different AI models, monitor a customer's token consumption, and then automatically apply the preset profit margin when generating invoices. This level of automation addresses a significant pain point for AI startups, which have been struggling to price their services effectively while covering the fluctuating costs of third-party AI models. Stripe's tool integrates with third-party gateways such as Vercel and OpenRouter, offering flexibility for developers. Notably, other platforms like OpenRouter already exist in this space, charging a 5.5% markup on token fees in their entry-level plan, though Stripe has stated it is not currently applying its own markup through its gateway. This move is part of a broader strategy for Stripe to become the essential financial infrastructure for the AI economy, a sector growing at an unprecedented rate. The company's leadership sees a fundamental shift in software economics, moving away from fixed costs to on-demand, usage-based models, a concept Stripe's CEO Patrick Collison compared to software being "cooked like a pizza" at the moment of use.