A new billing feature lets startups automatically mark up AI model costs—turning token usage from a liability into a profit stream.
AI startups face a familiar financial headache: every customer request triggers costs from underlying AI models.
Stripe thinks it has a solution.
On Monday, the payments giant unveiled a preview of a new AI billing feature designed to help companies pass model usage costs directly to customers—while adding profit margins automatically.
In practical terms, Stripe wants to transform AI infrastructure costs into something closer to a built-in revenue engine.
Turning Token Costs Into Margin
AI applications typically rely on models from providers such as OpenAI, Google Gemini, and Anthropic.
Each request consumes tokens, which developers must pay for through API pricing.
Stripe’s system tracks those costs and automatically adds a markup percentage chosen by the startup.
For example:
- A startup could apply a 30% margin on token usage
- The platform automatically tracks model API pricing
- Customer usage is measured and billing adjusts in real time
Stripe describes the workflow simply: if a company wants a consistent 30% margin across multiple LLM providers, the billing system handles the calculations automatically.
In effect, it turns unpredictable infrastructure spending into structured unit economics.
Why AI Pricing Is So Difficult
Pricing AI products has proven surprisingly complicated.
Many startups initially launched unlimited usage subscription plans, only to discover customers could generate enormous costs behind the scenes.
A single heavy user can quickly run up large token bills.
Companies have responded with new pricing models:
- Tiered monthly subscriptions with usage limits
- Rate-limited plans that charge extra for higher consumption
- Pay-per-use structures tied to token volume
Last year, for example, the AI coding assistant Cursor shifted some pricing tiers away from unlimited usage to rate-limited access with additional fees.
The challenge is even greater for agentic AI startups, where autonomous agents run multiple tasks automatically.
More usage means more value for customers—but it also means higher token costs for the startup.
Stripe Is Building an AI Infrastructure Layer
Stripe isn’t just tackling billing.
The company recently introduced an AI gateway, which allows developers to access multiple models from a single interface.
That means developers can:
- Choose the best model for a specific task
- Switch providers without changing infrastructure
- Optimize for price, speed, or accuracy
The billing feature integrates with that gateway.
But it also works with popular third-party AI routing tools, including gateways offered by Vercel and OpenRouter.
That flexibility suggests Stripe wants to position itself as the financial backbone of the AI stack, regardless of which models developers choose.
Competition Is Already Emerging
Stripe isn’t alone in trying to solve the AI cost management problem.
Some AI infrastructure platforms already offer similar tools.
For example:
- OpenRouter provides access to more than 300 models
- It charges a 5.5% markup over token costs on its entry plan
- It also includes budget controls and spending limits
Stripe’s product manager noted on X that Stripe currently does not charge its own markup on its gateway.
However, the new billing feature itself is still in waitlist mode, and the company has not said when it will launch broadly.
Why This Could Matter
AI applications are fundamentally usage-driven businesses.
That makes cost control and pricing strategy central to survival.
If Stripe’s system works as intended, startups could gain:
- Predictable profit margins across AI providers
- Automated tracking of model pricing changes
- Simpler billing for complex AI workloads
In short, the tool could help transform AI from an unpredictable expense into a structured revenue model.
For startups operating on razor-thin margins, that shift could make the difference between scaling profitably—or burning cash with every API call.
TL;DR:
Stripe is testing a new AI billing feature that lets startups automatically mark up AI model token costs when charging customers. The system tracks API pricing, usage, and margins across providers like OpenAI and Anthropic—helping companies turn AI infrastructure expenses into a predictable profit stream.
AI Summary:
- Stripe launched a preview AI billing tool for startups
- It allows companies to mark up token usage costs automatically
- The system tracks API pricing across multiple AI models
- Designed to help startups avoid losses from heavy AI usage
- The feature is currently in waitlist mode
