On August 19, Stripe announced it would acquire OpenRouter, the AI model gateway. OpenRouter routes 400-plus models from more than 80 providers, giving customers access to the most suitable model for their needs. The deal was valued at upwards of $7bn, taking OpenRouter’s valuation 5x in just three months.
This looks like a very interesting acquisition for Stripe, and it’s worth thinking about what Stripe is actually buying and what it signals about where the payments industry is headed.
Stripe has made this bet before
Stripe acquired Bridge in 2024. Most coverage at the time described it as Stripe’s entry into crypto. But Bridge was not just a crypto bet. It was Stripe’s bet on infrastructure. With Bridge, Stripe now owns the conversion layer between fiat and digital, so any company building in crypto must route through Stripe.
OpenRouter extends the same logic. Companies building AI products increasingly want flexibility. They do not want to be tied to a single model – the space is evolving too quickly for that. They want to use different models for different tasks, based on cost, speed, and performance. OpenRouter lets Stripe sit within that decision flow, in real time.
In payments, Stripe became a neutral rail trusted by merchants and optimised at scale. Now, tokens are becoming central to the new AI economy. This acquisition is Stripe’s bet on executing the same playbook in the AI world. What is most interesting is that while other payments companies are still debating what AI means for their core product, Stripe has gone ahead and bet on owning the infrastructure layer.
What the Playbook looks like
Stripe is the payment processor for a majority of AI companies. Last year, it launched token billing to enable usage-based billing for customers. Now, with OpenRouter, Stripe is adding to this layer by owning the routing as well. Stripe will have a say in which model handles the request, at what cost, and at what performance threshold.
This creates a full economic stack for AI businesses. Stripe is uniquely positioned to handle money coming in (revenue, subscriptions, and usage-based billing); money going out (payouts and treasury); and money spent in the middle (compute and AI tokens). With this unified stack across all three parts, Stripe becomes deeply entrenched and very difficult to replace.
An old-school version of this would be a payments company with reconciliation, reporting, and cash flow management deeply integrated into settlement cycles. In the new-gen model, Stripe has tied payments to crypto and AI, making itself virtually embedded at every financial decision point.
This has a few direct implications for payment companies. Stripe is expanding the definition of financial infrastructure well ahead of the rest of the industry. If a payments company only accepts money, it operates in a very narrow space. Stripe now also controls how these businesses optimise their highest operating costs, i.e., compute and tokens. For any AI-first business, this is a very sticky position.
Another point to note: according to a CNBC investigation, 46% of US enterprise token usage on OpenRouter runs on Chinese-origin models. So, Stripe now also inherits this compliance exposure, along with the business.
What it means
Stripe’s acquisitions align with a very ambitious and coherent story. They quickly identify the infrastructure the next generation of businesses will depend on and own it before the rest of the market catches up. The payment companies that should pay closest attention to this are those whose customers are building AI products and whose definition of financial infrastructure stops at payments. Stripe has expanded the definition of financial infrastructure in a way that will change the future for AI-first companies.