Merit AC
2026-08-19

Stripe is buying the switchboard that routes AI apps between models

OpenRouter built the layer that lets a company swap GPT for Claude for Gemini without rewriting its code. Stripe is acquiring it for a reported $7 billion-plus, betting that deciding which model handles a request is now a payments problem.

Stripe announced on August 19, 2026 that it has agreed to acquire OpenRouter, the startup that routes API requests across more than 400 models from over 80 providers so a developer doesn't have to hard-code a bet on any single one. Neither company disclosed a price; TechCrunch and other outlets had reported two days earlier that the deal was worth $7 billion or more, up roughly fivefold from the $1.3 billion valuation OpenRouter reached in a $113 million round only months earlier.

Why a payments company wants to own model routing

Stripe's own reasoning, per CEO Patrick Collison: "Tokens are the central currency for companies building with AI." That's a bigger claim than it sounds. OpenRouter doesn't just route traffic -- it sits at the exact point where a company decides what it's actually going to spend per request, weighing task complexity against price, speed, and reliability across competing models. Stripe already processes the resulting invoice through products like Token Billing; owning the routing layer means it now also influences the decision that generates that invoice in the first place. For any company trying to answer "what are we actually spending on AI, and is it buying anything," that's the same question this site's flagship tracker exists to answer -- just answered, in this case, by the vendor that gets paid either way.

OpenRouter's own post is explicit that the product isn't changing: "OpenRouter will continue to operate as it is: same mission, same name, same product, same roadmap." The company says it will keep operating independently under Stripe, expanding its roughly 90-person team while trying to hold onto its model-neutral positioning -- a claim worth revisiting once the acquisition actually closes, since "we won't play favorites" is precisely the kind of promise that gets tested the first time a parent company's own priorities point toward one model provider over another.

Sources

← All news