Stripe has agreed to buy OpenRouter for more than $7 billion, according to Bloomberg, with TechCrunch, SiliconANGLE and Quartz all reporting the same figure over the weekend and into Monday. Stripe has not issued a press release and told TechCrunch it does not comment on speculation.
Eighty-three days ago, OpenRouter raised $113 million at a $1.3 billion valuation in a Series B led by CapitalG, Alphabet’s growth fund, with NVIDIA’s NVentures, Menlo Ventures and Andreessen Horowitz alongside. That is roughly 5.4 times the price in under three months, paid for a company whose product is, in the plainest terms, a switchboard.
OpenRouter sits between an application and more than 400 models from OpenAI, Anthropic, Google, Meta, Mistral and others, and routes each request to whichever one fits on cost, latency and capability. Developers write to one endpoint instead of eleven. The company has said it serves around 8 million users.
The multiple nobody agrees on
Here the reporting splits. PYMNTS framed the deal at roughly 70 times annual revenue. Independent analyses citing a run-rate near $140 million put it closer to 50 times. Both cannot be right, and the gap is the point: OpenRouter’s revenue has been compounding fast enough that a two-month-old figure produces a materially different headline than a current one.
What is not in dispute is the direction. Reported run-rate estimates trace a path from the high tens of millions at the start of 2026 to somewhere north of $100 million by mid-year. On any of those numbers, Stripe paid a multiple that only makes sense if the asset is strategic rather than financial.
Our take: Stripe didn’t buy a router. It bought a meter. Every request passing through OpenRouter generates a record of which model was chosen, at what price, for what job — the closest thing that exists to a real-time census of production AI adoption. Stripe’s entire franchise is built on sitting in the middle of transactions and knowing things nobody else knows. Same trade, applied to tokens instead of dollars.
Why a payments company wants the routing layer
Stripe was valued at $159 billion in a February employee tender backed by Thrive Capital, Coatue and Andreessen Horowitz — up 74% from the $91.5 billion mark a year earlier. At that size, $7 billion is about 4% of the company. A large cheque, not a bet-the-firm one.
The strategic logic: AI inference is becoming a metered, per-unit, billable business, and metered billing is Stripe’s home field. The company has already built an agentic commerce protocol with OpenAI so software agents can transact. Owning the layer that decides which model gets a query — and therefore who gets paid — puts Stripe upstream of that flow rather than downstream of it.
The risk sits in the same sentence. A neutral router is valuable precisely because it is neutral. Model providers accept its distribution because it has no horse in the race. A router owned by the company that also processes the payments has a horse — or at minimum has to keep proving it does not.
What to watch
- Confirmation and terms. Stripe has not announced this. Until it does, the $7 billion figure is reporting, not filing.
- Whether the labs stay. Any large model provider quietly de-prioritising OpenRouter distribution would tell you the neutrality question is real.
- The take rate. Routers compete on it. If Stripe raises it, the thesis was rents. If it holds or cuts, the thesis was data.
- Comparables. Every gateway startup just got a price tag to point at. Expect the next round of AI infrastructure raises to be marked off this deal.
The broader read: the market has spent 2026 arguing about which model wins. Stripe just paid $7 billion on the premise that the answer keeps changing — and that the durable business is owning the switch, not the thing being switched between.