[Notifications](/login?return_to=%2Fgetlago%2Flago)You must be signed in to change notification settings -
[Fork 719](/login?return_to=%2Fgetlago%2Flago)
[1 revision](/getlago/lago/wiki/With-OpenRouter,-is-Stripe-becoming-the-Amazon-of-AI/_history)
Stripe has reportedly agreed to buy OpenRouter for more than $7 billion. Stripe has not confirmed the deal. Bloomberg reported it as finalized, and TechCrunch reported the same figure.
The price looks absurd if OpenRouter is an API proxy. It makes more sense if Stripe sees the beginnings of Amazon Marketplace for AI inference.
OpenRouter gives developers one account and one API for more than 400 models. Behind it, more than 70 inference providers compete to serve requests. OpenRouter says it now has 8 million users.
The code that forwards a request is not worth $7 billion. The right to decide where a large and growing pool of requests goes might be.
OpenRouter is usually described as a gateway. That sounds technical and fairly neutral: a request arrives, the gateway forwards it, a response comes back.
But OpenRouter also decides which suppliers get access to demand.
Its default routing logic first removes providers that have recently failed. It then favors cheaper providers among the remaining candidates, using the inverse square of price. In OpenRouter's own example, an endpoint charging $1 per million tokens is nine times more likely to receive the first request than one charging $3.
That is good for developers. They get lower prices, fallback capacity and fewer provider integrations to maintain.
For inference providers, it is a distribution system. Price and performance determine how much traffic they receive. A provider can win more demand by cutting its price or improving reliability. It can lose demand without a developer ever making an explicit decision to switch. Amazon Marketplace did not become powerful because listing products online was difficult. It gathered buyers in one place, then controlled how merchants reached them. Search ranking and the Buy Box could matter as much as the seller's product.
OpenRouter is not there yet. But it already admits providers, measures their performance and allocates demand between them. As more developers accept the default route, its routing rules start to determine provider revenue.
OpenRouter charges customers a 5.5% fee when they buy credits and says it does not mark up the underlying provider's token price. The visible fee is simple. The more interesting asset is the demand sitting behind it.
Stripe already knows OpenRouter well. OpenRouter uses Stripe for payments, invoicing, tax and fraud prevention. In January, the two companies also made OpenRouter available through Stripe Projects, so a developer or coding agent could provision an account and receive an API key from Stripe's command line. Stripe described the partnership here, and OpenRouter described the Projects integration here.
Owning OpenRouter would move Stripe earlier in the transaction.
Today Stripe can see a customer payment. OpenRouter can see which model was requested, which provider served it and what that inference cost. Put the two together and Stripe can connect the cost of producing an AI feature with the revenue earned from selling it.
That is useful because AI products have unusually variable margins. Two similar customer actions can have very different costs depending on the model, context length, cache behavior, retries and provider selected. A company may pass that cost through, add a markup, spend down credits or absorb it inside a subscription.
Stripe can sell more than payment processing around that activity: metering, credits, billing, tax, fraud controls and provider settlement. OpenRouter gives it a place to distribute those products at the moment the cost is created.
Stripe has already been assembling that stack. It acquired Metronome for roughly $1 billion to go deeper into high-volume usage metering and complex pricing. I wrote about why Stripe bought it instead of rebuilding Stripe Billing. Metronome is now marketed as a Stripe product, but it still has its own application and product surface alongside Stripe Billing. Buying the missing layer was faster than rebuilding it. Making two products feel like one is slower.
The reported PayPal bid belongs in the background too. Reuters reported that Stripe and Advent offered more than $53 billion for PayPal. PayPal would bring merchants and consumers. OpenRouter brings developers and inference providers. Both are networks in which several parties need money moved, reconciled and monetized.
Stripe could build an AI router for far less than $7 billion. It could not quickly reproduce OpenRouter's provider relationships or persuade 8 million users to move production traffic through a new one. The premium is for those relationships and that demand, not the routing code.
The Amazon analogy has an obvious limit. Many merchants cannot afford to leave Amazon. OpenRouter users still have options.
Developers can pin a provider, set a maximum price, sort for latency or throughput, bring their own provider keys, move to another gateway or integrate directly. Large teams can run open-source routing software themselves. Providers can sell through several gateways and keep direct customer relationships.
As long as switching remains cheap, Stripe cannot squeeze either side very hard.
OpenRouter has never been perfectly neutral. Its defaults already encode a view of what a good route is. That is the product. Developers accept it because the rules are documented and usually serve their interests.
Stripe will put more of its products into OpenRouter. That is the point of buying it. The risk is not that OpenRouter becomes commercially motivated. It already is. The risk is that developers begin to suspect routing, pricing or product decisions are being made for Stripe before they are being made for them.
If that happens, they can still route around it. Stripe may be buying its way into the Amazon position. It has not bought Amazon's lock-in.