Deconstructing OpenRouter: Stripe’s $7.5 Billion Acquisition Gamble
拆解 OpenRouter：Stripe 75 亿美元收购赌局
Summary
This report examines Stripe’s proposed acquisition of OpenRouter and the nearly 6x repricing implied by a roughly $7.5 billion deal. It argues that OpenRouter has proven inference demand aggregation at scale, but must convert portable routing flow into non-portable model intelligence and an agent transaction control point.
Why it matters
As model capabilities converge and gateway standards make switching a one-line change, the same interoperability that makes multi-model routing valuable also erodes its moat. The deal is a test of whether routing, telemetry, identity, budgets and settlement can become one durable control layer for the agent economy.
Key ideas
Model substitutability is simultaneously OpenRouter’s growth tailwind and moat-destroying force: selection and dynamic routing matter more, while OpenAI-compatible gateways lower switching costs toward zero.
Multi-provider inference liquidity delivers measurable redundancy, price and throughput benefits, but providers multi-home across competing gateways; it is a market utility rather than an exclusive moat.
Stripe’s bet is a transition from a replaceable inference pipe to an agent transaction control point linking routing, telemetry, identity, budgets and settlement.
Versions
Chinese and English versions of this report.
Social threads
Published or syndicated by
Related research
2026.07 · Intelligence Infra