Summary of Key Points
OpenRouter, an American model routing company, was acquired by the giant Stripe for $7-8 billion just three years after its establishment, representing a 14-fold increase in valuation within that time. Its core value lies in acting as the “Stripe of the AI world” – providing an interface that connects to over 400 models (including many Chinese open-source models), helping developers automatically select the most cost-effective and stable options. The competitive pricing and rapid rise of Chinese open-source models have been key factors in OpenRouter’s explosive growth in traffic. However, despite contributing more than half of the platform’s token revenue, Chinese models have not received a corresponding return on investment, allowing OpenRouter and Stripe to reap the benefits of this platform.
1. How Did OpenRouter Become Worth $7 Billion in Three Years? – By Becoming the “Middleman of the AI Era”
OpenRouter’s business model is straightforward: if you are a developer who wants to use AI models but doesn’t want to deal with each of the over 400 providers such as OpenAI or DeepSeek individually, OpenRouter offers a unified interface that allows you to switch between models effortlessly and automatically choose the most affordable and responsive ones. It’s like ordering food without having to download the apps of every restaurant; you can use a single platform like Meituan to handle everything.
Its growth rate is astonishing: the company was valued at only $500 million last year, but its valuation soared to $1.3 billion in its Series B funding round, and then Stripe offered more than $7 billion for it just three months later. The reason behind this success is OpenRouter’s grasp of the “fragmentation” trend in the AI market. With a vast number of models available, each with its own strengths (some excel at coding, others at inference), and significant price differences, developers need a platform to compare and switch between them. OpenRouter charges a small fee for each model invocation, similar to a toll, and the more traffic it generates, the more it earns.
2. Why Did Stripe Acquire OpenRouter? – To Control the “Cashier and Shelf” of the AI Era
Stripe started as a payment processing company (helping businesses accept credit card payments) and now aims to become the leader in AI industry settlements. It has already acquired Metronome, which helps AI companies track their token usage and issue invoices, as well as the cross-border payment tool Bridge. However, it lacked the ability to help developers choose the right models for their needs. By acquiring OpenRouter, Stripe gains control over the entire process: from model selection and invocation to payment. This means Stripe not only manages the money but also determines which models developers use, effectively gaining control over the distribution of AI traffic. It’s like a supermarket that not only sells products but also decides which ones to display on the front shelves, giving it significant influence.
3. Chinese Open-Source Models: The Driving Force Behind OpenRouter’s Growth
Half of OpenRouter’s traffic growth can be attributed to Chinese models. Last year, Chinese models accounted for less than 2% of the platform’s token revenue; this year, that figure has exceeded 50%. Why?
1. Competitive Pricing: Chinese models are 60%-90% cheaper than their American counterparts. For example, DeepSeek V4 Flash costs only $0.14 per million tokens, while GPT-5.5 costs $5 – a difference that is irresistible for developers who rely on a large amount of tokens.
2. Aggressive Marketing by Chinese Model Providers: Chinese model providers use free or low-cost offers to attract overseas developers. For instance, Xiaomi’s Hunter Alpha was launched anonymously and quickly rose on the rankings due to its free usage, while companies like DeepSeek and Zhipu also offer free quotas for testing purposes. The more developers use these models, the higher their rankings, which in turn attracts even more users, creating a positive cycle.
As a result, OpenRouter’s total token volume has increased from 5 trillion per week last year to 25 trillion this year, with almost all of this growth coming from Chinese models.
4. The Embarrassing Reality Behind the Success: Chinese Models Focus on Traffic, Not Revenue
Despite contributing more than half of the platform’s tokens, Chinese models generate much less revenue due to their lower prices. For the same amount of tokens, Chinese models may only bring a fraction of the income that American models do. On OpenRouter, high-priced models like Google and OpenAI may have a smaller token share, but they still account for a large portion of the total revenue.
More importantly, Chinese model providers compete mainly for rankings and developer attention, while the platform’s underlying benefits (such as user habits and traffic channels) go to OpenRouter and then to Stripe. It’s like merchants engaging in price wars on the platform; while they attract users, the platform earns the majority of the profits.
Some question whether the high token volumes on the rankings are artificially inflated (e.g., through self-testing by providers or automated invocations by agents), but for OpenRouter, as long as there are models available and developers willing to use them, its value remains. The more competitive Chinese models become, the more valuable it becomes.
5. What This Acquisition Indicates: The Competition for Key Positions in the AI Market Has Just Begun
The acquisition of OpenRouter reflects a larger battle for control over key positions in the AI ecosystem. Stripe aims to become the central hub for AI settlements, and OpenRouter provides the entry point for model traffic, allowing it to control the connection between developers and model providers.
For Chinese models, this is a reminder that relying solely on low prices and open-source models to attract traffic is not enough; they need to find ways to convert traffic into revenue. For example, they could build their own brands or create platforms similar to OpenRouter to retain more of the traffic for themselves.
In summary, the competition in the AI market has shifted from the models themselves to the control of key ecosystem components. Those who can influence developer choices and payment processes will have a significant advantage in the future. The aggressive growth of Chinese models currently serves to benefit other companies rather than Chinese model providers themselves.
The most interesting aspect of this news is how it uses a single acquisition case to highlight the complex competitive dynamics in the AI industry, as well as the opportunities and challenges faced by Chinese models. Behind the surface-level excitement, there are deeper business strategies at play.