Original link: https://www.techflowpost.com/zh-CN/article/30673
Disclaimer: This article is a reprint. Readers can obtain more information through the original link. If the author has any objection to the reprint format, please contact us and we will modify it according to the author's requirements. Reprinting is for information sharing only and does not constitute any investment advice, nor does it represent Wu Blockchain views or positions. OpenClaw has become a hit, but the company that has quietly made money in this wave is one you may not have heard of:
OpenRouter.
To use OpenClaw, you need to connect to various AI models to get things working. Claude, GPT, and DeepSeek each have their own pricing and APIs. What OpenRouter does is package these models together, allowing you to use them uniformly, and it earns the difference in price.
The person running this business is Alex Atallah. His company just received $40 million in funding led by a16z and is now valued at $500 million.
What you may not know is that his last company was OpenSea, the world's largest NFT exchange, which was valued at over $13 billion at its peak.
However, he chose to leave the market at the height of NFT's popularity, and a few months later, the NFT market collapsed.
Now, he's made money again in the AI craze.
From liquidity aggregation to large-scale model aggregation
Alex Atallah graduated from the Computer Science Department of Stanford University.
In 2018, he co-founded OpenSea with Devin Finzer. Their business is simple: others mint NFTs, and they provide a platform for buying and selling them, taking a 2.5% cut from each transaction.
OpenSea does not produce or speculate on NFTs; it only provides a marketplace and aggregates liquidity.
The NFT craze emerged in 2021, with hit NFTs like BAYC gradually becoming popular symbols. OpenSea's monthly transaction volume peaked at over $5 billion at that time, and Forbes estimated that he and Finzer's combined net worth was $2.2 billion.
In July 2022, he resigned as CTO, saying he wanted to do something new.
As we all know, NFTs collapsed, the market entered a complete freeze, and OpenSea's own business was in shambles. However, there's always someone to foot the bill for a good feast, and Alex left before the music ended.
In 2023, he started working on something called OpenRouter. To put it simply:
The large-scale model aggregation routing platform puts the APIs of hundreds of models behind a single interface, which developers can call, and charges a 5% fee each time.
You might ask, why not just look for OpenAI, Anthropic, Claude, or GPT to tune the model?
sure.
But nowadays, probably no one uses only one model. They use Claude for coding, Gemini for research, and DeepSeek handles the cheaper tasks. Each platform has its own registration, payment system, and API format...
Not to mention that many users want to use Claude and GPT, but there is no way to directly connect to the API from within China.
Therefore, OpenRouter is the path of least resistance. One interface, over 500 models, unified format, automatic switching, and everything is handled with a single key.
You might not have noticed when using OpenClaw that the default provider (API provider) in the configuration file was previously OpenRouter.
Image source: Zhihu user Feng, the alchemist.
When you call Claude or DeepSeek, the request goes to them first, then to the model vendor. Even OpenClaw's documentation states this:
If the system does not recognize your API key format, it will default to using OpenRouter.
How fast is this business growing?
In October 2024, the monthly revenue flowing through OpenRouter was $800,000. By May 2025, that figure had grown to $8 million.
Seven months, tenfold.
Over the course of a year, he handled more than $100 million. He took 5% of it, netting $5 million, while his team consisted of fewer than ten people.
Image source: sacra.com
a16z used his data to write an industry report called "The Current State of AI in the 100 Trillion Token Market"; Stripe specially customized a billing system for him.
With the explosive popularity of OpenClaw this year, more developers and enthusiasts have flocked in, burning tokens in various creative ways, which inevitably requires calling various large models, which has also completely boosted the business of OpenRouter.
Moreover, a16z led the investment in this company, valuing it at 500 million.
A shovel seller has become a shovel seller once again.
Different hot topics, same pattern
If you look closely at Alex's two businesses, you'll see that their structures are actually the same.
OpenSea's business model involves not creating NFTs themselves, but rather placing other people's created NFTs in one place, where buyers and sellers can trade, and they take a 2.5% cut. OpenRouter's business model involves not training models themselves, but rather placing other people's trained models in one place, where developers can use them, and they take a 5% cut.
This approach seems to have become his comfort zone, as the market structure for both NFTs and AI is very similar:
The supply side is extremely fragmented, and buyers on the demand side don't know where to find supply, while he stands in the middle as a shelf.
How fragmented were NFTs in 2021? Dozens of chains, hundreds of projects, and tens of thousands of new series every day. If you wanted to buy a BAYC, you couldn't possibly browse through every single project's website. OpenSea brings them all together; you choose and buy, and competitors offer to sell to you.
How fragmented will large-scale models be in 2025? OpenAI, Anthropic, Google, Meta, DeepSeek, Mistral, Zero1Wu... There are more than a dozen mainstream companies alone, plus hundreds of open-source communities.
You might be best off using Claude for coding today, but tomorrow Gemini releases a new version with even better search capabilities, and the day after that DeepSeek drops in price by half. Every time you switch, you have to modify the API.
Atallah himself once said something that made this logic very clear:
"OpenSea consolidates highly fragmented inventory into one place, and AI today is very similar to that."
He doesn't need to know which NFT will rise in value, nor does he need to know which model will win. He only needs to know one thing: the more dispersed the supply, the more valuable the middlemen become.
What's interesting is the timing.
When he left in July 2022, OpenSea's valuation was still high, and although the monthly transaction volume of NFTs had fallen from its peak, no one thought it was going to collapse. He said he wanted to "do something new from scratch," and six months later ChatGPT was released, marking the beginning of the era of large-scale models.
Did he see something, or was he just lucky?
I don't know. But one thing is certain:
When he registered OpenRouter in early 2023, there were almost no AI-powered large-scale routing products on the market. By the time everyone realized the need for a unified interface, he was already there.
Last time, he did the same thing in the NFT space. By the time everyone else had jumped in, he was already the biggest platform.
Does it matter whether AI is the one that's trending?
In every wave of hype, the question most people ask is: What will become popular?
Which NFT will rise in value in 2021? Which meme coin will increase 100 times in 2024? Which AI application will emerge in 2025? What can crayfish do in 2026?
Atallah's question might be different. I think his thought process is: no matter what becomes popular, where will the money go?
These two questions may seem similar, but they are actually completely different ways of betting.
Betting on "what will be popular" requires you to guess correctly at least once. BAYC will surge, PEPE will multiply a hundredfold, a certain AI product will be the next ChatGPT. Guess right and you'll be rich; guess wrong and you'll lose everything. Most people's experience is the latter.
When betting on "where the money goes," you don't need to guess any of them correctly. NFT prices rise, transactions occur on OpenSea, and they collect fees. The fiercer the AI model wars become, the more developers need a unified interface to switch between them, and the busier OpenRouter gets.
I'm not betting on who will win, I'm betting this battle will last a long time.
In retrospect, the platforms that made the most money in each cycle, regardless of the industry, were basically those in this position.
Gold prospectors came and went, while water sellers continued to collect money.
But I don't think just saying "selling water" or "selling shovels" is enough. A lot of shovel sellers die, but Atallah did something more specific: he always got the water right at the right spot.
You can't just build any tool and start collecting tolls. You have to be the one who brings together the scattered supply. The more dispersed the supply, the higher the switching costs, and the more pricing power the middle aggregator has.
This also explains why he entered the market at the very earliest stage both times. Because aggregation businesses have one key characteristic:
Once the first people arrive have signed off on the supplies, it becomes very difficult for those who arrive later to catch up.
So, I can summarize Atallah's exceptional qualities in two sentences:
First, don't guess who will win; just find intersections that everyone needs to pass through. Second, build the roads before others even realize they need them.
A genius never picks and chooses a table.
Right now, I feel like there are two very loud voices around me.
One view is that the AI Agent is a toy, and installing OpenClaw is useless except for burning tokens; another view is that this is just another AI hype, and no one will remember it in three months.
Both of these viewpoints may be correct.
But for someone like Alex Atallah, it doesn't really matter.
Whether OpenClaw is useful or not, he's still making money. If you decide today that lobster is boring and uninstall it, the tokens you burned over the past two weeks have already passed through his hands.
Some people think NFTs are dirty, a Ponzi scheme, a scam. Yet one person built a company on it with a valuation of $13.3 billion. Others think AI agents are a bubble, hype, and they can't see a viable business model. Yet another person built a company on it with a valuation of $500 million...
The truly talented may not need our respect for the field they're in.
He made money on the NFT table. He made money on the AI table. Nobody knows what will be on the next table.
But I guess he'll still be collecting tickets at the door.
