Author: David, Deep Tide TechFlow

OpenClaw is on fire, but the company quietly making money in this wave of hype is one you may not have heard of:

OpenRouter.

To use OpenClaw, you need to access various AI models to get to work, with Claude, GPT, and DeepSeek each having their own pricing and interfaces. OpenRouter's job is to bundle these models together, allowing you to use them uniformly through it, and it profits from the price difference.

The person running this business is Alex Atallah. His company just secured a $40 million investment led by a16z, and is now valued at $500 million.

You may not know that his previous company was OpenSea, the largest NFT marketplace in the world, which at its peak was valued at over $13 billion.

However, he chose to exit during the peak of NFTs, and a few months later, the NFT market collapsed.

Now, he is making money in this wave of AI hype.

From Liquidity Aggregation to Large Model Aggregation

Alex Atallah graduated from Stanford's Computer Science department.

In 2018, he co-founded OpenSea with Devin Finzer. The business was simple: when others mint NFTs, they provide a place for people to buy and sell, taking a 2.5% cut from each transaction.

OpenSea does not produce NFTs, nor does it speculate on NFTs; it simply provides a shelf, aggregating liquidity.

When the NFT boom hit in 2021, popular NFTs like Bored Apes gradually became cultural symbols. OpenSea's peak monthly trading volume surpassed $5 billion, and Forbes estimated that he and Finzer had a combined net worth of $2.2 billion.

In July 2022, he resigned as CTO, stating he wanted to work on something new.

Everyone knows what happened afterward; NFTs crashed, the market entered a deep freeze, and OpenSea's own business was in disarray. However, someone always pays for the feast, and Alex left before the music stopped.

In 2023, he began working on something called OpenRouter. To put it simply:

A large model aggregation routing platform that puts the APIs of hundreds of models behind one interface; developers come in to call them, and he takes a 5% fee each time.

You might ask, why not just go directly to OpenAI or Anthropic, calling Claude or GPT?

Of course you can.

But now, nobody is likely to use just one model—using Claude to write code, searching for information using Gemini, delegating cost-saving tasks to DeepSeek—each requires separate registration, individual top-ups, and the interfaces differ...

Not to mention, many users want to use Claude and GPT, and there's no way to connect directly to the API from within the country.

Therefore, OpenRouter is the path of least resistance. One interface, over 500 models, unified format, automatic switching, one key to solve it all.

You might not have noticed while using OpenClaw, that the default provider in the configuration file was previously OpenRouter.

Image Source: Zhihu user Feng Control Alchemist

When you call Claude or DeepSeek, the requests go to him first, and then are sent to the model providers. Even OpenClaw's documentation states:

If the system does not recognize your API key format, it defaults to OpenRouter.

How fast is this business growing?

By October 2024, the amount of money processed through OpenRouter each month is $800,000. By May 2025, this number is expected to become $8 million.

In seven months, a tenfold increase.

By the end of the year, more than $100 million will pass through his hands. He takes 5% of that, making $5 million, and the team consists of fewer than ten people.

Image Source: sacra.com

a16z used his data to write an industry report titled “The AI Landscape of a Trillion Dollar Token”; Stripe has customized a billing system specifically for him.

As this year, OpenClaw exploded, more developers and enthusiasts have flooded in, experimenting with various methods to burn tokens, which inevitably requires connecting to various large models, thus completely igniting OpenRouter's business.

Moreover, a16z led the investment in this company, giving it a valuation of $500 million.

Once again, a person selling shovels has become a shovel seller.

The Hotspots Are Different, But the Model is the Same

If you look closely at Alex's two businesses, the structures are actually similar.

OpenSea's business does not involve minting NFTs but provides a place for others' minted NFTs to be bought and sold, taking a 2.5% cut from each transaction. OpenRouter's business does not train models but places others' trained models in one location, allowing developers to call them, taking a 5% cut.

This approach seems to have become his comfort zone, whether in NFTs or AI; the overall market structure characteristics are very similar:

The supply side is extremely fragmented, and buyers on the demand side do not know where to find the supply, while he stands in the middle as the shelf.

How fragmented was the NFT market in 2021? Dozens of chains, hundreds of project teams, tens of thousands of new series every day. If you wanted to buy a Bored Ape, you couldn't possibly visit each project's official website and sift through them. OpenSea aggregated them, allowing you to choose and buy, with sellers bidding to sell to you.

How fragmented will large models be in 2025? OpenAI, Anthropic, Google, Meta, DeepSeek, Mistral, myriad others... just among the mainstream, there are a dozen companies, plus hundreds in the open-source community.

Today, you might find Claude best for writing code; tomorrow, if Gemini releases a new version with stronger search capabilities; the next day, if DeepSeek cuts its prices by half, every switch requires changing the interface.

Atallah himself has stated a clear logic:

"OpenSea integrated a highly fragmented inventory into one place; today's AI scenario resembles that very closely."

He does not need to know which NFT will rise nor which model will win. He only needs to know one thing: the more fragmented the supply, the more valuable the middleman.

Interestingly, there are aspects related to timing.

When he left in July 2022, OpenSea's valuation was still high; although monthly trading volumes of NFTs had declined from their peak, no one suspected a collapse was imminent. He expressed a desire to "create something new from scratch"; six months later, ChatGPT was released, marking the beginning of the large model era.

Did he see something or just get lucky?

I don't know. But one thing is certain:

When he registered OpenRouter in early 2023, there were virtually no AI large model routing products on the market. By the time everyone realized a unified interface was needed, he was already there.

Last time, he did the same thing in the NFT space. By the time everyone squeezed in, he was already the largest platform.

Does It Matter Whether AI is Hot?

In every wave of hype, the most common question people ask is: What will be hot?

In 2021, it was which NFT would rise; in 2024, which meme coin would increase a hundredfold; in 2025, which AI application would emerge; in 2026, what could be done with crayfish.

Atallah's question might be different. I think his thought process should be: regardless of what is hot, where will the money flow?

These two questions may seem similar, but they represent completely different betting strategies.

Betting on "what will be hot" requires a correct guess once. Bored Apes will rise, PEPE will increase a hundredfold, a certain AI product will be the next ChatGPT. Guess right, and you will become wealthy; guess wrong, and you go to zero. Most people's experience falls into the latter category.

Betting on "where the money will flow" means you don't need to guess correctly at all. NFTs rise, trading happens on OpenSea, and he earns transaction fees. The more fiercely the AI model wars rage, the more developers need a unified interface for switching back and forth, and OpenRouter becomes busier.

Not betting on who will win but betting that this battle will last a long time.

In retrospect, in every cycle, the players making the most money, regardless of the industry, are almost always those platforms.

Gold miners come and go, while the water sellers consistently collect money.

But I think merely saying "selling water" or "selling shovels" isn’t enough. The shovel sellers also have many that go out of business, while Atallah has done something more specific: he consistently positions himself at the aggregation point.

It is not enough to simply create a tool to collect road fees. You have to be the one bringing together scattered supplies. The more fragmented the supply, the higher the switching costs, and the more pricing power the aggregation layer has.

This also explains why he entered both markets in their early stages. Because aggregation businesses have a characteristic:

The early entrants sign up all the supply, making it hard for later entrants to catch up.

Thus, Atallah's key insights can be summed up in two sentences:

First, don’t guess who will win; just find the intersection everyone needs to pass through. Second, when others have not yet realized the need for the intersection, you have already built the road.

A Capable Person Never Chooses a Table

Right now, I feel two voices around me are particularly loud.

One says AI Agents are toys, and using OpenClaw has no use besides burning tokens; the other says this is just another wave of AI hype, and in three months, no one will remember.

Perhaps both viewpoints are correct.

But for people like Alex Atallah, it does not really matter.

Whether OpenClaw is useful or not, he is still collecting money. You may think the crayfish are boring and uninstall the app, but the tokens burned in the past two weeks have already passed through his hands.

Some people think NFTs are dirty, are a Ponzi scheme, are a scam. He created a company valued at $13.3 billion from it. Some think AI Agents are a bubble, just speculation, and fail to see the business model. He built a company valued at $500 million from it...

Capable people might not really need us to respect the fields they operate in.

He made money at the NFT table. He has earned again at the AI table. What the next table will hold is unknown to anyone.

But I guess he will still be at the entrance collecting tickets.