Doing the Hardest Work for the Least Money! Stripe's $7 Billion Acquisition Reveals the Truth About AI Profit Distribution

From OpenSea to OpenRouter, Alex Atallah has twice identified the pain points of fragmentation, and this time he sold to Stripe.

Written by: Thejaswini M A
Compiled by: Saoirse, Foresight News

In 2017, developer Alex Atallah believed that the crypto industry lacked a user-friendly front-end trading portal. He aimed to address this pain point and capture commercial value, leading to the founding of OpenSea. This platform aggregates trading of unique digital collectibles and was a darling during the NFT boom in 2020.

When the industry downturn hit, Alex Atallah did not wallow in disappointment. With the rise of open-source AI, he once again saw the fragmented problem he had previously tried to solve. The AI field was flooded with raw supply, and developers were overwhelmed by fragmented APIs and inconsistent billing systems. The time was ripe to create a unified front-end portal.

Thus, Alex Atallah founded OpenRouter, dubbing it the "Stripe of the AI field." Stripe provides developers with a unified API to handle all payment processes in one place; OpenRouter does exactly the same for AI models, which is the origin of its title as the "Stripe of the AI field."

I do not believe in the idea of wishful thinking, so I call it a self-fulfilling prophecy: Stripe has now acquired it for over $7 billion.

OpenRouter completed a $113 million Series B funding round in May this year, with a post-money valuation of $1.3 billion. Stripe's acquisition directly multiplied its valuation by five times.

This article will analyze what Stripe actually acquired and why real profits are generated at the API layer.

What Does OpenRouter Actually Do?

If a company wants to use AI models from OpenAI, Anthropic, and Google simultaneously, it needs to register accounts separately and write custom code for each. However, after integrating with OpenRouter, companies and developers only need to connect to one API to access all models at once. When an application makes a request, OpenRouter automatically forwards the request to one of the 500 supported AI models, selecting the one with the lowest cost, fastest response, best data privacy, or the currently available model.

The platform's business volume surged fivefold within six months, currently routing 250 trillion tokens weekly for 8 million users.

OpenRouter charges developers the raw fees for the models themselves, with a 5.5% fee for credit card payments and a 5% fee for cryptocurrency payments. Research firm Sacra estimates that by March 2026, the platform's total transaction volume of $900 million will correspond to an annual revenue of $50 million. Since then, business traffic has continued to rise, which is why Stripe paid a high premium of 50-140 times the sales multiple to acquire this business.

Stripe can easily afford this expense. The payment giant generated $6.8 billion in revenue last year, holds $3.2 billion in cash, and was valued at $159 billion in equity trading in February. According to its public annual report for 2025, the platform's total transaction volume reached $19 trillion, a 34% year-on-year increase, accounting for about 1.6% of global GDP. Patrick Collison and John Collison stated that Stripe serves over 5 million businesses, "covering all leading AI companies," specifically mentioning ChatGPT, Claude, Cursor, Replit, and Vercel. Stripe has not disclosed its overall revenue or broken down the revenue share from its AI business.

Stripe has completed its entire product layout, filling in all components needed for AI agents to pay autonomously.

At the infrastructure level, it spent $1.1 billion to acquire stablecoin infrastructure company Bridge. AI agents need funding storage and management solutions, so Stripe acquired wallet service provider Privy, with the transaction amount undisclosed.

Next is the usage billing platform Metronome, which is used to calculate computing power consumption. Giants like OpenAI, Anthropic, and Nvidia use it to account for resource consumption, and Stripe completed its acquisition in December 2025 for about $1 billion.

After that, Stripe, in collaboration with Paradigm, launched the payment public chain Tempo, which officially went live in March, along with a machine payment protocol that allows software to automatically complete service payments without human approval. Stripe also introduced the x402 protocol, enabling AI agents to purchase computing resources using cryptocurrency.

However, Stripe also hopes to see the flow of requests to determine which model will take on the tasks, and the OpenRouter acquisition conveniently solves this issue.

Since December of last year, Stripe CEO Patrick Collison has publicly stated that the AI industry is moving away from fixed monthly subscription models towards a pay-per-use billing model based on actual consumption. Stripe already had a backend usage tracking and billing system; this new transaction allows it to simultaneously control the front-end portal for developers accessing models.

Digital Transaction Settlement: A Dead End in Business

The crypto network can complete asset transfers and settlements, but where does the profit actually come from?

The underlying technology bears the heavy lifting and incurs huge costs, but it is the user-facing applications that can capture profits.

Verizon and AT&T have invested billions of dollars in laying pipelines and building 5G base stations to ensure data transmission, but they are essentially low-margin utilities. Profits are always generated at the API layer: for example, Apple sells smartphones that use the network, and Meta earns advertising revenue from applications within its platform.

Such cases are everywhere. Circle derives 94% of its revenue from investing customer deposits in U.S. Treasury bonds, even though its routing protocol handles 60% of cross-chain USDC transfers, the direct revenue generated is almost zero because the underlying blockchain for settlement belongs to a third party. Circle launched Arc (the mainnet will go live on September 16) precisely to migrate up the stack, relying on APIs to monetize and no longer lose transaction fees for free. Simply doing transfers and settlements is hard to scale unless a monopoly is formed. Visa and Mastercard also only handle data transfer settlements but can earn huge profits because they operate a closed system. Unlike public chains and the open internet, Visa belongs to a private network; merchants wanting to accept credit cards must use this infrastructure, paying a 2-3% fee for each transaction.

AI payment figures may seem large, but they lack real significance. The x402 protocol has processed over 153 million transactions since May 2025, with a total scale of $40 million, averaging just $0.26 per transfer. Even on low-cost public chains like Base and Solana, paying network fees for a $0.26 payment is not economically viable. In the Web3 space, millions of small transactions are often witch attacks—robotic small transfers to inflate data in hopes of obtaining future token airdrops or developers running stress test scripts in cycles.

In contrast, OpenRouter: annual transaction volume of $900 million, steadily extracting a 5% share. The underlying payment network does more work, but the actual value transferred is lower, only earning a few cents in meager profits.

Even dedicated public chains like Tempo launched without issuing a native token. This means it actively gives up the huge speculative profits usually associated with crypto networks. Making this trade-off is to provide a stable and predictable fee mechanism for enterprise clients.

Although Stripe's founders hope to build infrastructure capable of processing millions of transactions per second, Tempo's current actual utilization rate is extremely low. Since its launch in March, the network averages less than one real transaction per second. Stress tests have proven that the network can exceed 18,000 TPS, but the daily fees generated by real business are only $5.34, totaling about $620 in fees over thirty days. On-chain locked assets reach $40 million, but the economic returns from settlement business are minimal.

Data source: defillama

Only platforms that directly connect with customers can earn substantial profits. The original vision of the crypto industry was to aggregate AI models through crypto protocols, leveraging token discounts to achieve unbeatable low prices, but this path is heading towards failure. Crypto protocols can access cheap computing power, but it is centralized companies that control users and capture profits.

The decentralized computing network Chutes, built on Bittensor, has processed a cumulative 91 trillion tokens, covering 400,000 users. During the peak business period from May to June, Chutes output nearly 22 billion tokens daily through the OpenRouter interface. However, it relies on OpenRouter for as much as a quarter of its daily traffic. io.net outputs 4 billion tokens daily, also relying on this centralized platform for traffic distribution.

Data source: openrouter.ai

Decentralized networks like Chutes do not need to bear the expensive costs of data centers and corporate labor; computing power is provided by ordinary individuals, and participants earn crypto tokens as rewards, allowing them to quote much lower than large tech companies. The cost advantage is significant, but the shortcoming lies in trust, which is precisely the entire value of the API layer.

Quality control is a huge challenge for OpenRouter. When an application initiates a call, OpenRouter forwards the request to the lowest-cost available independent server. However, the quality of servers varies; some devices may downgrade hardware or alter model parameters to cut costs. A survey of 42 servers showed that nearly one-third completely hide their configuration parameters. To address this, OpenRouter plans to launch a "certification zone" for trusted service providers. For crypto networks, this issue is even more challenging: malicious nodes, driven by profit motives, may actively run downgraded versions of models to inflate profits.

Projects like DGrid and Bittensor are researching theoretical solutions to detect model output degradation, but large-scale real-time verification remains an unsolved problem.

Service availability is another significant real-world obstacle. Unlike large tech companies that have massive elastic server clusters, networks like Chutes are prone to crashes during traffic surges. Developers must still treat centralized aggregation platforms as a fallback solution to ensure stability, ultimately paying the toll fees they initially sought to avoid.

In terms of revenue, the project's publicly disclosed income diverges from actual market demand. Bittensor claims a subnet revenue of $43 million in the first quarter, but independent analysis verifies that the real external paid revenue is at most $15 million, with Chutes' actual revenue being less than $2.5 million. Chutes can receive newly minted crypto tokens worth $54,000 daily, relying on token inflation to maintain operations rather than a steady stream of real paying users.

DGrid has built a gateway compatible with OpenAI standards, connecting over 200 models, and completing settlements through x402, relying on a "quality proof" mechanism to audit outputs. The platform claims that in the first six months after launch, 13,000 users generated $20 million in revenue, paving the way for the issuance of the DGAI token. However, this revenue essentially comes from selling a lifetime membership priced at $1,580, allowing members to receive future token shares, which is a disguised token pre-sale and not continuous API call consumption.

Of course, decentralized networks can also handle real traffic, as evidenced by Dippy. This AI character application, with 8.6 million users, abandoned traditional cloud service providers and fully migrated to Bittensor's Targon network. Although on-chain financial income is difficult to verify completely, this is a significant breakthrough, proving that consumer-grade applications can run on decentralized infrastructure.

When platforms remain independent, they can fairly allocate traffic. Once they fall into the hands of companies like Stripe, conflicts of interest will emerge. Stripe holds its own payment tools, wallets, and blockchain products, naturally favoring its own system for traffic flow, squeezing out competitors.

The biggest selling point of the crypto narrative is the absence of centralized control. Decentralized systems cannot be acquired by companies and will not modify rules behind the scenes to favor their own applications. Stripe can develop excellent software, but it can never provide a truly neutral and unbiased underlying channel.

The business logic of the API layer is completely valid, but the crypto industry is still far from mastering the API layer.

Stripe's transaction effectively marks the public market price for businesses "standing in the middle of traffic." Reports indicate it paid $7 billion to acquire a business with an annual revenue of about $50 million, because this platform connects 8 million developers with about 500 AI models, making this business immensely valuable. If you are developing an open-source model routing project, you can now take this valuation data to approach venture capital.

However, the situation has also become awkward. In the past, decentralized computing networks relied on integrating with OpenRouter to grow without the hassle of customer acquisition. Now this traffic gateway belongs to Stripe, which has sufficient commercial motivation to prioritize its own product traffic. If you initially chose OpenRouter for its neutrality, hoping to call low-cost domestic models without interference, you may inevitably feel concerned.

Only routers that truly belong to no one will not deliberately favor traffic. But to achieve commercial viability, three major challenges must be overcome: proving the trustworthiness of the actual running code of nodes; achieving high availability without relying on centralized backups; and creating a business model that can survive without token subsidies. Currently, the industry has not achieved this.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.