Original author: E.H.
Original translation: Deep Tide TechFlow
We analyzed all x402 transaction data from December 2025:
- 63 million payments
- 7.5 million USD in USDC
- Over 1,100 projects participating
This is the first real-world case proving that AI agents can provide payment services at scale.
Here are our findings
http://blockrun.ai/state-x402-2025.pdf
http://blockrun.ai/state-x402-2025-deck.pdf
A Quick Review of History:
HTTP 402 "Payment Required" status code was defined in 1997 but has never been practically applied.
The reason is simple: processing a $0.01 payment with a credit card incurs a fee of $0.30, resulting in a fee of up to 3000%. Micropayments were doomed to struggle from the start.
Until the emergence of stablecoins, which changed everything.
Overview of December 2025 Data:
- 63 million transactions
- 7.5 million USD in USDC circulation
- 64,000 independent buyers
- 10,000 independent sellers
- Average transaction amount of only $0.12
In comparison: Visa's minimum viable transaction amount is about $15, while x402 is processing payments nearly 100 times smaller and achieving scalable operations.
The ecosystem is experiencing explosive growth:
- Over 1,100 independent projects
- Over 4,800 mainnet endpoints
- 6 active payment coordinators
Most service providers are small, with no obvious dominant players.
Translation: Now is the critical moment for "seizing territory."
76% of service fees are at $0.10 or below
Best range? $0.01-$0.10, accounting for 69% of all services.
This is the realm of micropayments—an area where credit cards cannot compete at all.
AI agents pursue efficiency optimization; they pay on demand, always charging per use.
So what are people actually building?
🥇 Data services: 31%
🥈 AI/Large Language Models (LLM): 25%
🥉 Blockchain: 15%
Not meme projects, nor speculative ventures.
But solid infrastructure: APIs, analytics, inference services.
"Real money" has already entered the scene.
Base vs Solana—who is superior?
- Base: 53% (the main stronghold for developers)
- Solana: 37% (the preferred choice for production environments)
Which one to choose? It depends on your use case.
"Isn't this all just fake trading volume?"
Let's be honest:
Artemis data shows that 47% of the trading volume is for ranking manipulation.
But—this only accounts for 14% of the total transaction amount.
Small transactions = cleaner cash flow.
The feasibility of micropayments has been validated.
The billion-dollar question yet to be solved:
What happens when AI agents make payment errors?
x402 has already solved the question of "how AI agents pay."
But it still hasn't resolved:
- Agent verification (KYA, Know Your Agent)
- Reputation systems
- Dispute resolution mechanisms
Whoever can solve these problems will become the ultimate winner.
- 63 million transactions in a single month
- Average transaction amount: $0.12
The protocol is working.
The economic model is viable.
Developers are already in place.
The only remaining question is: What will you build on this foundation?
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
