When AI Starts Paying for Itself

marsbitPublished on 2026-03-04Last updated on 2026-03-04

Abstract

The article "When AI Starts Paying for Itself" discusses the emergence of the x402 protocol, which enables AI agents to autonomously make micro-payments for services like data and computation. In 2025, Coinbase and Cloudflare revived the long-dormant HTTP 402 status code ("Payment Required") to create a seamless payment layer for the internet. The protocol allows an AI agent to receive a payment request, authorize it with a cryptographic signature, and complete the transaction in under a second—with no human involvement, accounts, or traditional banking infrastructure. Supported by low-cost Layer 2 blockchains and stablecoins, x402 processed over 100 million transactions within months. Its V2 update added multi-chain support and session-based authentication. Google later integrated a similar model into its Agentic Payments Protocol (AP2). However, trust between autonomous agents remains a challenge. ERC-8004, an Ethereum standard, addresses this by providing on-chain identity (via NFT-based IDs), reputation tracking, and task verification systems. The ecosystem faces risks: speculative "x402-themed" meme tokens have surged without real utility, technical vulnerabilities exist, and competing standards from Google and a16z threaten fragmentation. Furthermore, regulatory frameworks for AI-driven transactions are undeveloped. In summary, x402 and ERC-8004 aim to create a trustless, open economic network for AI agents—but must overcome technical, economic, and competitive h...

Author: Clow, Baihua Blockchain

An AI wrote a piece of code and needed data to verify it.

It sent an HTTP request, and the server returned a number: 402.

Then, it paid 0.001 USDC. In less than a second, the data came back.

This transaction had no account, no password, no bank card, no KYC. The entire process involved no human participation.

This is no longer science fiction. By the end of 2025, the x402 protocol had processed over 100 million such transactions; in the first 30 days of this year, another 15 million were added.

The "Payment Code" That Slept for Thirty Years

In 1990, the drafters of the HTTP protocol reserved a specific status code: 402, Payment Required.

The meaning is straightforward—"Payment is required to enter here."

But this status code was never officially used. It just hung there in the protocol documentation, dormant for thirty-four years.

The reason is simple: the first builders of the internet never imagined that machines would one day be the ones paying. Credit cards, bank accounts, KYC verification—these payment infrastructures were designed for humans and are completely ineffective against autonomously running code.

AI Agents need to call APIs, purchase data, and acquire computing power at millisecond speeds. The account registration and fee thresholds of traditional payments are a solid wall.

In 2025, three conditions aligned simultaneously.

The total supply of stablecoins surpassed $300 billion, with L2s like Base reducing single transaction costs to sub-cent levels; the AI Agent ecosystem, led by OpenAI and Anthropic, began large-scale commercialization; Coinbase engineers dug up that dusty 402 and decided to activate it.

In May 2025, Coinbase, jointly with Cloudflare, officially released the x402 protocol. In September, they collaborated with Cloudflare to jointly announce the establishment of the x402 Foundation. A forgotten status code returned to the center of the internet. Cloudflare manages roughly 20% of global web traffic—this meant x402 had no shortage of infrastructure entry points from day one.

Machines, For the First Time, Learned to Pay

The design of x402 is surprisingly simple.

An AI Agent initiates an HTTP request, the server returns a 402 status code along with the payment requirements: how much, which chain, which token. The Agent authorizes the payment using an EIP-712 encrypted signature, stuffs the payment information into the request header, and resends the request. The server verifies it and returns the resource.

The entire process takes less than a second. No account, no subscription, no API Key.

This turns "payment" into part of the internet. Like GET and POST, it's just an HTTP action. Any service can charge machines by adding a single line of middleware.

Data proves this logic works. About seven months after the protocol's release, it processed over 100 million transactions. According to the Cambrian Network Q1 2026 report, over 15 million transactions occurred in the past 30 days, from over 400,000 buyers and over 80,000 sellers. The AI Agent群体 from Virtuals Protocol is the largest single source of transactions; they use the protocol to autonomously settle collaboration fees among Agents.

On December 11, 2025, x402 V2 launched. This upgrade pushed the protocol from "usable" to "user-friendly": support for multiple chains like Base, Solana, and Avalanche operating in parallel; introduction of a Session mechanism (wallet becomes an identity credential, allowing repeated access without on-chain interaction each time); integration with ACH bank transfers and credit card networks—the payment systems of Web2 and Web3 were connected for the first time within this protocol.

Google subsequently integrated x402 into its Agent2Agent (A2A) protocol, releasing the Agentic Payments Protocol (AP2); machines paying is becoming an infrastructure consensus among large tech companies.

Trust is the First Hurdle of the Agent Economy

The payment problem is solved, but an even more fundamental question remains unanswered.

"Commerce can’t happen if people don't trust each other."

Davide Crapis, AI Lead at the Ethereum Foundation and one of the co-drafters of ERC-8004, directly pointed out the core obstacle of the Agent economy: When an AI Agent needs to hire another Agent to complete a task, how does it know the other party isn't a scammer? Where is the transaction record? How is reputation transferred?

ERC-8004 is Ethereum's answer to this question. Drafted in August 2025, it officially launched on the Ethereum mainnet on January 29, 2026. It establishes three lightweight registries on-chain:

  • Identity Registry: Each Agent gets an on-chain ID based on an ERC-721 NFT, portable, transferable, and verifiable across chains. An AgentCard (JSON format) records capabilities, endpoints, and x402 payment support status;
  • Reputation Registry: Feedback signals between Agents are archived on-chain—accuracy, timeliness, reliability scores. Only indices are kept on-chain, with data hashes pointing to off-chain storage to reduce Gas costs;
  • Verification Registry: After task completion, the result data hash is put on-chain requesting verification, giving "whether the task was truly completed" cryptographic provability.

The drafting team spanned four major crypto ecosystems: Marco De Rossi from MetaMask, Davide Crapis from the Ethereum Foundation, Jordan Ellis from Google, and Erik Reppel from Coinbase. EigenLayer, ENS, The Graph, and Taiko have all expressed support. Less than a month after the mainnet launch, registered Agents on Ethereum mainnet surpassed 24,000, with about 49,000 across all EVM chains combined.

A typical workflow looks like this: Agent A discovers a service provider through the ERC-8004 identity registry, filters the reputation registry to choose a high-scoring Agent B, completes the payment using x402, and attaches payment records with reputation feedback after the task ends—payment history becomes an anchor of credit. This chain is what Cambrian Network calls the "Agent Economy Operating System": payment + identity + reputation, three layers combined.

How Deep Are These Waters?

The data looks good, but a few things need clarification.

Tokens and protocols are two different things.

The ecosystem token related to x402 reached a market cap of over $9 billion on CoinGecko, with daily trading volume exceeding $230 million. But many of these "x402 concept tokens" are Meme tokens with no substantive binding to the protocol itself. Buying an x402-related Token is not the same as buying into the growth of the x402 protocol. This market has always been good at mixing narrative and reality, and this time is no exception.

Technical risks remain unresolved.

x402's EIP-712 signature mechanism requires continuous security audits. The ERC-8004 reputation registry faces the threat of Sybil attacks—batch registration of fake identities—and the current economic incentive mechanisms are not yet perfect. There is an economic tension between high-frequency micropayments (e.g., $0.0001 per transaction) and L2 fees (which can still reach $0.05); extremely small transactions can still be eaten up by fees at this stage.

The protocol war is not over.

x402, Google's AP2, and a16z's ecosystem ACP, three sets of protocols coexist. If developers split between these three standards, network effects will be significantly diluted. Furthermore, OpenAI and Anthropic could completely bypass on-chain protocols and build their own closed-loop payment systems—they have users, data, and scale advantages, which is competitive pressure x402 cannot ignore.

The regulatory landscape is another blank slate. For a payment autonomously initiated by an AI Agent, who is the transaction主体? Where is the trigger point for KYC/AML responsibility? No major jurisdiction has provided a definitive answer.

Summary

Someone once said something fitting to quote here: "Inscriptions in 2023 allowed humans to engrave value onto the chain; x402 in 2025 allowed machines to autonomously pay for value on the network for the first time."

If HTTP connected global computers into an information network, the combination of x402 and ERC-8004 attempts to connect billions of Agents into an open service and data market—no account needed, no approval needed, just one request, one payment, one result.

However, whether the protocol can prevail in fragmented competition, whether the trust mechanism can be truly perfected, whether the Agent economy can move from demo to real business... these are still open questions.

Before the narrative lands, distinguishing between "the value of the protocol" and "the tokens炒作 around the protocol" is perhaps the most important thing for every participant to figure out.

Related Questions

QWhat is the x402 protocol and what problem does it solve?

AThe x402 protocol is a system that enables AI agents to autonomously make micropayments for services like data access or computational resources via HTTP requests. It solves the problem of traditional payment systems (requiring accounts, passwords, credit cards, KYC) being incompatible with machine-to-machine transactions by integrating payment as a native HTTP action, allowing for seamless, account-less transactions in under a second.

QWhat historical internet standard did the x402 protocol revive and why was it unused for decades?

AThe x402 protocol revived the HTTP 402 status code, 'Payment Required,' which was defined in 1990 but never officially used. It remained dormant for over three decades because the early internet was built for human users, and the existing payment infrastructure (credit cards, bank accounts, KYC) was not designed for autonomous machines or AI agents.

QHow does the ERC-8004 standard complement the x402 protocol in the AI agent economy?

AERC-8004 establishes a trust layer for the AI agent economy by providing on-chain registries for agent identity (via NFT-based IDs), reputation (storing performance metrics and feedback), and task verification (cryptographic proof of completion). It complements x402 by enabling agents to discover reliable service providers, verify their credibility, and use x402 for payment, creating a combined system for payment, identity, and reputation.

QWhat are some challenges and risks associated with the x402 protocol and ERC-8004?

AChallenges include technical risks like the need for ongoing security audits of EIP-712 signatures, susceptibility to Sybil attacks on reputation systems, and the economic tension between micropayment values (e.g., $0.0001) and layer-2 transaction fees (e.g., $0.05). Additionally, there is protocol competition (e.g., from Google's AP2 and a16z's ACP), potential fragmentation among developers, and unresolved regulatory questions regarding KYC/AML responsibilities for autonomous AI transactions.

QWhat real-world adoption and transaction volume has the x402 protocol achieved according to the article?

ABy the end of 2025, the x402 protocol processed over 100 million transactions. In the first 30 days of 2026, it handled an additional 15 million transactions, with over 400,000 buyers and 80,000 sellers. Virtuals Protocol's AI agents were the largest single source of transactions, using x402 for autonomous settlement of collaboration fees among agents.

Related Reads

Why Do You Always Lose Money on Polymarket? Because You're Betting on News, While the Pros Read the Rules

Why do you always lose money on Polymarket? Because you bet on news, while the pros study the rules. This article explains how top traders ("che tou") profit by meticulously analyzing market rules, not just predicting events. Polymarket, a prediction market platform, often sees disputes over event outcomes due to ambiguous rule wording. For instance, a market asking "Who will be the leader of Venezuela by the end of 2026?" was misinterpreted by many who bet on Delcy Rodríguez, assuming she held power. However, the rules specified "officially holds" as the formally appointed, sworn-in individual. Since Nicolás Maduro was still recognized as president officially, he won the market—even being in prison. To resolve such disputes, Polymarket uses a decentralized arbitration system via UMA protocol. The process involves: 1. Proposal: Anyone can propose a market outcome by staking 750 USDC, earning 5 USDC if unchallenged. 2. Dispute: A 2-hour window allows challenges with a 750 USDC stake; successful challengers earn 250 USDC. 3. Discussion: A 48-hour period on UMA Discord for evidence and debate. 4. Voting: UMA token holders vote in two 24-hour phases (blind then public). Outcomes require >65% consensus and 5M tokens voted; otherwise, four re-votes occur before Polymarket intervention. 5. Settlement: Results are final and automatic. Unlike traditional courts, Polymarket’s system lacks separation between arbitrators and stakeholders—voters often hold market positions, creating conflicts of interest. This leads to herd mentality in discussions and non-transparent outcomes without explanatory rulings, preventing precedent formation. Thus, success on Polymarket hinges on deep rule interpretation, not just event prediction, exploiting gaps between reality and contractual wording.

marsbit1h ago

Why Do You Always Lose Money on Polymarket? Because You're Betting on News, While the Pros Read the Rules

marsbit1h ago

DeepSeek Funding: Liang Wenfeng's 'Realist' Pivot

DeepSeek, a leading Chinese AI company, has initiated its first external funding round, aiming to raise at least $300 million at a valuation of no less than $10 billion. This move marks a significant shift from its founder Liang Wenfeng’s previous idealistic stance of rejecting external capital to maintain independence. Despite strong financial backing from its parent company, quantitative trading firm幻方量化 (Huanfang Quant), which provided an estimated $700 million in revenue in 2025 alone, DeepSeek faces mounting challenges. Key issues include a 15-month gap in major model updates, delays in its flagship V4 release, and the loss of several core researchers to competitors offering significantly higher compensation. The company is also undergoing a strategic pivot by migrating its infrastructure from NVIDIA’s CUDA to Huawei’s Ascend platform, a move aligned with China’s push for technological self-reliance amid U.S. export controls. However, DeepSeek lags behind rivals like智谱AI and MiniMax—both now publicly listed—in areas such as product ecosystem, multimodal capabilities, and commercialization. The funding round, though relatively small in scale, is seen as a way to establish a market-validated valuation anchor, making employee stock options more competitive and facilitating talent retention. It also signals DeepSeek’s transition from a pure research-oriented organization to a commercially-driven player in the global AI ecosystem.

marsbit1h ago

DeepSeek Funding: Liang Wenfeng's 'Realist' Pivot

marsbit1h ago

Trading

Spot
Futures

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片