ERC-8004 gathers over 20,000 "digital immigrants," turning the blockchain into a testing ground for AI Agents

Author: Nancy, PANews

Once, AI Agents were just supporting characters in science fiction; now, they are preparing to take over the real world. From simple chatbots in the beginning to intelligent agents capable of autonomous decision-making, cross-platform tool invocation, and complex task completion, the AI Agent economy is on the brink of explosion.

Currently, Ethereum, which urgently needs fresh blood, is gathering a special group of “digital immigrants.” With the launch of the ERC-8004 protocol, Ethereum is becoming one of the core experimental grounds for AI Agents.

Over 20,000 Agents on Chain, Ethereum and Base as Main Testing Grounds

At the end of last month, ERC-8004 officially went live on the Ethereum mainnet, with plans to deploy as singleton instances across all major Layer 2 networks within the coming weeks.

This marks an important step in the evolution of the Ethereum ecosystem toward AI-native infrastructure. ERC-8004 introduces discovery mechanisms and a portable reputation system, enabling AI agents to interact across organizations and carry reputation records across different platforms, providing a foundational protocol for a global AI service interoperability market without centralized gatekeepers.

According to data from 8004scan, as of February 13, the number of Agents deployed based on ERC-8004 has approached 21,000.

Looking at deployment distribution, ERC-8004 has covered 16 networks including Ethereum, Polygon, BNB Chain, Base, Monad, Arbitrum, Celo, and plans to add Plasma, Metis, and Soneium.

Ethereum remains the core hub, with over 11,000 Agents, accounting for more than half; the rest are mainly distributed across Base, Gnosis, and BNB Chain, each with several thousand.

However, in terms of deployment pace, initial Agents were mainly concentrated on Ethereum mainnet. As scale expanded, new deployments gradually shifted toward Base. Overall, most Agents on ERC-8004 are locked between Ethereum and Base ecosystems. Mature development environments, liquidity depth, and user bases are likely key factors attracting developers.

Regarding participants, among these Agents, prolific teams have begun to emerge, with it being common for a single address to operate multiple Agents.

In application types, the current ERC-8004 ecosystem features diverse players, including technical Agents focused on DeFi infrastructure and on-chain tools, market analysis and investment assistants, content generation and creative applications, and some Agents have even issued tokens to explore independent economic models.

However, in terms of actual interactions, the ecosystem is still in early stages. To date, feedback has totaled about 15,000 entries, averaging less than one per Agent. This indicates that most Agents are still in cold start, with limited actual use and user interaction. Notably, 73.6% of feedback comes from Base, far surpassing Ethereum, BNB Chain, Avalanche, and other networks, suggesting that active interactions and scenes are currently concentrated on Base. This may be related to the recent popularity of OpenClaw igniting a wave of Agentic ecosystems on Base.

Recognition metrics also highlight the current top-tier effect within the ERC-8004 ecosystem. Fewer than ten Agents have received thousands of stars, with most still awaiting market validation.

A noteworthy phenomenon is that most Agents support the x402 protocol, enabling autonomous, real-time micro-payments, and are beginning to become genuine on-chain economic participants, further driving machine collaboration and agent economy explosions.

Overall, the ERC-8004 ecosystem remains in early exploratory stages, but a decentralized on-chain collaboration network for intelligent agents has already begun to take shape.

This wave of on-chain AI Agent enthusiasm has also attracted many AI projects across various chains, including Chainlink, Filecoin, Render, Internet Computer, Bittensor, Virtuals, Bankr, Clawnch, and others. Meanwhile, some public chains, exchanges, and crypto projects are increasing their layouts—for example, Coinbase recently launched an AI agent wallet with built-in security features; Farcaster supports OpenClaw agents to autonomously create accounts; Virtuals announced a $1 million monthly Agent incentive plan; and AI.com, acquired at high prices by Crypto.com founder, is positioning itself in AI agent services.

Making Ethereum a Home for AI: Four Short-term Construction Directions

AI is the next core narrative Ethereum is advancing toward. Last September, the Ethereum Foundation (EF) established a dedicated dAI team with the goal of building Ethereum into an infrastructure for AI development.

However, in the current era of rapid AI development, Ethereum, which champions the “world computer” narrative, faces practical challenges. The computing power of large GPU clusters far exceeds that of blockchains, and training and inference of large models seem difficult to complete quickly on Ethereum.

Thus, Ethereum has chosen a different AI strategic path. Instead of competing in raw computing power with centralized giants, it positions itself as a trust and verification layer for the AI ecosystem.

Vitalik Buterin, Ethereum’s co-founder, endorses the vision of making Ethereum a home for AI and recently shared his latest thoughts on the relationship between Ethereum and AI. He believes that Ethereum’s spirit and ideals align closely with the development path of AI—both emphasize enhancing individual freedom, decentralizing power structures, and improving societal resilience, rather than allowing centralized forces to expand unchecked.

But Vitalik also points out that Ethereum in the AI era should not simply copy existing solutions but should pursue differentiated innovation, deeply integrating the values of cryptography and AI to build a future that benefits human freedom, safety, and decentralized collaboration.

In fact, the current AI agent race has already begun, quickly moving from technical exploration to deep commercialization. Some well-funded accelerationists pursue “stronger models and faster progress for safety,” but this trend risks centralization of power.

In Vitalik’s view, the core issue of AI development is not about computing power or model size but about the direction chosen. He opposes blind acceleration without constraints and calibration, emphasizing that Ethereum’s AI development should adhere to two bottom lines: protecting human freedom and agency, and preventing AI or power structures from marginalizing society, as well as avoiding systemic risks from AI out-of-control or asymmetric offense-defense.

The recently implemented ERC-8004 standard, promoted by EF, can provide on-chain identity, reputation, and behavior verification for AI agents, allowing them to prove themselves on-chain and giving users choice, thus avoiding monopolization by centralized platforms. This replaces the past model where platforms set rules and AI served as both executor and judge, with a verifiable decentralized system—aligning with Vitalik’s emphasis on Ethereum’s core values of decentralization and censorship resistance.

Note: Vitalik’s depiction of Ethereum’s role at the AI intersection

For short-term infrastructure development, Vitalik proposes four key directions:

First, building trustless, privacy-friendly AI interaction tools. This includes local LLM tools, ZK API payments supporting anonymity, cryptographic solutions to enhance AI privacy, and client-side verification of server-side TEE and cryptographic proofs. He notes this effectively extends Ethereum’s privacy roadmap into LLM computation scenarios, allowing humans to maintain control during AI interactions.

Second, Ethereum as an AI economic interaction layer. This layer encompasses AI API calls, employment between bots, margin mechanisms, on-chain dispute resolution, and ERC-8004 reputation systems. He believes that enabling payments, collateral, arbitration, and reputation management via on-chain mechanisms can make decentralized AI architectures more feasible. This economic layer is not about financialization per se but about facilitating multi-agent collaboration without reliance on a single organization.

Third, creating a cyberpunk-style self-verifying world. In the past, ordinary users couldn’t audit code line-by-line or fully verify system security. With local LLMs, people can leverage AI for various assistance—using Ethereum applications without third-party UIs, generating and verifying transactions with local models, auditing smart contracts locally, understanding formal verification (FV) proofs, and verifying trust models of applications and protocols. This moves the concept of “full self-sovereignty” from idealism toward practical reality.

Fourth, reshaping markets and governance mechanisms. Many decentralized governance and market design theories (like prediction markets, secondary voting, combinatorial auctions) have long been limited by human attention and cognition. The advent of LLMs can greatly expand human judgment capacity, making these institutional designs operationally feasible.

In short, Vitalik does not see AI as an isolated technological revolution but as part of a larger framework of decentralized civilization building. In this framework, AI extends human capabilities, while cryptography constrains power structures—AI brings intelligence, cryptography provides defense and autonomy.

In this trillion-dollar new market of AI Agents, Ethereum’s new AI story is about building an ecosystem that promotes human freedom, prevents superintelligent entities from losing control, and advances decentralized AI architecture.