Deep dive into AI Agents: Evolution, Applications, Token Role, and Capitalization

Compiled by: Felix, PANews

Since the beginning of 2025, AI-related on-chain activity has surged 86%, with daily active unique wallets (dUAW) reaching approximately 4.5 million. This brings AI’s dominance to 19%, second only to games at 20%. Considering that its market share was only 9% at the beginning of the year, this is a significant change.

This explosive growth is not just due to hype, it reflects a structural change in the way users interact with dApps. Whether through DeFi, social agents, or autonomous gaming assistants, AI agents are becoming the new on-chain interaction layer. They are not replacing users, but expanding users, automating, optimizing, and interacting on their behalf.

The data confirms this statement: AI has dominated the discussion of Web3 in the past month and will likely define its next phase of development. This report explores the development history of AI agents, application scenarios, the role of tokens, and the scale of capital supporting this transformation.

Key Takeaways

In June, AI’s on-chain dominance rose to 19% from 9% in January, transaction activity increased by 86%, and the number of daily active independent wallets reached 4.5 million.
As of June 2025, AI agent projects have raised $1.39 billion, up 9.4% from the total funding for the whole of 2024.
Since November 2024, 17,124 agents have come online on Virtuals Protocol, with an average of more than 85 new agents added every day.
Although AI Token’s market capitalization is down 64% from its early June high, it still currently stands at $5.9 billion, with a 24-hour trading volume of $1.4 billion.
Europe (26.2%) and Asia (21.9%) are the regions with the most AI dApps, followed by North America (15.8%).

1. What is an AI agent?

AI agents are autonomous software programs that can perform tasks, make decisions, or interact with users based on goals, prompts, or real-time data. Traditional AI agents cover industries such as finance, healthcare, or customer service, while the Web3 ecosystem is giving rise to crypto-native agents with unique capabilities and roles.

In Web3, AI agents are moving towards specialization. Some act as on-chain DeFi agents, executing trades, managing yield strategies, or acting as portfolio “managers.” Others are social agents, representing users in decentralized social applications, managing profiles, and even responding to messages and interactions. In the gaming space, a new class of native gaming agents is emerging. These AI companions are trained on the game context, game mechanics, or player behavior to act as helpful guides or even opponents.

This is not just theoretical talk. According to cookie.fun, which tracks the agent economy, there are currently 1,748 AI agents active in various environments. Since its launch in November 2024, Virtuals Protocol, which allows users to create and deploy their own AI agents, has launched more than 17,000 agents. The actual number may be even higher, as multiple chains are currently prioritizing the construction of infrastructure for AI agent creation, training, and deployment.

The agency economy is taking shape and growing rapidly.

2. AI Agent Tokens: Utility, Hype, and Capital

In Web3, most AI agents are not launched standalone, but come with tokens. The role of these tokens depends on the vision of the project and can serve a variety of functions. In many cases, they are intended to support community governance, provide access to advanced features, or serve as a mechanism for raising funds for development, especially when training and maintaining AI agents still requires a lot of resources.

For others, tokens are simply a capitalization strategy, a way to capture liquidity, reward early adopters, or ride on market momentum. In some cases, they are more like meme coins in AI guise, tokens that are issued around vague proxy narratives, lack technical substance, and simply ride on the coattails of speculative enthusiasm.

Despite all the noise in the market, the AI agent token market has made impressive progress. As of the time of writing, the total market value of AI-related tokens has reached $5.9 billion, accounting for 0.18% of the entire crypto market. The daily trading volume in the past 24 hours is also considerable, exceeding $1.4 billion.

However, the recent trend is not encouraging. At the beginning of this month, the AI agent space was valued at $16.6 billion, which means the space has lost 64% of its value in just a few weeks. The drop reflects broader market conditions, not just attitudes toward AI. But it also illustrates the volatility inherent in early-stage spaces, where token hype often outstrips actual utility.

AI agents are here to stay. But their tokens?

3. Top blockchains that empower AI dapps

While AI agents are often the visible layer used to interact with users, perform transactions, or provide in-game assistance, their success depends largely on the underlying infrastructure. Blockchains that support high-volume AI dapps are effectively laying the foundation for the deployment, training, and interaction of AI agents.

Between January and June 24, 2025, the following blockchains were the most active in terms of AI dapp usage:

Matchain dominates with nearly 1.9 million daily active users, indicating that AI infrastructure is thriving, driven by social or agent-oriented dapps. opBNB and Nebula follow closely behind, both showing strong usage related to lightweight to gamified AI services.

While not all of these dapps have agents deployed yet, the momentum is clear. As AI agent frameworks mature, these blockchains may usher in the next wave of autonomous agents, whether in DeFi, games, or social applications.

We are probably still in the infrastructure phase of the AI agent boom, and these networks are leading the way.

4. Where do users come from?

AI agents may exist on-chain, but their users are spread across the globe. Understanding where user engagement is coming from can provide a clearer picture of adoption trends, localization needs, and potential market opportunities.

According to Dapprader traffic data, Europe leads in AI-related dapp interactions between January and June 2025, accounting for 26.2%. It is followed by Asia at 21.9%, followed by North America at 15.8%. South America accounts for a smaller proportion of 2.5%, but its user base is growing. Interestingly, 33% of traffic comes from unspecified or difficult to categorize regions, collectively referred to as "others".

This global distribution shows that AI agents are not limited to a single region. Whether it is a DeFi agent managing transactions in Asia, a social agent representing users in Europe, or a gaming partner interacting with players in North America, the demand is diverse and increasingly cross-continental.

As the industry matures, expect to see more regionalized agent behaviors, better language modeling, and even agent personalities with specific geographies. Currently, the playing field is vast, and the race for user attention is global.

5. Money is pouring into AI agents

The AI narrative continues to dominate news headlines and funding rounds. While centralized AI giants like OpenAI, Anthropic, and Mistral have raised billions in funding, the AI agent economy in Web3 is also starting to gain traction.

As of 2025, AI agent projects have raised $1.39 billion, up 9.4% from 2024. This shows that investors are increasingly convinced that autonomous on-chain agents may be the next frontier. While this number is still far lower than the amount invested in centralized AI, it is worth noting that the amount of funding in the AI agent space is now comparable to or even better than other Web3 sub-sectors such as blockchain gaming.

This contrast is significant. Centralized AI investment still dominates, with tens of billions of dollars flowing into model development, chips, and infrastructure. But in the Web3 space, investors are increasingly looking to AI agents as a new “primitive” that can reshape how users interact with protocols, navigate DApps, and even automate personal financial management strategies.

The momentum is building. If it continues, 2025 could be the first year that AI agents attract more funding than any other Web3 vertical.

6. Conclusion

The rise of AI agents marks a profound shift in how users interact with decentralized systems. From DeFi traders and social partners to gaming native assistants, agents are rapidly evolving from experimental bots to infrastructure.

The numbers bear that out. Over 17,000 virtual agents have come online since the end of 2024. Even in the downturn, the AI token market is $5.9 billion and growing. This year alone, agent-focused startups have raised $1.39 billion in funding. And user engagement is global, with strong performance in Europe, Asia, and North America.

Still, challenges remain. Many tokens are driven more by hype than utility. Not all agents deliver on the promise of autonomy. And cross-chain infrastructure remains uneven. But as tools mature and real use cases become more widespread, the agent economy is approaching a new baseline where interactions with on-chain AI are no longer the exception, but the norm.

The race to create smarter agents, stronger ecosystems, and clearer standards is already underway. And it’s still in its infancy.