AI memes, agent tokens and the Pump.fun ecosystem posted the largest 30-day gains

Only three categories of tokens stood out with total gains in the past 30 days – AI agents, AI memes, and tokens from the Pump.fun ecosystem. Data from Alphanomics shows that despite the accumulation of other tokens, they are still underperforming as attention shifted to the hottest trends.

AI agent tokens are the best performers for the past 30 days of trading, marking the peak of a trend that only started at the end of 2024. The collection of AI agent tokens grew by 71.7% on average, driven by the top projects that gained the biggest share of attention. The sector lost around 5% of its valuation in the past 24 hours, though it retained its leading position.

AI memes rallied by 44.4% in the past 30 days, retaining momentum from social media cults and their proximity to some of the AI agents. The pure AI narrative of utility and DePIN tokens expanded by 16.1% for the same time period.

The third most active category of tokens was the Pump.fun ecosystem, gaining a total of 38% in the past month. Pump.fun tokens gained prominence for their organic origins, with some of the memes being selected for their wide adoption and lack of cabal ownership.

Pump.fun is being used to launch AI agents, independent of other platforms like ELIZA or Virtuals Protocol, with most of the tokens having sufficient liquidity from their creators to move to Raydium.

Those relatively new narratives diminished the hopes for a wide altcoin rally. The new narratives displaced a brief expansion for ‘dinosaur coins’ at the end of 2024. After that, the trend shifted to projects that could gain immediate social media appeal, while offering a functioning product.

All of the three categories had outlier tokens, which drove the general trend. Most of the tokens in those categories remain risky and can quickly incur deep losses. Even experienced whales have posted losses while trying to snipe the most promising AI agents.

There is also a crossover between ecosystems, where GRIFT and AVA give a boost to both Pump.fun and AI memes. AVA is also the hottest rising AI agent token, during a period of heightened competition.

The AI space showed it can outperform in the short term, though the gains have been tested after Bitcoin (BTC) dipped under $94,000. With Ethereum (ETH) in the $3,200 range and Solana (SOL) under $200, the markets may break the 30-day trend.

AI agents turn to the green

AI agents retained a total valuation above $15B, with more than $2.4B in daily trading. The tokens of most agents are often shifting to the red, due to their low liquidity and profit-taking from early whales.

However, AI tokens also proved resilient, often switching back to gains within hours. For Ethereum, the growth drive came from Virtuals Protocol (VIRTUAL). The chain carried a wider selection of utility and task-specific AI agents.

Solana, on the other hand, is more skewed toward personality agents, with AVA taking the lead.

Solana AI agents shifted to the green, while competing for the biggest mindshare. | Source: Bubblescanner

Solana AI agents aim for memetic appeal while building communities around their content and personalities. Agents like AI16Z and Eliza also have DeFi and utility features, while still focusing on their content and presentation.

Base has already grown its own mix of utility and memetic agents. So far, only the top chains have built AI ecosystems, while BSC still has less than 20 agents. L2 chains like Arbitrum and Optimism are just starting out with the agentic trend.

AI agents, however, are not simply displacing other projects and tokens. Just like the Web3, gaming, NFT, and other trends before that, older tokens provided the liquidity and infrastructure. The success of AI tokens hinges on the availability of DEX, lending, market makers, and an entire ecosystem capable of reacting quickly to new communities and trends.

The AI agent space is yet to expand the real capabilities of agents, which currently focus on content and user engagement. For trading and on-chain activities, most of the tasks are still in the testing stage.

Source