Introduction
We found that every time we start researching the industry, the crypto market always experiences a big rally. BTC has broken through 10,000 USDT, and ETH has even stood above 2,500 USDT. Haha, is this the so-called reward for diligence? Just kidding, it's actually because the United States and the United Kingdom reached a trade agreement, and the initial talks between the US and China went well (first revealed on Trump's Twitter, then confirmed by both countries). This still indicates that the entire crypto industry remains a high-risk financial speculative market, following the overall economic environment, and has not yet formed an independent market like gold and oil. This has further solidified our determination to find long-term development breakthroughs in the crypto field, and AI + WEB3 is currently the track we value most.
In this article, we continue the previous train of thought and review the five waves of AI development in WEB3 over the past few years. We'll first summarize it in a table, of course focusing on the current fifth wave:
Review of Five Waves of Web3 AI Projects
Since the intersection of blockchain and AI, the crypto field has experienced five major AI concept waves. Each wave was led by representative projects or events, showing an evolution from pure concept speculation to gradual technical exploration.
[The rest of the translation follows the same professional and accurate approach, maintaining the original structure and meaning while translating to English.]$GOAT Price Change, Source: CoinMarketCap
04, Fourth Wave - a16z Ecosystem AI + Rise of AI Agent Basic Concepts
After previous hype waves subsided, a group of AI projects focusing more on technological and ecosystem construction emerged in late 2024. Unlike previous community-driven hype, this stage saw the rise of AI platform projects with notable backgrounds or serious visions. Typical examples include a16z and Virtuals, viewed as representatives of the "a16z AI ecosystem" (dramatically mimicking the name of the famous venture capital firm a16z). These projects share a common goal: building AI Agent infrastructure, attempting to become the "app store" for Web3 intelligent agents.
$AI16z was a phenomenal AI Meme coin that rose in November 2024, alluding to the famous venture capital firm a16z. The project essentially positioned itself as a decentralized AI trading fund on Solana, or an "AI Investment DAO", primarily using AI agents to obtain on-chain and off-chain market information, analyze community sentiment, and automatically execute trades. This novel concept instantly sparked market attention: just one day after launching in late October 2024, its market value skyrocketed from zero to around $80 million, igniting speculative enthusiasm. Surprisingly, a16z founder Marc Andreessen even endorsed it on Twitter, personally resharing AI16z's concept introduction and link. Backed by a notable figure and the "AI Investment + Meme" narrative, AI16z became the hottest crypto topic that month.
While many in the industry questioned its high hype-to-substance ratio, it undeniably opened up imagination for the "AI + DAO autonomous investment" model. To further enhance token value, the AI16z team proposed an economic plan in December 2024 to evolve into an "AI Layer1": launching a Launchpad specifically serving AI agents, making the AI16z token the base currency and fee medium for intelligent agent transactions, and designing node incentives and value buyback mechanisms. This stage shows the project team attempting to transition from pure hype to building a sustainable ecosystem, supporting token value through technical capabilities and economic models rather than just storytelling.
During the same period, the Virtuals protocol was another representative project that was highly successful in ecosystem promotion. Its team constructed an engaging narrative, attracting numerous developers to create AI agent projects on its platform and drawing many retail investors. At its peak, the top projects in the Virtuals ecosystem had Fully Diluted Valuations (FDV) exceeding $5 billion, with the Eliza framework's corresponding token reaching about half of that, around billions of dollars. Other standout projects (like AIXBT) reached market caps of hundreds of millions of dollars. During this time, capital gave extremely high valuations to "AI agent platform/framework" projects. However, user scale and actual revenue were far from matching these valuations - for instance, despite Virtuals claiming a thriving ecosystem, its daily revenue was reportedly less than $500 after launch, creating a stark contrast with its astronomical valuation. As the market cooled in early 2025, these projects' token values were generally halved or collapsed, proving that grand visions and narratives alone cannot sustain prices.
Therefore, the representatives of the fourth wave have not truly become the "app store" for AI Agents, but more resembled Pump.fun, a platform concentrated with AI Agent Memes. While everyone was waiting to see if they could be reborn, the fifth wave surged in.
05, Fifth Wave - "Hundred Flowers Blooming" Closer to Practical Applications
After a brief downturn, the WEB3 AI sector reheated in early Q2 2025, with hot spots becoming noticeably more "substantial": focusing on next-generation AI Agent real platforms and tools; new AI ecosystem chains specifically suitable for AI project operations; decentralized data utilization and value storage; and even combining WEB3 AI with other high-tech tracks like robotics. These presented a "hundred flowers about to bloom" form, and we look forward to the day these practical projects truly "bloom together". Below we list some cases from the fifth wave for reference:
(Translation continues in the same manner for the rest of the text)The emergence of the first AI chain, Hyperion, is aimed at solving the problems encountered by all current L1 and L2 chains: WEB3 AI projects require larger-scale computing power to meet frequent read and write operations and computational states while maintaining decentralization, which previous EVM environments could not support. The Metis team's newly released Hyperion solution can be simply understood as: reshaping a new chain most suitable for running WEB3 AI or game projects (both requiring high-performance computing) by using Metis SDK and Alith AI Agent framework to address the challenges faced by development teams. Hyperion still uses $Metis as its sole token, which makes us believe that the Metis team genuinely wants to do something substantial in the AI sector, rather than just shouting empty slogans - this is likely the reason for the recent stirring of the Metis token's trend.
If the first AI chain Hyperion was just a small test by the Metis team, as many have heard other public chains claiming to create high-performance AI public chains, then the mysterious second AI chain LazAI is their full-scale campaign into the AI field. LazAI will be an entirely new Layer1 aimed at solving the value pricing mechanism for data contributions by individuals (including personal and institutional). At first glance, it sounds complex, but an example will make it clear: everyone knows that giants like Alibaba or ByteDance are profiting by selling personal data privacy to merchants, and you can do nothing about it unless you stop online shopping and short video browsing. However, LazAI leverages blockchain's decentralized characteristics, ensuring that individual data remains owned by individuals. Originally, AI requires inputting massive individual data, which might be valuable to certain projects or merchants. Users can now obtain the value of their own data, which can be represented through tokens. Moreover, LazAI can use algorithms to distinguish the value of different data, thereby minimizing the crypto industry's pain point of studios or data farmers forcibly inputting useless data. Therefore, "decentralized classification and storage of personal data + data value pricing mechanism" will become LazAI's most significant feature, which we believe is the direction the entire industry should strive towards: how to use WEB3's inherent characteristics to improve the shortcomings of the WEB2 world.
We must say that such a grand narrative of Hyperion + LazAI can only be proclaimed by a project team like Metis, which appears genuine and credible, and we are beginning to believe they might truly be working towards this goal.
To be continued.
