Editor's Note: After the AI Agent bubble burst, Web3 AI enters a new phase: moving from hype to infrastructure and practical applications. Decentralized AI (DeAI) gains attention due to data privacy and autonomy demands, with ecosystems like Bittensor gradually rising. The future will focus on combining intelligent agents with underlying technologies that have real use cases, driving industry maturity.
Below is the original content (slightly edited for readability):
AI agent market value soared to over $20 billion within months, then collapsed just as quickly. However, the field is gradually maturing: infrastructure, decentralized AI, and practical uses are becoming dominant. The next wave is forming, and here's why we should pay attention.
In the fourth quarter of last year, we witnessed the explosive growth of the "AI agent" sector, growing from zero to over $20 billion in just a few months - from amusing, charming, entertaining, and somewhat "unintelligent" AIs to those promising to make you rich overnight through trading and investment. Not
only that, but there were various "get rich" agents and numerous investment-oriented DAOs - organizations composed of humans or AIs (3,3) to invest in other AI agents.
From Hype to Infrastructure
We've learned that when a new field emerges (especially with new catalysts like Web2 AI, Trump supporting Crypto and AI), people don't care about fundamentals at all. As long as a project looks lively, has a cool demo, and generates buzz, its market value can easily reach billions.
@virtuals_io became a representative of this ecosystem, perfectly executing market promotion, capturing public attention, telling the most attractive story, and building the best narrative. This not only attracted developers to build projects but also drew retail traders to participate in the hype.
Then there's @elizaOS, taking a different path - open-source AI, allowing any developer to use this "gold mining shovel" to create their own intelligent agent. Around this concept, a massive community quickly formed, with adoption speed extremely fast, and GitHub stars and forks growing rapidly (and still increasing).
Some top projects in Virtuals once valued over $5 billion, Eliza's peak was about half of that, and other standout projects reached eight or nine-digit valuations, with AIXBT reaching a peak market cap of around $1 billion.
Now the situation is completely different, with newly launched and well-performing agents averaging $3-10 million in market value; established and continuously operating ones are between $10-50 million. The valuation ceiling has been significantly compressed, with total market value dropping from the peak of $20 billion to the current $4-6 billion range.
Infrastructure Momentum & Web2 Acceleration
The current market focuses more on "pure fundamentals" - especially infrastructure and decentralized AI, as AI models in the Web2 domain are rapidly evolving: Meta's Llama, OpenAI's GPT, Musk's Grok, DeepSeek, Alibaba's Qwen are releasing new, more efficient generations every month. For example, ChatGPT's latest image generation feature immediately sparked a viral trend of Ghibli-style images.
More critically, the Web2 consumer layer is developing even faster - because AI model capabilities are becoming stronger, many previously impossible functions are now achievable. Tools like Lovable, Bolt, Cursor, and Windsurf allow developers to iterate products more quickly. Agentic workflows and AI agents are ubiquitous, with extremely low entry barriers and almost zero user switching costs - don't like this App? You can immediately find an alternative with better UI and pricing.
Data Sovereignty Awakening
Meanwhile, more people are beginning to think: "There are centralized AI agent Apps everywhere, so who does my data belong to? Where did my data go? When I share private conversations with AI, will it really keep them confidential? Or will they be used elsewhere?"
This concern becomes more important because @OpenAI recently updated ChatGPT's "memory" function - it can now review all your conversations to provide more personalized responses. This means everything you say to AI is remembered and can be repeatedly referenced.
Bro... these features are indeed cool and might trigger a wave of "personalized AI agents" like dedicated AI assistants, AI co-pilots, personal secretaries, AI therapists, companion AIs, etc.
But imagine if these AIs use centralized technology and your data is "owned" or controlled by someone else - the consequences could be severe.
The Rise of Decentralized AI (DeAI)
I made several predictions last year, one of which was: Decentralized AI will begin to emerge in the second quarter of 2025. At that time, AI infrastructure focusing on privacy protection, data transparency, verifiability, and user ownership will receive more attention and broader use - because demand for these will only grow stronger.
Current trends can be divided into three main categories (though they do intersect):
·Web2 AI Investment Trend
Some YC (Y Combinator) incubated companies are frantically launching vertical AI agents; a16z (Andreessen Horowitz) is also positioning for future consumer-level AI trends through various investment arguments; Perplexity AI recently launched its own AI fund, directly investing in the AI track.
·Web3 AI Investment Trend (DeAI infrastructure investment, distributed training, inference networks, etc.)
·Web3 AI Retail Trend (AI agent ecosystem, consumer-facing agents, consumer-level AI applications)
Web2 vs Web3 AI: Completely Different Atmospheres
For Web2, due to its much larger total addressable market (TAM) - meaning many enterprises want to transform/optimize their business with AI, improve workflows, attract more potential customers, increase conversion rates and sales, improve customer retention, reduce management costs, and achieve more efficient operations - these enterprises seek solutions that solve their specific pain points and meet their vertical market needs.
This "optimization demand" attracts many young entrepreneurs to try improving traditional workflows with AI agents. Compared to traditional SaaS, AI agents can significantly save costs or generate more leads, allowing these startups to charge higher subscription fees (which is why we see many startups achieving 7-8 digit ARR in just a few months).
Web3 venture capital trends are very different. Because blockchain is naturally the ideal infrastructure for decentralized AI, such as verifiable/immutable transaction records, trustless environments, decentralized computing power, AI inference and training under minimal trust premises (these terms might be complex, but you get the idea).
In summary: The future direction is - people know how their data is processed, understand AI's thinking logic, own their data, models, and application scenarios, and can be incentivized to share without censorship. And Web3 VCs are continuously investing in these futures.
Why Retail Investors Like AI Agents (Even If They Don't Understand DeAI)
For Web3 retail investors, DeAI is difficult to understand, as it requires mastering numerous technical terms and understanding its core value (sometimes it seems like an alien language). Therefore, they are more inclined to flock to projects that are "easiest to understand" - such as starting with chatty, humorous, and interactive AI agents.
But as they continue to stay in the industry, they will realize: these things cannot create sustained value for users (yes, many AI agents are actually useless and lack originality). This awakening, coupled with the current sluggish and volatile market environment, has led to market clearing - useless agents are being eliminated, while useful ones are surviving despite significant valuation declines.
People are beginning to realize: there must be truly useful AI products and landing scenarios. This awakening is driving teams to either develop real products or integrate with AI projects that have genuine technology, such as @AlloraNetwork and @opentensor (Bittensor).
This transformation has two benefits:
1) It makes people start paying attention to less understandable but truly valuable infrastructure;
2) It allows AI agents to obtain actual application scenarios to showcase to the community.
Before this change: Agents could only perform basic functions (chatting, post analysis)
After this change: Agents master more advanced and practical skills (AI-driven betting, trading, market-making, mining, etc.)
Agents like @AskBillyBets and @thedkingdao have already become representatives showcasing Bittensor subnet capabilities, bringing cool technology into public view.
Bittensor Ecosystem
What I find particularly interesting about the Bittensor ecosystem is: it's a decentralized AI ecosystem that you can directly participate in investing. Currently, most DeAI projects are in very early stages, where only VCs or strategic investors can participate, and ordinary people can't invest because many projects haven't issued tokens yet.
But on Bittensor, anyone can stake their $TAO in the subnets they want to support, thereby exchanging it for the subnet's Alpha Token (equivalent to "All In" on these DeAI projects).
I've previously complained about poor user experience in bridging and trading, but in terms of technology, products, and atmosphere, especially teams like @rayon_labs, they are truly strong and have great vibes.
What I like about Rayon Labs is that they are building consumer-facing UI/UX optimized products. Considering that the dTAO mechanism is determined by the market for how much incentive emission each subnet can obtain and its valuation, this makes "whether each subnet can build easily understandable and usable products" increasingly important.
Rayon has many cool subnets (possibly the coolest being Gradients, an autoML platform where you can easily train models), and even cooler is their latest flagship product - Squad AI Agent platform, where you can create an agent by dragging and dropping modules (similar to node builder style, like @figma).
One final note: I'm currently in the early stages of deeply understanding Bittensor, and I will write a dedicated article later sharing some interesting discoveries and how to seize opportunities. If you also want to understand other trends and changes in the market, you might want to check out the article I recommend below.
Welcome to join BlockBeats official community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Communication Group: https://t.me/BlockBeats_App
Twitter Official Account: https://twitter.com/BlockBeatsAsia
