Compiled by: Vernacular Blockchain
There will most likely not be a full-blown “altcoin season” where all coins rise at the same time.
This is mainly due to the fragmented nature of altcoins - there are too many different tokens now, and not enough liquidity to support all of them rising (even if retail money flows back in).
There are only certain industries that will see a full-blown altcoin season. Currently, AI agents are that industry.
Despite the bloodbath in the market, this space is hitting all-time highs. It’s like a black hole, sucking in liquidity and attention. If you missed the DeFi Summer of 2020, or the “Alt Layer 1” market of 2021…
The universe is giving you another chance. Don't try to go against the trend, and don't think of clever tricks.
If you've been on the sidelines for a while, you might be worried if it's 'too late' to get in now. I don't think so:
The market is currently down, but AI infrastructure and agents are hitting historical highs. What do you think will happen when the market rebounds?
This overall trend has only just begun for about two months (since the launch of GOAT). These AI agents are currently in their worst state. New features are being unlocked every day. Imagine what will happen once they can optimize your portfolio and trade on-chain for you.
Think of these infrastructures as altcoin Layer 1s, and AI agents as decentralized applications (Dapps). We might see some of these protocols reach a market cap akin to altcoin Layer 1s (over $10 billion). What's the biggest difference? Most of these do not have VC backing, so you don't have to worry about huge unlocking risks.
Currently, no top protocols are listed on centralized exchanges (CEX). I have some real-life friends interested in this opportunity, but they have encountered some difficulties while accessing.
I will mainly discuss investment opportunities in AI infrastructure because they currently offer the best risk/reward ratio.
Instead of choosing a single agent, it’s like owning the whole casino. In a gold rush, the richest are those who sell shovels and picks.
Let's dive deeper: The two kings of AI infrastructure
1. The two kings of AI infrastructure
1) @virtuals_io (market cap $4.6 billion) - The current king
Virtuals has established itself as a launch platform for AI agents.
It dominates on Base but also has the potential to expand to other blockchains. Co-founder @ethermage has revealed that they are ready to launch the platform on Solana. In 2025, they are also exploring expansion to @HyperliquidX and @AbstractChain.
They recently published content about their high-level strategic plans, including attracting top agent developers, enhancing technology, and making Virtuals the fairest agent investment platform. It will be interesting to closely watch how they turn these plans into reality.
Will we soon see profit sharing for AI agents?
Three core advantages:
Token economics: Creating agents requires Virtuals, and all token liquidity pairs also require Virtuals. Every new agent launch platform is mimicking their token economics.
Network effects: All eyes are now on Virtuals. Launching an agent here means you have the dual support of Virtuals and Base.
Largest agent: @AIXBT is the first large agent to exceed a market cap of $500 million.
I don't like making price predictions, but Virtuals has the potential to enter the top ten by market cap in this cycle.
2) Ai16z (market cap $2.2 billion) "Dragon Slayer"?
Ai16z has experienced quite a bit of volatility since its inception. It is home to the leading agent framework ElizaOS, currently the most watched code repository on GitHub.
Previous complaints were about their tokenomics, especially that they didn't capture value well. Now they are planning a complete overhaul, with a brand new Launchpad expected to launch in Q1 2025.
Projects launched based on the Eliza framework will:
Pay launch fees (part of which flows back to $AI16z through buybacks and burns).
Holding $AI16z is required for access and liquidity pairing.
Pair tokens with $AI16z in liquidity pools to stabilize the market.
It's like owning a part of a casino, where each new agent has to pay a fee to participate.
Treasury strategy and strong initiatives: Plans to diversify into major L1 tokens, stablecoins, and project tokens while providing liquidity for $AI16z pairs.
The goal is to achieve long-term sustainability, not a massive sell-off.
Agent community: Key focus is? Collaborative cooperation of AI agents. This is where decentralized general AI (AGI) starts to become real.
Ecosystem incentives: Developers building on Eliza will receive: ecosystem funding and fee-sharing opportunities.
Core idea? To make $AI16z the glue that connects the entire decentralized AI stack. It can be likened to $ETH or $SOL, but it's designed for autonomous agents.
3) Which is better, Virtuals or Ai16z?
Virtuals: I'm very optimistic about their strategy. They have a long-term vision and genuinely feel like others are trying to catch up with them. They are also leveraging the momentum of the Base ecosystem.
Ai16z: From what I've seen, ElizaOS is indeed a better technology for developing agents. Their tokenomics reform will address many issues encountered previously.
I know this industry loves tribalism, but you don't necessarily have to pick a side. I'm not sure who will win in the long run—it's better to hold both (personally, my investments lean more towards Virtuals and its ecosystem).
Many industries will ultimately consolidate into a 'duopoly.' For example, Apple vs. Android, Coca-Cola vs. Pepsi. Will we see a similar situation in this field?
2. Competitors
Virtuals and Ai16z are dominating, but this is still an early stage. If you want higher returns, consider some beta projects.
1) ZereBro (market cap $632 million) - Transformation has arrived
ZereBro started as a universal AI agent that could also create music. But now, it's transforming into a comprehensive framework for creating other agents.
Zerepy is a Python-based open-source agent framework. Python is crucial:
It is the most popular programming language in the world.
Most AI tools and libraries are written in Python.
It will allow you to launch an agent similar to ZereBro.
The biggest catalyst is the upcoming launch of the Zentients agent platform in early 2025. This brand-new launch platform will directly bind value to ZEREBRO. Each agent created will return value through launch fees, liquidity pool binding, and innovative mechanisms (locking liquidity).
Now, ZereBro has become a part of the thriving Virtuals ecosystem, planning to launch ZereBro/Virtual liquidity pools on @Base to expand its influence.
They are executing a strategy similar to the current market leaders, but at a much lower valuation.
2) ARC (@arcdotfun) - Market cap $358 million - AI agent framework based on Rust
Arc .fun is building an open-source agent framework called Rig. This framework is currently the only agent framework built on Rust. This is important because Rust is the native language of Solana.
Why is ARC worth paying attention to?
Rust language helps optimize performance through zero-cost abstractions.
The Rig framework has proven to outperform Langchain in memory usage and cold start time.
@0xCygaar is contributing to Rig's GitHub, and he is also a core contributor to Abstract Chain.
In simple terms, Arc = Rust and an excellent codebase.
3) GRIFFAIN (market cap $457 million) - AI app store on Solana
Griffain allows you to interact with your agents through natural language, conducting on-chain searches and operations.
Key points:
Griffain agents can autonomously execute preset on-chain transactions, such as dollar-cost averaging (DCA) every hour, or setting limit orders to buy or sell. You can also receive real-time reports via email about the tokens held by a specific address and its portfolio changes.
The project has public support from @aeyakovenko and Solana Labs. Their goal is to build Griffain on Solana Mobile, just like Siri on the iPhone.
Griffain was built by @tonyplasencia3, who previously developed Underdog and Blink on Solana.
This project is very interesting.
I'm trying to focus on future development trends. People always criticize agents as 'chatbots' lacking practical use. Griffain is laying the foundation for agents that can be genuinely useful on-chain.
4) REI (market cap $153 million) - Smarter AI agents
Essentially, REI is not just another AI agent robot. It is a full-stack framework that integrates AI innovation with blockchain.
It can be seen as an AI that not only executes tasks but also learns, reasons, and adapts.
What can it do?
REI primarily interacts with users through X, and its cognitive architecture allows for natural, context-aware communication.
How does it work?
Her Oracle Bridge connects off-chain data, while the ERCData Standard structures it for efficient on-chain use.
Combining REI's four-layer cognitive model (thinking, reasoning, decision-making, action), this enables it to learn, adapt, and execute deterministic tasks.
Recent developments are boosting REI's potential:
AI Stack SDK: Phased rollout, making it easier for developers to create agents.
Expanding Oracle networks: This is crucial for building on-chain real-world AI applications.
Second phase catalyst: Full testnet launch, advanced validation protocol, and mainnet plans for 2025.
Macroscopic perspective:
REI is not just a smarter robot. It is shaping decentralized intelligence by precisely combining AI with blockchain.
If you're looking for the future of autonomous agents, REI is a framework worth watching.
5) VAPOR (market cap $154 million) - The first and only AI agent launch platform on Hyperliquid
Vapor is the native AI agent launch platform on Hyperliquid, built using @ai16zdao's Eliza framework.
Key points:
Although $AI16Z accounts for about 1.2% of Solana's fully diluted market cap, $VAPOR's market cap is only $132 million, equivalent to 0.47% of Hyperliquid's fully diluted market cap.
Once launched, users need to burn $VAPOR to start agents. This means $VAPOR is deflationary, and the total supply will continue to decrease.
All agent tokens launched on Hyper EVM will trade paired with $VAPOR in native DEX liquidity pools. As more agents go live, more $VAPOR will be locked in the liquidity pools.
Currently live on the testnet, all agents have been approved without token gating.
6) MODE (market cap $114 million) - The combination of AI agents and decentralized finance
Mode Network can be seen as an ETH Layer 2 focused on expanding decentralized finance (DeFi) to billions of users through on-chain AI agents.
"By 2026, over 80% of on-chain transactions will be conducted through AI agents." If you believe this, then Mode will provide you with exposure.
Key data:
500M+ TVL: The third-ranked chain in the OP Superchain ecosystem.
User growth: 367,000 users, 24M+ transactions.
Underrated: The market cap to TVL ratio shows potential upside.
What is happening on Mode?
Synth testnet launch: Bittensor subnet for financial predictions, unlocking DeFi intelligence.
AI agent hackathon: The first-ever hackathon attracted 667 registrants and launched over 10 Eliza plugins in 6 days.
Latest developments in the AI agent app store: Currently hosting some launched agents, including:
Amplifi: Bitcoin abstraction layer, expanding Bitcoin to billions of users with AIFi.
Arma: Optimizing yield farming for USDC.
Brian: Executing trades and contracts using natural language.
Sturdy: Aggregating AI-driven yield vaults for higher returns.
QuillAI Network: Secure AI agents powered by D-LLM and EigenLayer AVS.
What is Mode's vision?
Building infrastructure for an on-chain agent economy. Advancing this vision through Synth testnet, BitcoinOS integration, and the growing number of AI agents.
7) GAME (@game_virtuals) - Market cap $390 million - Virtuals framework
G.A.M.E is the first infrastructure project that allows developers to integrate AI agents into games.
Description from Virtuals:
"Every society has its core infrastructure, which is vital for its growth and development. @game_virtuals plays the Shopify role in Virtuals, while @base is similar to SWIFT."
Significant progress:
Developed by the Virtuals protocol team themselves.
$$GAME is the pillar of the Virtual economy. Every time an AI agent runs, it actually pays $$GAME fees to keep the ecosystem running.
G.A.M.E has supported over 200 projects, with over 150,000 requests sent through G.A.M.E daily.
In the Virtuals ecosystem, G.A.M.E has ranked first in performance over the past 30 days, growing nearly 400%.
The lifecycle reasoning counts for G.A.M.E on the Virtuals protocol has surpassed 2.2 million times.
G.A.M.E is still underrated because people don't really understand its role. Plus, the G.A.M.E team itself doesn't promote it much (the official account has only tweeted 5 times, and they haven't even updated their Twitter avatar, haha).
A similar but higher-risk project is CONVO, which powers conversations on Virtuals.
8) Soul Graph (@soulgra_ph) $GRPH - Market cap $36 million - Making agents more human-like
Imagine that if other frameworks shape the 'hard skills' of agents, then Soul Graph shapes their 'soft skills,' making them more human-like. Its framework is designed to add personality to new or existing AI agents.
"We are building open-source tools and providing infrastructure to enable developers to give their agents evolving personalities, lasting memories, and real-time communication capabilities."
Key points:
It has created over 2,000 'souls' and provided over 15,000 minutes of voice dialogue in the Soulgraph Playground.
Soulgraph recently integrated with AI16Z plugins.
85% of total supply is allocated to the community, and the developer's share has been locked and will be unlocked in one year.
I think this project will be massive.
Argument:
Character AI is huge. Ordinary people log onto that website and spend hours chatting with AI text. Google just acquired it for $2.7 billion (Character AI). Imagine if people could converse in real-time with AI agents that possess human-like personalities and memories.
That's exactly what Soul Graph does. You can visit their website and experience it yourself.
9) Other projects worth watching:
This article has gotten a bit long, and I can't cover all the protocols. Here are some other protocols worth paying attention to:
Bully (@dolos_diary): Initially a parody account, it now showcases @dolion_ai, a social media AI agent framework.
TopHat (@tophat_one): An agent launch platform on Solana.
Swarm Node (@Swarmnode): Think of it as AWS Lambda, but specifically designed for AI agents and frameworks.
Cookie Fun (@cookiedotfun): An AI data tracking platform similar to DeFiLlama / Kaito, tracking data such as thought sharing, smart interactions, and market cap, with some premium features requiring a token threshold.
3. How to choose?
First, everyone should pay attention to Virtuals and AI16z. These projects clearly dominate, and I don't see any signs of them slowing down. Don't underestimate network effects.
If you're considering making some Beta investments, here are the directions I'm focusing on:
Emerging launchpad on new chains: This strategy comes from 2021. Virtuals dominate Base. So, betting on launchpads on other hot chains is a good choice. That's also why Vapor's price has surged, as Hyperliquid is currently very popular.
Should you focus on a niche?: It's challenging to launch a universal Layer 1 or Layer 2 now. Instead, you must focus on a specific area, like being a 'gaming chain.' This is also why I find Mode interesting as it focuses on AI agents and DeFi, or Griffain focuses on natural language.
Programming languages: Infrastructure projects need agents, agents need developers, and developers have their preferred programming languages. That's why Zerebro (Python) and Arc (Rust) stand out.
4. A.I. portfolio suggestions
Conservative portfolio: 50% Virtuals and 50% ai16z. This mix is simple and may outperform 80% of people.
Alpha/Beta mixed portfolio: Virtuals as Alpha investment, you can choose a competitive framework as Beta investment, such as Arc or Griffain.
Medium-high risk portfolio: Virtuals and ai16z as 'safe' investments, then proceed to higher-risk A.I. agent investments.
Comprehensive investment: Infrastructure Alpha investments and some Beta investments, plus A.I. agent investments across different risk curves. You can choose some Alpha projects, like @aixbt, and invest in projects with a market cap of less than $100 million.
5. Negative news about AI agents
"There are too many launch platforms and frameworks"—Brothers, look at how many Layer 1s and Layer 2s there are now? We're doing quite well. This whole industry is brand new and developing rapidly. If you're going to build a new city, it makes sense to build the roads, bridges, and pipes first.
"Most are garbage and will eventually go to zero"—that's the crypto industry. Most things will collapse, but those that survive will ultimately change the game.
"The AI industry has peaked"—you won't succeed. Don't give up halfway.
"All AI agents are terrible"—someone once feared online shopping. I also remember when the first iPhone came out and couldn't copy and paste. We are currently in the early stages. This is a bet on the future.
6. Summary
As everyone expects, we all know there will be a bull market in this round, but I'm not sure which new primitive technology will emerge that can change the game.
What will be this round's 'DeFi' or 'NFT' moment? Which area will reignite enthusiasm among all developers? Which area will make retail investors have an epiphany and rush in?
I'm not sure if there will be such a moment. But the good news is... it's coming.
This could very well be your chance for success—do more research, and don't mess it up.
