Author: @xingpt
Before analyzing the AI Agent sector, it might be better to step back and look at what crypto has experienced in this cycle as a whole:
Decoupling Bitcoin from the crypto space.
Before this round, Bitcoin was almost synonymous with the crypto space. Buying Bitcoin meant buying crypto assets, thus acknowledging crypto and decentralization.
However, after the approval of the Bitcoin spot ETF, from the US president to listed companies, the purchase of Bitcoin and acknowledgment of its value seems to have become mainstream opinion; however, the existential significance of crypto, especially Ethereum and Altcoins, does not seem to have gained recognition from mainstream society and funds.
The reasons are complex.
The main reason lies in asset positioning: Bitcoin is viewed as an alternative asset linked to gold, with its attributes of hedging inflation and preserving value beyond sovereign currencies widely recognized.
Ethereum and other altcoins, in the eyes of Wall Street's old money, are still considered tech meme stocks without mature, sustainable business models. However, compared to hard tech companies like Nvidia, Microsoft, and Amazon that have users, products, and demand, the valuations of Ethereum and other crypto coins are not low, and their return elasticity is insufficient, making them low-risk-return assets from an asset allocation perspective.
As shown in the figure below, Ethereum's Sharpe Ratio is lower than that of Meta, Google, and other tech companies. Bitcoin's Sharpe Ratio is only surpassed by the extraordinary Nvidia of this cycle.
Another important factor is that the overall macro interest rates and the degree of monetary easing cannot be compared to the massive liquidity injection during the previous cycle's pandemic. Coupled with the booming AI industry, the attraction of crypto to off-market funds is insufficient. The reasoning is simple: there is only so much money, and if it is spent on AI stocks and GPUs, there is no way to buy altcoins and Ethereum.
The M2 money supply in USD has not yet recovered to the high point of 2022 (data source: CEIC).
The internal circular ecosystem of the crypto space is extremely unbalanced (Imbalance).
Since the crypto space lacks the ability to attract off-market funds, can on-market funds leverage sufficient purchasing power?
We roughly estimate the total amount of on-market funds using the total amount of stablecoins + contract holdings, and it’s not difficult to see that the current total amount of on-market funds has far exceeded the last bull market. But apart from BTC, most altcoins have not reached new highs. Where exactly is the problem?
The root of the problem lies in the imbalance of supply and demand. The supply side has a large number of new projects with extremely high valuations, most of which have not found practical application scenarios (PMF) and have few real users.
The existence of these projects stems from the excessive financing of crypto VCs during the bull market of 2022. Due to the excessive funding raised by crypto VCs, most of this funding is limited to five years, with around three years for investment plus a two-year exit period. Funds ignore project quality in a frenzy to complete investment tasks.
So who will provide buying power for these projects?
Previously, the exit channels were mainly centralized exchanges, but after the FTX incident, centralized exchanges became the target of criticism. The long arm of regulation has made centralized exchanges suffer, worrying not only about being fined heavily but also about the risk of founders going to jail. This causes the objectives of centralized exchanges to shift from expanding users and increasing trading volume to obtaining profits.
User-oriented exchanges must inevitably offer benefits to users, including lowering the valuations of new projects, sharing opportunities for early project participation (IEO), and a series of offline activities to attract new users and expand circles.
Under heavy regulation, exchanges have simultaneously contracted offline and regional expansion efforts while also actively or passively halting profit-sharing operations such as IEOs. This has also led to insufficient demand and buying momentum in the market.
AI Agents have unique advantages compared to Meme coins.
As we all know, the core application scenarios in the crypto space are asset trading and asset issuance; and only by involving users in new modes of asset issuance and trading can wealth effects be generated, which in turn triggers the phenomenon of leverage for funds within the circle and the entry of funds from outside the circle during bull markets.
However, under the premise of high valuations for on-market projects and severe supply-demand imbalance, Meme coins became the first track to break through.
Meme is characterized by its lack of VC financing and fair launch, creating wealth effects through a rapid surge in low market cap, and driving new pathways for asset issuance (pump.fun) and asset trading (Gmgn, TG bot).
One of the most important features of Meme is that it has no actual use; this financial nihilism can deconstruct the VC setup and suits a minority of crypto users with IQs of 50 and 150. For the majority of practitioners and institutions with an IQ of 100, the difficulty of participation is still too great. It's hard to imagine explaining to a fund's LP that the reason to invest in Moodeng is that it is too cute, while the reason to sell it is that it has become fat and no longer cute (I still love Moodeng).
But AI Agents can unify the consensus of the majority: with fund LPs, one can tell stories about investing in AI infrastructure; with IQ 50 and IQ 150 degens, one can discuss the logic of on-chain memes and gold dogs; and with IQ 100 crypto practitioners and crypto VCs, one can discuss the logic of investing in AI Agent sector projects.
In short, AI Agents are the greatest common divisor of the web3 industry in this cycle.
How I view AI Agent projects.
How should IQ 100 view AI Agents - From Dapp to AgentApp.
IQ 100 includes the vast majority of practitioners and investment institutions in the crypto space, including myself, so let’s start analyzing AI Agents based on the investment framework I am familiar with.
I believe that for investment institutions in the crypto space, the most important point is to understand that AI Agents are reshaping the upstream and downstream industry chain and valuation logic of crypto.
Through the two waves of crypto bull markets from 2017-18 and 2020-21, the industry chain and valuation logic of blockchain projects have gradually taken shape.
Underlying public chains; the market cap ceiling is Ethereum, currently at 400 billion USD; Solana, the second-largest, has a market cap of about 1/4 of Ethereum, and may reach 1/3 or even 1/2 in the future.
Middle layer: such as the oracle Chainlink, with a fully circulating market cap of 20 billion USD, about 5% of Ethereum;
Basic protocols like DeFi, Uniswap FDV market cap is 13 billion USD; AAVE has a market cap of about 5 billion USD; they represent 3% and 1.25% of Ethereum, respectively.
The underlying logic of DeFi is based on smart contracts, and the functional limitations of smart contracts also restrict the innovation of other applications in the crypto space.
Now that AI is incorporated into the underlying tech stack of blockchain, the AI layer becomes a parallel technical base alongside smart contracts, known as the fully-onchain AI Agent layer.
Image source: (https://x.com/karsenthil/status/1874471383066984706)
This can also explain another problem: Why previous AI projects did not lead to new narratives. Because whether it is token-incentivized projects for sharing GPUs, data, data labeling, etc., they still treat blockchain as an incentive layer and have not stepped out of the realm of smart contract applications (DAPP). The application of AI Agents, however, exists as an adhesive between the blockchain base layer and off-chain data, and serves as a better UX interface. (Reference: https://x.com/jolestar/status/1872935141326373237)
So, based on this logic, let's look at valuations. If the leading middleware of DeFi, Chainlink, can account for 5% of Ethereum's market cap; analogously, if we boldly assume that the leading framework of AI Agents can also account for 5%; currently, ai16z's market cap is about 2.5 billion USD, with another 8-10 times the space; this is just an example of ai16z, and there may be other better Agent frameworks emerging.
Platforms like Virtual that come with a built-in framework are equivalent to Chainlink + Uni. Chainlink + Uni currently has a market cap of 33 billion USD, while Virtual's 5 billion USD calculation still has six times the space.
Freysai (FAI) is somewhat like AAVE, discreet but delivers extremely high product quality. The verifiable applications of AI TEE will also become standard for future AgentApps, with a market cap ceiling of 1.25%-3% of Ethereum, corresponding to 5 to 10 billion USD.
Other leading comparisons are similar to Spore, equivalent to the stable projects or Launchpad of the last cycle, aixbt is analogous to DeFi aggregators like 1inch, with a lower limit of 1 billion USD valuation and an upper limit depending on market developments.
Other AI projects can be referenced independently; I will not elaborate on them one by one.
How should IQ 100 view AI Agents - How to change the internal competition ecology of the crypto space.
In this round, most of the wealth effect occurred on-chain since the inception of Meme, but on-chain memes still present a high operational threshold for off-market users and institutions with IQ 100, as most users still prefer to trade on exchanges.
But the biggest advantage of AgentApps lies in interaction.
Buying assets: In the past - from centralized exchange apps and websites, recharge and place orders to buy coins;
Agent Era - Agent applications can directly buy coins using natural language and even intelligently assist in trading and investment decision-making.
Financing: In the past - conceptualize, assemble a team and package it, find VC for seed round financing, gather top VCs or exchanges to take over and raise valuations;
Agent Era - Directly access GitHub and products, spread the word in the community, and the community directly funds it.
**Issuing Tokens:** In the past - launch on the testnet, announce VC financing high valuation information from six months ago, attract studios to test the net, negotiate terms with exchanges and get cut, then issue tokens and dump to recover costs;
Agent Era - AI automatically issues tokens, Agents hold private keys, Agents add liquidity pools, and Agents promote in the community;
The entire cycle follows several important standards:
Projects are open-source, applications are real and visible, and codes can be queried and verified.
Funds are relatively safe, with private keys in the hands of Agents to avoid Devs withdrawing liquidity.
Financing and token issuance transparency, avoiding the issues of exchange token listings, opaque airdrop rules, and the core circle of VC collusion.
Of course, issuing tokens also has issues such as on-chain front-running, KOL information advantages, etc., but compared to past opaque operating methods, significant progress has clearly been made.
AgentApps that can win users' trading entry points are highly likely to be benchmarked against the valuations of exchange platform tokens.
How should IQ 150 view AI Agents?
As I am Mid-Curve, I would like to reference my friend Alen's opinion from over a month ago (
https://x.com/qiqileyuan/status/1858357959807635854), he believes AI
The future of Agents will form a new AI society, where the AI population will create a social economy exceeding trillions of dollars, and Bitcoin and crypto will become important assets in the monetary and economic cycle of this AI economy.
For AI beings, AGI is the brain, robots are the body, and crypto provides autonomous identity and an economic system.
In summary: Don't think about what AI can do for you, but think about what you can do for AI.
How should IQ 50 view AI Agents?
Whoever has big breasts should all in.
At what stage is the Agent project development?
According to cookie.fun data, the total market cap of AI Agents is approximately 18.6 billion USD; about 64% of the total market cap of DeFi projects at 29 billion USD; 75% of the total market cap of GameFi projects at 24.6 billion USD; and 62% of all Layer 2 market cap at 30 billion USD. (Classification of sectors based on Coingecko)
Data source: cookie.fun
Although the market cap statistics are somewhat rough, judging by market sentiment, AI Agents have just passed the halfway point, roughly equivalent to the sun around noon to 1 PM, at its prime.
The chart made by Messari indicates a more optimistic trend. AI Agents are nearing mid-game but the subsequent growth is expected to be even more substantial.
Reference document: https://x.com/Defi0xJeff/status/1873272066834841699;
So, where might the possibilities for future hype lie?
I analyzed the various applications of Agents using the development chart favored by IQ 100 individuals, created by ChatGPT:
In addition to the currently popular Launchpad and Framework models, the directions I personally find promising include: Agent powered crypto exchange: including intent trading, on-chain data analysis, and intelligent advisory-driven exchanges; completely achieving decentralized listings, decentralized asset custody, and decentralized token issuance, with investments and financing all driven by Agents. Different types of Agent exchanges may vary in terms of the degree of Agent involvement in investment decisions, risk preferences, etc.
Agent powered Stablecoin: an evolved version similar to stable projects, using AI to automatically rebase and maintain pegging.
Application of Agentization, similar to tokenization, allows different types of applications plus agents to run, such as games, NFTs, and physical assets; all applications will eventually incorporate Agent services into their core functionalities. (Excessive Agentization can also be seen as a bubble indicator, such as when every chain starts to develop AI Agents.)
To summarize in one sentence, the era of AgentFi has just begun.
