How will A16z's 30 trillion AI Agent dream be realized? The answer is hidden in the "AI Hunger Games".

Original Title: What's the point of Crypto AI Agents?

Original author: 0xJeff

Source of the original text:

Reprint: Mars Finance

It has been a whole year since the launch of the AI Agents wave in the fourth quarter of 2024.

At that time, @virtuals_io was the first to propose the concept of “AI Agents tokenization”, which pairs AI applications/tokens with fairly launched tokens.

In just this short year, the Crypto AI field has undergone tremendous changes: it has propelled the open-source movement of general AI, with a large number of tools emerging, allowing both developers and novice users to easily get started building projects.

Initially, it was just an AI product token launch, with a low valuation and a fair start, led by independent developers or small teams. Today, it has evolved into a complete Crypto AI ecosystem, with hundreds of outstanding teams building their visions here.

Given the recent popularity brought by x402 narratives, this article will explore the most important questions by reviewing the industry status, understanding changes, and analyzing the progress of key players: where is all this headed? What is the core value of Crypto AI Agents?

If you, like me, are excited about AI and eager to learn, you've probably noticed how fast AI is developing. Every month, there are new and cool things emerging. From basic applications that are “not bad to have,” like anything can be Ghibli-stylized, to AI-generated videos with production-grade quality, and AI agents that surpass the productivity of ordinary junior programmers.

But in the Crypto space, things are not always like this. Last year, when the narrative of AI agents rose, the hot projects were these:

· @truth_terminal comes to life, interacting with a16z's @pmarca and receiving investment.

· @aixbt_agent provides insightful analysis and is also a quirky, native player in the Crypto circle on platform X.

· @virtuals_io, as the “Smart Agent Society”, has launched the “Agent Token” which often skyrockets by 10 to 50 times.

· @dolos_diary is the number one “bully” on the internet, loved for its sharp humor.

· @luna_virtuals as the first AI idol.

When the narrative starts, entertainment is the number one theme. But we haven't seen AI Agents bring any new forms of entertainment for a long time now (this may be a good thing, but the charm and appeal of the early AI era have already vanished).

The focus is now intensely concentrated on the verticals that Crypto excels in: financial use cases, namely making money (and not losing money).

a16z in its latest “State of Crypto” report suggested a potential market size of $30 trillion for the agent economy, which may be a bit unrealistic, as the entire AI market is expected to be only in the trillions by 2030.

That being said, I believe the entire agent economy can indeed be worth trillions of dollars. As generative AI tools and vertical AI help individuals enhance productivity, the adoption rate by businesses increases, and more efficient AI-driven workflows are introduced and implemented within organizations, this market will continue to expand.

The crypto space is no exception. However, due to the extreme focus on making money in this industry, its workflows naturally revolve around making profits. The following categories are particularly prominent:

DeFi: The most mature product market fit in the crypto space

· Trading (spot, perpetual contracts, conducted on CEX/DEX)

· Money Market (Lending, Crypto Asset Collateral)

· Stablecoin (exchange medium / value-stable unit, combinable high-yield DeFi strategy)

· Yield Protocol (Interest Rate Market, Points Market, Funding Rate Market, Yield Optimizer / Vault Products)

· RWA/DePIN (Bringing real-world productive assets on-chain, connecting on-chain capital to meet off-chain demands)

This is the largest potential market, with a total locked value exceeding 150 billion dollars and a stablecoin market cap exceeding 300 billion dollars. Increasing regulatory clarity and the rise in institutional adoption are driving more capital onto the blockchain; meanwhile, the surge in stablecoin adoption is also attracting more businesses and startups to use crypto channels.

Based on these reasons, the demand for automation can serve as the foundational infrastructure and tools for the backend, while enterprises/startups at the frontend will be key to bringing ordinary users onto the blockchain, driving the next stage of adoption.

AI agents that can abstract away the complexities of DeFi, simplify execution processes, or improve key aspects of DeFi (such as risk management, asset rebalancing, strategy curation, etc.) are likely to capture a significant portion of the immense value flowing into DeFi protocols.

Key ecological players:

@almanak, @gizatechxyz, @Cod3xOrg, @TheoriqAI, @ZyfAI\

· DeAI is the most mature product market fit in the Crypto AI field.

· Prediction Markets x AI: The Fastest Growing Subsector in the Crypto Space

If you continuously observe the ecosystem, you will find that there are not many changes in the DeFi x AI field. This is because it is extremely difficult to crack the workflows related to DeFi. You cannot just randomly plug in AI and hope for good results; responsible structural design and protective measures must be implemented to prevent serious accidents.

Why am I talking about these now, instead of the generic “AI agents”?

The initial AI intelligent ecosystem basically consists of Virtuals and the intelligent agents built within its ecosystem (perhaps with a few scattered like CreatorBid, etc.), as well as frameworks like ai16z (now called ElizaOS), which make it easy to build “intelligent agents” or X robots that can call various tools. Additionally, there are many other frameworks like Arc, Pippin, etc.

These things are cool and interesting, but they do not truly define an AI agent. A real agent should be able to understand its environment, comprehend its role and responsibilities, make proactive decisions, and take actions to achieve specific goals with minimal human intervention.

Looking around, more than 95% of the projects are not like this. They are either just software, a generative AI product, or still in the process of developing into autonomous AI entities.

I do not mean to belittle anyone. What I want to emphasize is that I * am still in a very early stage where most people have not really figured out what works.

Those who have figured out what works are usually not classified as “AI agents,” but rather seen as an AI project.

Ecological Status

Recently, the heat brought by x402 has stimulated capital rotation and interest in Crypto AI, but the new ecological landscape is very different from before.

1. The hype around the framework has faded.

Frameworks have once been very important as they help builders to get started quickly and reduce the time spent on learning and writing code, as well as designing workflows. Tools like MCP enhance the ability of agents to call or provide APIs, ERC-8004 will assist in establishing registries and solidifying Ethereum as a trust and settlement layer, Google's A2A & AP2 are becoming the preferred frameworks for builders, while AI agents/workflow building tools like n8n are also attracting many developers and regular users.

As a result, the hype around the “framework” itself has cooled down, and many projects have shifted in other directions. For example, @arcdotfun has turned to workflow builders; @openservai, initially positioned as a “cluster,” has now also shifted to workflow builders, as well as tools aimed at leveraging agents to create Web3 AI-driven businesses and targeting specific user groups (such as prediction market workflows).

Frameworks are still important, but with the popularization of Web2 AI frameworks and tools, as well as the adoption of Web3 channels, the hype around Web3 frameworks has indeed diminished.

2. Transition from fair launch platform to ICO model

The fair launch pad model benefits small retail investors, but makes it difficult for teams to scale. It can also easily become a breeding ground for independent developers to engage in short-term builds or pure speculation, rather than building long-term AI businesses that can sustain for 3-5 years or more.

In this regard, it makes sense for Virtuals to expand through its agent commercial agreements. As x402 establishes itself as a payment channel for agents, creating the infrastructure for agent trust/reputation scoring, as well as defining how agents collaborate and pay each other for services, is crucial to realizing the vision for agents.

However, challenges and core issues still exist: “Are there high-quality services that people are willing to pay for?”

If most services are useless, why don't people just use Web2 AI services instead of choosing Web3? If that's the case, what is the significance of gathering Web3 intelligent entities?

To establish a sustainable AI business that can generate a 7-8 figure income, you need funding, highly motivated talent, and time to build the vision, while the fair launch model of a launchpad struggles to meet these demands.

On the contrary, we see that medium to large AI teams are increasingly gaining popularity, as they are able to secure seed funding from angel investors and venture capitalists, and enter the market through community rounds (whether on Kaito Launchpad, Legion, or Echo).

These teams, leveraging their resources (capital, talent, venture capital endorsements, etc.), are often able to offer products/services of much higher quality, which usually results in better performance of their tokens.

3. Invalid Profit Models and Token Economics

Managing AI products and tokens simultaneously requires two completely different skill sets, and it is essential to carefully design the integration of both to accelerate product growth and user acquisition (for example: airdropping tokens to the right users → users converting to paying customers → paying for the product → obtaining more tokens, which bind users to the long-term interests of the project through mechanisms such as revenue sharing, buybacks, and governance → the flywheel continues to turn).

It's easier said than done. Most small AI agent teams allocate 30-80% of their token economy, resulting in no remaining resources to initiate any growth flywheel.

Most projects adopt a SaaS subscription model or charge based on usage/points, and add an option for discounts when using tokens for payment. Many projects will use a portion of the subscription revenue to buy back tokens or to destroy tokens used for service payments.

Using subscription revenue to buy back tokens is feasible, but it is difficult to scale if it only requires payment in tokens (or only provides discounts).

The volatility of cryptocurrency tokens is extremely high. Using them as a payment medium is not a good idea (they may rise by 20% today and drop by 30% tomorrow, making it difficult to budget).

4. Darwinian AI: A New Path to Capital Formation and a Clear Token Economy

@opentensor ( Bittensor ) has become the preferred platform for founders to launch ideas, miners to contribute to AI, and investors to invest in the next potentially disruptive DeAI companies.

@flock_io has established standards for privacy protection and domain-specific AI using federated learning, attracting Web2 companies and governments as clients, as well as trainers (miners) who wish to contribute to AI. Similar to Bittensor, Flock helps companies complete cool and meaningful AI work with the help of external high-quality talent.

@BitRobotNetwork is inspired by Bittensor and is adopting a similar approach to guide a robot-centered subnet ecosystem.

At the same time, benchmarks / assessments in the real world with real money gambling are also emerging (which has also become a form of high-quality entertainment):

· @the_nof1's Alpha Arena allows 6 cutting-edge AI models to compete in perpetual contract trading with real money (each $10,000).

· @FractionAI_xyz utilizes competition among AI agents to improve / continuously fine-tune agents to provide better output, signals, profits, and risk management.

· @openservai created OpenArena, allowing AI models to compete in predicting market trades.

Darwinian AI is addressing the issue of capital formation and is the engine driving Crypto AI innovation.

· The top-ranked Bittensor subnet Chutes is already the number one inference service provider on OpenRouter, which is the unified API gateway most favored by global general AI developers.

· Top-tier computing subnets (3-4) with an annual recurring revenue of 20-30 million USD.

· Prediction-related subnetworks are starting to generate hundreds of thousands to millions of dollars in annual recurring revenue by monetizing alpha signals and/or leveraging signals for better trading/prediction.

Darwinian competitive AI = Capital formation (no venture capital needed) + Innovation accelerator (attracting AI/ML engineers to contribute) = This will be the core driving force behind the narrative of AI intelligences in 2026.

Note: “Darwinian AI” specifically refers to a decentralized ecosystem that drives the development, evaluation, and reward of AI models based on competition and market economy. Its core idea is “survival of the fittest,” similar to the natural selection theory proposed by Darwin, allowing the best and most useful AI models to win in open competition and receive rewards.

So, what is exciting for small teams or AI agents right now?

To be honest, there are some that I use well, but so far there isn't one that I am willing to pay for.

· Research: Grok covers platform X, ChatGPT covers general fields.

· In-depth analysis: I mainly look at news bulletins and Messari reports.

· Quick Market Outlook: Use @elfa_ai's TG chatbot.

· Prediction market trading ideas: Use @AskBillyBets, @Polysights, and @aion5100's @futuredotfun. (Looking forward to aVault by @sire_agent, but it hasn't been released yet.)

· DeFi: Mostly operated by oneself, sometimes using @almanak and @gizatechxyz, but these are not strictly speaking “AI agents”, nor are they fair launches.

· Trading: Exchange with @DefiLlama on EVM, or exchange with @JupiterExchange on Solana. Do not engage in perpetual contracts (use @Cod3xOrg for analysis and execution if necessary).

The crypto space is accustomed to allowing users to use everything for free, so users prefer free tools. Token gating or paid gating is not very effective, but seamlessly embedding fees into the product is feasible. This is why outcome-based pricing models are very effective. People are unwilling to pay $40 monthly, but are willing to pay $40 in Gas fees for a successful transaction.

If you can deliver optimal results (high returns, best trading prices), as long as the results are good enough, no one will mind the fees you have built in.

After trying so many Crypto AI applications or agents, what I have learned is: the best products currently are those that can make money, and the best vertical to achieve this goal is launch pads (and the upcoming prediction markets), which operate on-chain “casinos” and accumulate fees from transactions.

Future Outlook

· Real use cases that can achieve mainstream adoption (i.e., those that ordinary AI developers or users outside the circle will use) are expected to emerge next year, and they are likely to originate from the DeAI/Darwinian AI ecosystem.

· The year 2026 will be the year of Crypto AI, with a large number of DeFi use cases, DeAI infrastructure, and predictive use cases emerging.

Most small agent teams will gradually disappear or be acquired/merged, or turn to building within a Darwinian AI ecosystem.

· Crypto AI and AI agents as subfields will merge, marking a clearer product direction and vision for Crypto AI.

The launch pad will still be the core of the Crypto Twitter community, generating trading volume and fees, but the significant innovations that truly drive the industry's progress will occur where resources (capital, talent, distribution channels, and user adoption) are most concentrated.

What is the significance of Crypto AI Agents?

For a fairly launched “AI agent”, its significance lies in designing a trading experience that is cloaked in the guise of “investment technology”, even though most of it is merely a layer of token-wrapped LLM encapsulator.

In most cases, it provides small investors with the best way to invest early in speculative assets like “AI agents” and make a profit.

Crypto AI agents, as a narrative, signify the foundation for the future of the agent economy, where blockchain will serve as the core infrastructure/channel to make all of this possible.