Original title: Who Makes Money from Agents?
Original author: Jonah Burian
Original translator: Peggy
Editor's note: If Agents really become the next billion users of blockchain, then the more important question might not be "how much transaction volume they will bring," but rather if this world really comes to pass, who will make money?
In the past, whether it was the theory of "Fat Protocols" or "Fat Applications," it was assumed that on-chain users are humans. Humans care about whether the interface is user-friendly, whether the brand is trustworthy, and whether the pathways are convenient, so the application layer can capture value by mastering user entry and transaction flow. But Agents are different. They directly call APIs, have no brand loyalty, and can switch between different protocols, aggregators, and trading venues at low cost.
This means that Agents may rewrite the value distribution logic of Web3. The application layer could shift to "headless" models, opening up wallets, aggregators, and deposit/withdrawal capabilities as APIs for Agents; the protocol layer may also regain opportunities due to Agents bypassing intermediate layers; but in a more radical scenario, Agents could drive the entire on-chain stack towards price competition, compressing the profit margins of applications, aggregators, and infrastructure to near marginal costs.
What is truly worth noting is that Agents do not just lead to higher frequency of existing on-chain transactions, but may create new activities that have not existed before: continuous portfolio rebalancing, machine-to-machine payments, and new types of markets that only make sense in the context of automated high-speed execution.
Therefore, the core question of the Agent era is not simply to judge whether value will flow to protocols or applications, but rather who can ensure that Agents still choose to return here when they have unlimited alternatives. The answer may no longer be UX and brand, but rather liquidity, latency, settlement certainty, or some kind of new business model that has yet to be named today.
Here is the original text:
Many people are imagining that Agents will become the next billion users of blockchain. But few ask the follow-up question: if this world truly comes to pass, who will make money?
Historically, all theories on value capture in the crypto industry presumed that users were human. The "Fat Protocols" theory held that the protocol layer is best at monetizing user engagement. In contrast, my colleagues and I proposed the "Fat Applications" theory in "How to Capture Value" and "The Great Repricing," suggesting that the application layer performs better.
However, Agents change who "users" are. Consequently, existing value capture theories are no longer reliable.
“Fat Protocols” Theory
In 2016, @jmonegro wrote "Fat Protocols." For nearly a decade thereafter, this article has become one of the most mainstream value capture theories in the crypto industry.
Its core idea is that, in the internet era, value primarily flows to the application layer, such as @Google and @facebook, while the underlying protocols like TCP/IP and HTTP capture little value. But the crypto industry is poised to reverse this. Since blockchain data is openly shared, applications will be commoditized; and the protocol tokens required for using the network will capture corresponding speculative value as usage increases. Every successful application will drive demand for tokens. Ultimately, the protocol layer will experience compound growth faster than any application above it.
For a long time, this judgment appeared correct. Bitcoin and Ethereum's market capitalizations exceeded those of any companies built on top of them. This model was valid because, at that time, protocol layers were scarce, expensive, and hard to replace. In 2017, Bitcoin and Ethereum indeed held scarcity; there weren't a dozen general-purpose L1 competing for the same workload. The scarcity of block space was so acute that holding the underlying asset felt like holding a share of rights across all applications that relied on that network.
But now, there are credible alternatives at every layer of the infrastructure stack: multiple high-throughput L1s, dozens of L2s, and competing modular settlement layers and data availability layers. Block space has shifted from scarce to abundant. As cross-chain bridges and aggregators render the underlying chains nearly invisible to users, switching costs have also declined rapidly. Infrastructure has become interchangeable, and interchangeable items ultimately compete only on price. Therefore, the pricing power of protocol layers has waned with the disappearance of scarcity.
“Fat Applications” Theory
By 2026, entities capturing significant economic value will no longer be protocols but applications, such as @phantom, @coinbase, and @Polymarket.
In my view, the reason lies in the fact that the most valuable assets in the crypto industry are user relationships. If you control the user interface and transaction flow, you control distribution; and as long as users encounter a certain on-chain product, you can monetize it: exchanges, lending, staking, minting, deposit/withdrawal channels, and so on. That's probably why investment institutions are so obsessed with new banks (neobanks).
Applications will also push infrastructure towards pure price competition, compressing infrastructure profit margins to near marginal costs. I documented this strategy in "How to Capture Value." The same dynamics are also occurring in the stablecoin space, which I discussed in another article.
Prices reflect this theory. Spencer and I called this shift "The Great Repricing": in this cycle, value flows toward the layer that possesses user relationships.
Why Will Agents Disrupt Everything?
The "Fat Applications" theory presumes that users are human, and humans value user experience, brand, and convenience. But Agents do not prioritize these aspects. They directly call APIs, possess no brand loyalty, and can switch trading venues at zero cost.
When users become software, possessing user relationships is no longer as defensible. The front-end moat that the entire "Fat Applications" theory relies on will depreciate as well.
So, in the Agent era, who will capture value?
Applications Move Towards Headlessness
One possibility for the future is that application layer winners will continue to thrive, only they will abandon the UI.
Wallets and aggregators have built out the most challenging components: integration capabilities for numerous protocols, routing logic, identity, and deposit/withdrawal infrastructure. The natural next step is to open this capability as APIs for Agents, allowing Agents to route transactions just like today's human users via @phantom or @JupiterExchange.
In this world, the "Fat Applications" theory still holds, only it has lost its front end. Those companies that excelled in the human user era will re-platform and become headless infrastructure. We have already seen traditional SaaS companies like Salesforce transforming in this direction.
Protocols Rise Again
Another possibility is that Agents entirely bypass the intermediary layers.
If integration is simple enough, for example, with clear API documentation, standardized RPC, and predictable execution semantics, then Agents have little reason to pay aggregators to do things they could handle themselves.
The advantage of aggregators during the human user era came from user experience and complex routing capabilities. But Agents do not require user experience, and routing itself is a solvable engineering problem, which Agents are getting increasingly adept at handling.
If the future unfolds this way, then the "Fat Protocols" theory will experience a second life.
The Pricing Power of the Entire Stack Will Collapse
Another possibility is that Agents will exert commoditization pressure throughout the whole stack.
They are rational enough. They will always choose the cheapest trading venue, having no loyalty or friction. Applications will lose the UX premium they once charged to human users. Aggregators and infrastructure will also lose their pricing power, as there will no longer be the inertia of human users helping them withstand price competition.
In this scenario, it will be challenging for any layer in the stack to capture significant value. The entire supply chain will be compressed to near marginal costs, with economic surplus flowing to the side that possesses Agents, or to the end users represented by Agents. Crypto will become a utility, and utilities are typically not easy places to make money.
Agents Will Create Previously Unfeasible New Activities
The simple version of this perspective is: Agents will perform tasks that humans have already been doing, just at a higher throughput; even if profit margins are compressed, as long as transaction volumes increase significantly, the overall pie will still grow larger.
But I think there's a more interesting version: Agents will make a class of previously unfeasible activities feasible. For instance, consistently rebalancing portfolios at execution costs below 1 cent; machine-to-machine commerce between Agents; and markets that only become meaningful when pricing and transaction speeds are so fast that humans cannot genuinely keep up.
These activities will not appear in our current framework for observing on-chain activities, because we assume there is always a human participant in on-chain activities.
If this is the real change that Agents bring, then the question is no longer how to distribute the existing pie, but rather how many new economic activities will be brought on-chain, and which layers will be best suited to serve these new activities.
A Yet-To-Be-Named Business Model
In every cycle, we attempt to guess where value will flow, often assuming that known business models will naturally extend into the future. But such assumptions often miss those business models that have yet to emerge.
When the internet was first constructed, no one predicted the emergence of the attention economy. The business model that seems obvious today—slicing user attention into different fragments, auctioning it to advertisers, and letting a company extract a substantial portion of global advertising spend—was very foreign at the time. It only appears inevitable in hindsight.
AI appears to be one of the largest technological disruptions in decades. In a world dominated by Agents, a portion of value capture will likely flow toward some business models that no one is seriously discussing today. Ultimately, the participants capturing value may not be those currently in the market's spotlight.
What Should We Focus on Next?
The most likely outcome is not that one paradigm completely replaces another. Humans and Agents will coexist as users in the crypto industry for a long time, and the value capture maps for the two types of users will not be the same.
As long as humans still interact directly with blockchain, the "Fat Applications" theory remains valid: consumers willing to pay for user experience, brand, and convenience will continue to pay a premium to applications that own user relationships. Meanwhile, the layer where Agents conduct transactions will be governed by a different set of theories—exactly which one depends on how the scenarios mentioned above ultimately evolve.
In my view, for builders, the most worth pondering question on the Agent side is: what will cause an Agent to return to you rather than route directly to the next cheaper alternative?
The answer may not be user experience. It could be liquidity, latency, settlement guarantees, or something else.
At @bcap, we are spending significant time thinking about this question, both in investment committee meetings and discussions with engineering teams. We do not yet have a definitive answer. If you are building products around Agents and have your own insights on value capture in the Agent era, we would love to talk to you.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
