There is no future for developing infrastructure, is application the future direction?

Author: Adrian; Compiler: Luffy, Foresight News

In every crypto cycle throughout history, the most lucrative returns on investment have been achieved by early betting on new underlying infrastructure primitives (PoW, smart contracts, PoS, high throughput, modularity, etc.). If we look at the top 25 tokens on CoinGecko, we will find that there are only two that are not native tokens of L1 blockchains (excluding pegged assets): Uniswap and Shiba Inu. This phenomenon was first theorized in 2016 by Joel Monegro, who proposed the "Fat Protocol Theory". Monegro believes that the biggest difference between Web3 and Web2 in terms of value accumulation is that the value accumulated by the cryptocurrency base layer is greater than the sum of the value captured by the applications built on it, and the value comes from:

• The blockchain has a shared data layer on which transactions are settled, thereby promoting positive-sum competition and enabling permissionless composability.

• Token appreciation -> Introducing speculative participants -> Initial speculators converted to users -> Users + token appreciation attract developers and more users, etc. This path forms a positive flywheel.

Fast forward to 2024, the original argument has gone through countless industry debates, and there have been several structural changes in industry dynamics that challenge the original claims of the fat protocol theory:

1. Commoditization of block space: With Ethereum block space at a premium, competitive L1s have emerged and become asset class definers. Competitive L1s are often valued at billions of dollars, and builders and investors are attracted to competitive L1s almost every cycle. New "differentiated" new blockchains emerge in each cycle, which excite investors and users, but eventually become "ghost chains" (such as Cardano). Although there are exceptions, in general, this has led to an overabundance of block space in the market without enough users or applications to support it.

2. Modularity of the base layer: As the number of dedicated modular components increases, the definition of the "base layer" becomes increasingly complex, not to mention the value generated by deconstructing each layer of the stack. However, in my opinion, this shift is certain:

• Value in modular blockchains is decentralized across the stack, and for a single component (e.g. Celestia) to receive a higher valuation than the integrated base layer requires its component (e.g. DA) to become the most valuable component in the stack and have “applications” built on it, which in turn have more usage and fee income than the integrated system;

• Competition between modular solutions drives cheaper execution/data availability solutions, further reducing fees for users

3. Towards a “chain abstraction” future: Modularity inherently creates fragmentation in the ecosystem, leading to cumbersome user experiences. For developers, this means too many choices for where to deploy applications; for users, it means overcoming many obstacles to get from application A on chain X to application B on chain Y. Fortunately, many of our smart people are building a new future where users can interact with crypto applications without knowing the underlying chain. This vision is called "chain abstraction". The question now is where will value accrue in this chain abstract future?

I believe crypto applications are the main beneficiaries of the shift in how we build infrastructure. Specifically, intent-centric trading supply chains, with order flow exclusivity and intangible assets such as user experience and brand, will increasingly become the moat for killer applications, allowing them to be commercialized more efficiently than today.

Order Flow Exclusivity

The MEV landscape has changed dramatically since the Ethereum merger and the introduction of Flashbots, MEV-Boost. The dark forest that was once dominated by searchers has now evolved into a partially commoditized order flow market, and the current MEV supply chain is dominated by validators, who capture ~90% of MEV in the form of bids from every participant in the supply chain.

Ethereum's MEV supply chain

Validators have taken most of the extractable value from order flow, which has made most participants in the trading supply chain dissatisfied. Users want to be compensated for generating order flow, applications want to retain value from users' order flow, and searchers and builders want to make greater profits. Therefore, participants who pursue value have adapted to this change, and they have tried a variety of strategies to extract alpha, one of which is searcher-builder integration. The idea is that the higher the certainty of inclusion of the searcher's packaged block, the higher the profit. A large amount of data and literature shows that exclusivity is the key to capturing value in a competitive market, and applications with the most valuable traffic will have pricing power.

This is similar to Robinhood’s business model. Robinhood maintains a “zero-fee” trading model by selling order flow to market makers and taking kickbacks. Market makers like Citadel are willing to pay for order flow because they are able to profit from arbitrage and information asymmetry.

This is further evident in the increasing number of transactions being conducted through private memory pools, which recently hit an all-time high of 30% share on Ethereum. Applications realize that the value of all user order flow is extracted and leaked into the MEV supply chain, and private transactions allow for more customizability and commercialization around sticky users.

With the advent of the era of chain abstraction, I expect this trend to continue. Under an intent-centric execution model, the trading supply chain is likely to become more fragmented, with applications directing their order flow to the network of solvers that can provide the most competitive execution, driving solver competition to drive down margins. However, I expect the majority of value capture to move from the base layer (validators) to the user-facing layers, with middleware components being valuable but with low margins. Frontends and applications that can generate valuable order flow will have pricing power over seekers/solvers.

Possible ways of accumulating value in the future

We are already seeing this happening today, with lending protocols regaining the liquidation bid order flow that would have otherwise flowed to validators by leveraging niche order flow in the form of application-specific ordering (such as oracle extractable value auctions, Pyth, API3, UMA Oval).

User Experience and Brand as Sustainable Moats

If we further break down the 30% of private transactions mentioned above, most of them come from front-ends such as TG Bots, Dexes, and wallets:

Although people have always believed that crypto-native users have a poor attention span, they finally saw a certain degree of retention. Both brand and user experience can become a meaningful moat.

User Experience: Alternative front-end forms that introduce a new experience by connecting a wallet on a web application will undoubtedly attract the attention of users who need a specific experience. A great example is Telegram bots like BananaGun and BONKbot, which have generated $150 million in fees by enabling users to trade Memecoin from the comfort of a Telegram chat.

Brand: Well-known brands in the crypto space can drive up fees by earning the trust of their users. Fees for in-wallet app swaps are notoriously high, but a killer business model because users are willing to pay for the convenience. MetaMask swaps, for example, generate over $200 million in fees per year. Uniswap Labs’ front-end fee swaps have netted $50 million since launch, and transactions that interact with the Uniswap Labs contract in any way other than the official front-end do not incur this fee, yet Uniswap Labs’ revenue is growing.

This shows that the Lindy Effect in applications is consistent with or even more pronounced than in infrastructure. Typically, the adoption of new technologies (including cryptocurrencies) follows a sort of S-curve, and as we move from early adopters to mainstream users, the next wave of users will be less mature and therefore less price sensitive, allowing brands that can achieve critical mass to profit in creative ways.

The S-curve of Cryptocurrency

Conclusion

As a cryptocurrency practitioner who focuses primarily on infrastructure research and investment, this article is by no means intended to deny the value of infrastructure as an investable asset class in cryptocurrencies, but rather a shift in mindset when thinking about entirely new categories of infrastructure that enable the next generation of applications to serve users above the S-curve. New infrastructure primitives need to bring entirely new use cases at the application level to attract enough attention. At the same time, there is enough evidence that there are sustainable business models at the application level where user ownership directly guides the accumulation of value. Unfortunately, we may have passed the market stage of L1s where betting on every new shiny L1 will bring exponential returns, although those with meaningful differentiation may still be worth investing in.

That said, I have spent a lot of time thinking about and understanding the different "infrastructures":

• AI: agent economies that automate and improve end-user experience, compute and inference markets that continuously optimize resource allocation, and verification stacks that extend the compute capabilities of blockchain virtual machines.

• CAKE stack (https://frontier.tech/the-cake-framework): Many of my points above indicate that I believe we should move towards a future of chain abstraction, and the design choices for most components in the stack are still large. As infrastructure supports chain abstraction, the design space for applications will naturally grow and may cause the distinction between application/infrastructure to become blurred.

• DePIN: I have believed for some time that DePIN is the killer real-world use case for cryptocurrency (second only to stablecoins), and this has never changed. DePIN leverages everything that cryptocurrency is good at: permissionless coordination of resources through incentives, bootstrapping markets, and decentralized ownership. While each specific type of DePIN network still has specific challenges to solve, validating a solution to the cold start problem is huge, and I'm very excited to see founders with industry expertise bringing their products to the crypto space.