Written by: Jiawei @IOSG
Three years ago, we wrote an article about Appchains, prompted by dYdX's announcement to migrate its decentralized derivatives protocol from StarkEx L2 to the Cosmos chain, launching its v4 version as an independent blockchain based on Cosmos SDK and Tendermint consensus.
In 2022, Appchain may have been a relatively marginal technical option. As we approach 2025, with the launch of more Appchains, particularly Unichain and HyperEVM, the competitive landscape of the market is quietly changing, forming a trend centered around Appchains. This article will start from this point to discuss our Appchain Thesis.
The choice of Uniswap and Hyperliquid
Source: Unichain
The idea of Unichain emerged quite early. Dan Elitzer, founder of Nascent, published 'The Inevitability of Unichain' in 2022, pointing out the necessity of launching Unichain based on Uniswap's size, brand, liquidity structure, and the demand for performance and value capture. Since then, discussions about Unichain have been ongoing.
Unichain officially launched today, in February, with over 100 applications and infrastructure providers building on Unichain. The current TVL is approximately $1 billion, ranking in the top five among many L2s. In the future, it will also launch Flashblocks with 200ms block time and the Unichain validation network.
Source: DeFiLlama
As a perp, Hyperliquid has clearly had a need for Appchain and deep customization from day one. Beyond its core product, Hyperliquid has also launched HyperEVM, which is protected by the HyperBFT consensus mechanism, just like HyperCore.
In other words, beyond its powerful perp products, Hyperliquid is also exploring the possibility of building an ecosystem. Currently, the HyperEVM ecosystem has over $2 billion in TVL, and ecosystem projects are starting to emerge.
From the development of Unichain and HyperEVM, we can intuitively see two points:
The competitive landscape of L1/L2 is beginning to differentiate. The combined TVL of Unichain and HyperEVM ecosystems exceeds $3 billion. These assets were supposed to be deposited in general-purpose L1/L2 like Ethereum and Arbitrum in the past. The independence of top applications has directly led to the loss of core value sources such as TVL, transaction volume, transaction fees, and MEV on these platforms.
In the past, L1/L2 and applications like Uniswap and Hyperliquid had a symbiotic relationship, where applications brought activity and users to the platform, and the platform provided safety and infrastructure for applications. Now, Unichain and HyperEVM have become platform layers themselves, forming direct competition with other L1/L2s. They are not only competing for users and liquidity but are also starting to compete for developers, inviting other projects to build on their chains, which significantly changes the competitive landscape.
The expansion paths of Unichain and HyperEVM are distinctly different from current L1/L2s. The latter often first builds infrastructure and then uses incentives to attract developers. In contrast, the model of Unichain and HyperEVM is 'product-first'—they first have a market-validated core product with a large user base and brand recognition, and then build ecosystems and network effects around that product.
This path is more efficient and sustainable. They do not need to 'purchase' ecosystems through high developer incentives but instead 'attract' ecosystems through the network effects and technological advantages of core products. Developers choose to build on HyperEVM because there are high-frequency trading users and real demand scenarios, not because of nebulous incentive promises. Clearly, this is a more organic and sustainable growth model.
What has changed in the past three years?
Source: zeeve
First, the maturity of the tech stack and the improvement of third-party service providers. Three years ago, building an Appchain required teams to master full-stack blockchain technology. However, with the development and maturity of RaaS services like OP Stack, Arbitrum Orbit, and AltLayer, developers can now combine various modules on demand as easily as selecting cloud services, significantly reducing the engineering complexity and initial capital investment required to build Appchains. The operational model has shifted from building infrastructure to purchasing services, providing flexibility and feasibility for application layer innovation.
Secondly, there are brand and user perception. We all know that attention is a scarce resource. Users are often loyal to the application brand rather than the underlying technical infrastructure: users use Uniswap because of its product experience, not because it runs on Ethereum. With the widespread adoption of multi-chain wallets and further improvements in UX, users are nearly unaware when using different chains—their touchpoints are often primarily wallets and applications. When applications build their own chains, users' assets, identities, and usage habits are all consolidated within the application ecosystem, creating strong network effects.
Source: Token Terminal
Most importantly, the pursuit of economic sovereignty by applications is gradually becoming prominent. In the traditional L1/L2 architecture, we can see that the flow of value shows a clear 'top-down' trend:
The application layer creates value (the trading of Uniswap, the lending of Aave)
Users pay fees to use applications (application fees + gas fees), part of which goes to the protocol and part to LPs or other participants.
Among them, the gas fees flow 100% to L1 validators or L2 orderers.
MEV is divided among seekers, builders, and validators in varying proportions.
Ultimately, L1 tokens capture other values beyond app fees through staking.
In this chain, the application layer that creates the most value captures the least.
According to statistics from Token Terminal, in Uniswap's total value creation of $6.4 billion (including LP earnings, gas fees, etc.), the allocation received by protocols/developers, equity investors, and token holders is less than 1%. Since its launch, Uniswap has generated $2.7 billion in gas revenue for Ethereum, which accounts for about 20% of the settlement fees collected by Ethereum.
But what if an application has its own chain?
They can retain gas fees, use their own tokens as gas tokens; internalize MEV by controlling the orderer to minimize malicious MEV and return benign MEV to users; or customize fee models to achieve more complex fee structures, etc.
In this light, seeking the internalization of value becomes an ideal choice for applications. When the bargaining power of applications is large enough, they will naturally demand more economic benefits. Thus, high-quality applications have a weak dependence on the underlying chain, while the underlying chain has a strong dependence on high-quality applications.
Summary
Source: Dune@reallario
The above diagram roughly compares the revenue of protocols (in red) and applications (in green) from 2020 to the present. We can clearly see that the value captured by applications is gradually increasing, reaching about 80% this year. This may, to some extent, overturn Joel Monegro's famous theory of 'fat protocols and thin applications.'
We are witnessing the paradigm shift from the 'fat protocol' theory to the 'fat application' theory. Looking back at the past pricing logic of projects in the crypto space, it was mainly centered around 'technical breakthroughs' and the drive of underlying infrastructure. In the future, it will gradually shift to a pricing method anchored on brand, traffic, and value capture ability. If applications can easily build their own chains based on modular services, the traditional 'rent-seeking' model of L1 will be challenged. Just as the rise of SaaS has diminished the bargaining power of traditional software giants, the maturity of modular infrastructure is also weakening the monopoly position of L1.
The market capitalization of leading applications in the future will undoubtedly exceed that of most L1s. The valuation logic of L1 will shift from the previous 'capturing the total ecological value' to a stable, secure decentralized 'infrastructure service provider.' Its valuation logic will be closer to that of public goods generating stable cash flow, rather than 'monopolistic' giants that capture most ecological value. Its valuation bubble will be somewhat squeezed. L1 also needs to rethink its positioning.
Our viewpoint on Appchains is that due to the brand, user perception, and highly customizable on-chain capabilities, Appchains can better sediment long-term user value. In the era of 'fat applications,' these applications can not only capture the direct value they create but also build blockchains around themselves, further externalizing and capturing the value of infrastructure—they are both products and platforms; serving end users and other developers alike. In addition to economic sovereignty, top applications will also seek other forms of sovereignty: the right to make decisions on protocol upgrades, transaction ordering and censorship resistance, and ownership of user data, etc.
Of course, this article mainly discusses in the context of top applications that have launched Appchains, such as Uniswap and Hyperliquid. The development of Appchains is still in its early stages (Uniswap's TVL on Ethereum still accounts for 71.4%). Protocols like Aave, which involve wrapped assets and collateral and rely heavily on composability on a single chain, are not very suitable for Appchains. In contrast, perpetual protocols that only rely on oracles for external demand are more suitable for Appchains. Also, Appchains are not necessarily the best choice for mid-tier applications; a case-by-case analysis is needed. I will not elaborate further here.
