The greatest crypto investments historically in each cycle have been made by betting early on new base layer infrastructure primitives (PoW, Smart contracts, PoS, high throughput, modular etc.). If we look at the top 25 assets on Coingecko, only two tokens are not L1 blockchain native tokens (excluding pegged assets) - Uniswap and Shiba Inu. The first rationalization of this phenomenon was by Joel Monegro in 2016 who proposed the “fat protocol thesis”, which argues the biggest difference in value accrual between web3 and web2 is that crypto base layers accrue more value than the collective value captured by applications built on top of them, whereby the value is derived from:
Blockchains having a shared data layer in which transactions are settled on, which fosters positive-sum competition and enables permissionless composability
The positive flywheel of token appreciation > onboarding speculative participants > conversion of initial speculators into users > users + token appreciation attracting developers and more users and so on.
Fast forward to 2024 - the original thesis has gone through numerous industry debates, alongside several structural changes in industry dynamics that have risen to challenge the original claims of the fat protocol thesis:
1. Commoditization of blockspace - perhaps driven by the realization of the infrastructure premium and the fact that successful alt L1s end up becoming “category definers” (e.g. Solana for high throughput, Celestia for Data availability etc.) that command multi-billion-dollar valuations, builders and investors alike are drawn to the alt-L1 trade almost every cycle, where there are new blockchains with insert differentiation each cycle that get investors and users excited and end up becoming ghost chains (cough Cardano). While there are certainly exceptions, this has overall resulted in a market where blockspace is overly abundant without sufficient users or applications to power them.
2. Modularization of base layers - With an increasing number of specialized modular components, defining “base layer” becomes progressively complex, let alone deconstructing value that accrues to each layer of the stack. However, in my opinion, what is certain about this transition is that:
The value in a modular blockchain gets fragmented across the stack, and for an individual component (e.g. Celestia) to command a valuation greater than an integrated base layer it would require their component (e.g. DA) to be the most valuable component of the stack, and to have “applications” (Modular blockchains) built atop that generate more usage and therefore fees than integrated systems;
Competition among modular solutions drive cheaper execution/data availability solutions have further squeezed fees lower for users
3. The progression towards a “chain-abstracted” future - Modularity inherently creates fragmentation across ecosystems, which has resulted in cumbersome UX. For developers, this means choice overload for where to deploy applications, for users, this means jumping through multiple hurdles just to go from app A on chain X to app B on chain Y. Luckily, identifying this problem doesn’t take a genius - and we have plenty of smart folks building towards a future where users interact with crypto applications without knowing the underlying chain(s) that power the application. This vision has been coined as “chain abstraction” - which is a thesis I’m excited about. Now the question being where will value accrue in a chain abstracted future?
I argue here that crypto applications are the main beneficiary of the shift in the way we are building infrastructure. Specifically, an intent centric transaction supply chain with orderflow exclusivity and intangibles such as UX and brand will increasingly become moats for killer apps and allow them to monetize much more effectively than the current state.
Ethereum’s MEV landscape has changed significantly since the merge and the introduction of Flashbots and MEV-Boost. A once dark forest dominated by searchers has now evolved to a somewhat commoditized orderflow marketplace, where the current MEV supply chain is heavily dominated by validators, who are receiving ~90% of MEV generated in the form of bids up each player in the supply chain.
Validators capturing a majority of extractable value from orderflow leaves most players in the transaction supply chain unsatisfied - users want to be compensated for generating orderflow, dapps want to retain value from their users’ orderflow, searchers and builders want greater margins. As such, value hungry participants have adapted by experiementing with numerous strategies to extract alpha, one of which is a searcher-builder integration - the idea being having higher certainty of inclusion for searcher bundles will result in higher margins. There’s extensive data and literature that shows the exclusivity is the key to value capture in a competitive market, and applications with the most valuable flow will hold pricing power
This has also been seen in retail equity trading via brokers like Robinhood - which sustains “0 fee” trading by selling orderflow to market makers, and monetize in the form of capturing rebates. MMs like Citadel are willing to pay for this flow due to their ability to profit via arbitrage and information asymmetry.
This is further evident in an increasing number of transactions going through private mempools, which recently hit an all time high on Ethereum of 30%. Dapps realize that value of all user orderflow is being extracted and leaked up the MEV supply chain - and that private transactions allow for more customizability and monetization around sticky user flow.
On the eve of @ethcc 2024, Ethereum's Private Transaction share has nearly doubled – surpassing 30% of all confirmed transactions for the past two days. This trend appears to be accelerating. Is this good or bad overall? My take: it's both. 
This is a trend I expect to continue as we move towards a chain abstracted future. With an intent-centric execution model, the transaction supply chain likely becomes more fragmented, with applications gating their orderflow to networks of solvers that can offer the most competitive execution, which drives solver competition to squeeze margins lower. However, I expect the majority of value capture to be shifted from the base layer (Validators) to the user facing layer, with middleware components being valuable but operating on low margins - i.e. frontends and apps that can generate valuable orderflow will hold pricing power over searchers/solvers.
We are already seeing this play out today with niche forms of orderflow utilizing application specific sequencing such as oracle extractable value (OEV) auctions (e.g. Pyth, API3, UMA Oval), serving as a means for lending protocols to recapture liquidation bids that would have gone to validators otherwise.
If we further breakdown the 30% of private transactions mentioned above - a majority come from front-ends like TG Bots, Dexes, and wallets:
For how much crypto-natives have been touted for having low attention spans, some level of retention is finally being seen. Apps illustrate that both brand and UX can serve as a meaningful moat -
UX - Alternative forms of frontends that introduce net new experiences from connecting your wallet on a webapp unsprisingly draws attention from users that demand specific experiences. A great example are telegram bots like Bananagun and bonkbot that have generated >150M in fees(
https://dune.com/whale_hunter/dex-trading-bot-wars
) to allow users to trade memecoins in the comfort of their alpha telegram chats.
Brand: Established brands in crypto can upcharge based on earning users’ trust. Wallet in-app swaps are known to have notoriously high fees but serve as killer business models that allow users to pay for convenience. Metamask swap for example, has generated >200M in fees annually. Lastly the icing on the cake - Uniswap Lab’s front-end fee switch has netted a cool 50M since launch. Transactions that interact with uniswap labs contracts in any way other than the official frontend aren’t charged this fee - yet their revenues have only increased.
This demonstrates that lindy effects in applications exist just as much, if not more than in infrastructure. Typically, new technology adoption (crypto included) follows some sort of S-curve, and as we move past early adopters into more of the majority - the next wave of users will be less sophisticated and therefore less price sensitive, enabling brands that are able to reach critical mass the ability to monetize in creative (or simple) ways.
As someone who primarily focuses on researching and investing in infrastructure, this post is not by any means aimed to dismiss infrastructure as an investable asset class within crypto - rather a shift in mindset when thinking about net new categories of infrastructure that enable the next generation of applications serving users higher up the S-curve. New infrastructure primitives need to illustrate net-new use cases on the application level to make them compelling. At the same time, there's sufficient proof of sustainable business models on an application level, where user ownership directly leads to value accrual. We unfortunately have likely passed the phase of the market where punting on every new shiny L1 will bring exponential returns; though those with meaningful differentiation may still deservingly capture mindshare and value.
Instead, a non-exhaustive list of “infrastrcuture” that I’ve been spending more time thinking about and understanding include:
AI: from the agent economy that automates and improves the end-user experience, compute and inference marketplaces that continue to optimize for resource allocation, and the verification stack that extends the capabilities of blockchain VM compute.
The CAKE stack: many of my above points suggest that I believe we should build towards a chain abstracted future, and the design choices for most components within the stack are still huge. With infrastructure enabling chain abstraction, the design space for applications should naturally grow, and could lead to the discinction between app/infra become less binary.
DePIN: For a while now I’ve believed DePIN is crypto’s killer real world use case (#2 behind stablecoins) and this has not changed. DePIN leverages everything that crypto is already good at: permissionless coordination of resources via incentives, boostrapping marketplaces, and decentralized ownership. While there are still specific challenges to solve for each specific type of network, validating the solution to the cold start problem is already huge, and I’m super excited to see founders with industry-specific expertise bring their product on crypto rails.
I guess it should go without saying that if you’re building anything that resonates with the above, please reach out as I’d love to chat. I’m also all ears for feedback or counterarguments, as frankly investing would be much easier if I’m entirely wrong about this.
“The most exciting applications on
Ethereumblockchains are probably ones we haven’t even thought of” - Vitalik Buterin, 2014
Adrian
123