Exchange Trilemma: The Challenge of Balancing Scalability, Decentralization and Capital Efficiency
Understanding The Exchange Trilemma Trade-off
In the wake of last year's GameStop crisis, and the subsequent 3AC and FTX issues, it has become abundantly clear that opaque systems that lack communication from leadership can quickly spiral out of control.
Lack of transparency has left many in the decentralized finance industry feeling powerless and frustrated, including Jay Jog, a former engineering lead at Robinhood. As a response, Jog has become an advocate for on-chain solutions, which provide trustless and transparent systems. This led to him to initially creating a decentralized Robinhood prior to Sei Network. But after looking into every Layer 1 and Layer 2 and all the other infrastructure that is suitable for building an exchange, he instead began to observe the Exchange trilemma. That theory is explained here and is the holy grail for the space that retail investors should long for.
As of now, every exchange is only able to offer two out of three of the trilemma's key components - decentralization, scalability, and capital efficiency. Let’s dive in:
Scalability
Scalability refers to the capacity of an exchange to handle large amounts of trading volume without suffering from performance issues. This can be broken down into two questions: Does the exchange have high throughput and liveness, and is it easy to provide liquidity?
Scalability is especially important for exchanges that are aiming to become global players, as they must be able to handle large numbers of trades with minimal latency. Centralized exchanges (CEXs) are generally more scalable than decentralized ones, as they have more control over the infrastructure they use and can thus better handle large numbers of trades.
Decentralization
Decentralization, on the other hand, refers to the degree to which an exchange is able to operate without any single point of control. It inquires if there is trust in a single entity. Decentralized exchanges (DEXs) are better at this than centralized ones, as they are able to spread out the responsibility for managing trades across multiple nodes. This makes them more resistant to manipulation or censorship, as no single entity has the power to shut them down.
Capital efficiency
Finally, capital efficiency is the ability of an exchange to maximize the return on investment for its users. This brings up two queries: Is the majority of liquidity being utilized for trading? And are liquidity providers usually suffering losses?
This is important for both centralized and decentralized exchanges, as users are looking for ways to maximize their returns on their investments. Central limit order books (CLOBs) are particularly good at this, as they are able to facilitate trades quickly and efficiently.
What are the existing types of exchanges and where do they fall under these criteria?

CEXs are a popular choice for scalability and capital efficiency, as they are fast, user-friendly, and have low fees. However, CEXs (ie. Binance) are not decentralized, meaning users do not have control over their funds and data. On-chain CLOBs and concentrated liquidity AMMs (CLAMMs) are decentralized and capital efficient, but have run into issues with scalability - orderbooks (ie. Serum) have issues with throughput and liveness, and CLAMMs have issues with simplicity. Finally, constant product/function AMMs (ie. Uniswap V2) are a third option that provide scalability and decentralization, but they’re not capital efficient.
Is there a solution to the Exchange trilemma?
For the past few years, the blockchain industry has been searching for a solution to the 'trilemma' - how to achieve scalability, security and decentralization - but iterating on exchange mechanisms has yet to provide the answer. While the underlying infrastructure has been rewritten to a degree, we believe that a complete rewrite of this infrastructure is going to be the key to solving the trilemma.
Generally, innovations tend to follow app-infra cycles, where first the necessary infrastructure is created to enable new types of applications, and then more specialized infrastructure is developed to support these applications. A great example of this is in the database industry, where Oracle created the initial infrastructure that enabled an explosion of applications, and now we are seeing technologies such as Databricks Warehouse, which is specifically designed to support AI.
Something similar is likely going to happen in the blockchain industry; the development of general-purpose infrastructure, such as Ethereum, led to the creation of decentralized exchanges, and now there is a need for specialized infrastructure that can support these exchanges. When infrastructure for exchanges is built correctly, the decentralized exchange experience is extremely close if not identical to that of the centralized exchange experience.
When it comes to infrastructure for exchanges, there are three main categories:
- rollup, a layer 2 scaling solution that reduces gas fees and increases throughput but has the upper bound of 6K TPS on Ethereum and does not scale
- a shared security approach, which distributes the responsibility of validating transactions across multiple stakeholders but cannot change the consensus level
- sovereign chain, which is an independent chain with the ability to interact and transact with other blockchains and can be optimized to meet specific needs
The Exchange Trilemma can be addressed through leveraging the Cosmos SDK and Tendermint Core. Three major features can be added to improve performance. The first, 'Twin Turbo Consensus', optimizes the block propagation and block processing elements of Tendermint and Cosmos, reducing the time to finality between 300ms and 600ms. The second, Parallelization, increases throughput up to 22K OPS. Lastly, a native order matching engine, built into the chain itself, enables exchanges to create new markets. This has multiple benefits: it helps exchanges scale, prevents frontrunning, and provides a better experience for market makers through order batching. By employing these measures, exchanges can scale whilst maintaining decentralization and capital efficiency.
About Jay Jog
Jay is the co-founder of Sei, the first sector specific L1 blockchain, specialized for trading. He was previously an engineering lead at Robinhood, where he saw the company 10x in employee count and go through major milestones such as the IPO. In 2021, he started building a decentralized Robinhood with his friend, eventually leading to the formation of Sei.
About Sei
Sei serves as the infrastructure for the future financial system to support capital markets, starting with exchanges. As the first Layer 1 specialized for trading, Sei is the fastest chain to finality, optimizing every layer of the stack to give exchanges an unfair advantage. Exchanges are the killer app of crypto, but current Layer 1s hold them back. Most L1s fall into two extremes, general-purpose and app-specific. Sei unlocks a new design space in between as a sector-specific L1. There are currently 100+ teams from Solana, Polkadot, Terra, and NEAR building on top of Sei going into mainnet. The team is backed by Multicoin, Delphi, Tangent, and several MMs like Hudson River Trading, and GSR.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.