The upcoming L2 project, MegaETH, has been referred to as “real-time Ethereum” with sub-millisecond latency and the ability to process over 100,000 transactions per second (TPS). The project has just announced a seed funding of over $20 million at a valuation of over $100 million.
The star-studded financing was led by Dragonfly Capital, with participation from Ethereum founder Vitalik Buterin, Consensys founder Joe Lubin, Lido/Flashbots strategic lead Hasu, prolific crypto trader Cobie, and EigenLayer founder Sreeram Kannan, among other prominent figures.
The involvement of these prominent figures has sparked some attention in the market for this project.
Today, we will discuss how MegaETH is innovating on the contemporary Ethereum Virtual Machine (EVM) blockchain to provide industry-leading performance and decentralized assurance.
What Makes MegaETH Special
High-performance alternatives to L1 require their nodes to execute the same tasks without specialization, thus making a fundamental trade-off between performance and decentralization. In contrast, MegaETH creates differentiated roles for nodes with different hardware requirements using Ethereum’s L2 technology.
MegaETH decouples transaction processing tasks from full nodes and creates three main roles for infrastructure operators: sequencers, attestors, and full nodes.
While the actual block production of MegaETH becomes increasingly centralized, the specialized hardware requirements for nodes ensure trustless block validation and provide industry-leading decentralized assurance.
A single active MegaETH sequencer will be responsible for sorting and executing user transactions, eliminating the consensus process during normal operations and propagating state deltas (i.e., changes in the blockchain state) to full nodes through a peer-to-peer network, which will then apply the state deltas to update their local state. It is worth noting that MegaETH transactions will not be re-executed by full nodes to validate the block integrity; instead, they will indirectly verify blocks using proofs provided by attestors.
Even the highest-performing existing L2 (opBNB for BNB) imposes significant limitations on its applications. Despite opBNB’s relatively high throughput target of 100M Gas per second, it can only process 650 Uniswap exchanges per second, compared to modern Web2 databases that can achieve 1M TPS.
In addition, these networks often have “long” block times of over 1 second, which is impractical for applications requiring real-time performance, such as high-frequency trading.
While blockchains often turn to one-off solutions, such as parallelization to pursue scalability, the benefits of this particular approach are limited by the fact that many transactions contain dependencies, resulting in only modest improvements in blockchain speed.
MegaETH does not optimize only a few components of its stack like its competitors, but aims to identify and address a range of issues plaguing existing blockchains by building a new system. This ambitious goal requires pushing node hardware to the limit while maintaining decentralization through specialization, and creating a system that aims to approach the theoretical performance limits of decentralized blockchains.
To achieve this, MegaETH sequencers will store their entire state in memory and become the first blockchain to implement in-memory computing, a key feature of high-performance Web2 applications that should allow MegaETH to increase state access speeds by 1,000 times compared to the alternative solid-state drive storage method used by competitors.
Thanks to a just-in-time (JIT) compiler, compute-intensive applications on MegaETH will see a 100x performance improvement, as the compiler will convert smart contract code into “native machine code” for MegaETH, which is a set of instructions that server CPUs can directly interpret and execute, thereby improving the speed and efficiency of smart contract execution.
Maintaining the Ethereum Merkle Patricia Trie (MPT) is a core data structure representing the current state and related information of all assets, and is a primary bottleneck for all EVM implementations. However, MegaETH is creating a new state trie from scratch that will maintain the complete state trie, compatible with EVM, while minimizing disk I/O operations and storing TB-level state data.
Finally, MegaETH must propagate 100,000 transactions per second to its full node network; an efficient peer-to-peer protocol will deliver state updates from sequencers to full nodes with low latency and high throughput, allowing moderately connected full nodes to stay synchronized at the maximum update rate.
Conclusion
The significant performance improvements of MegaETH compared to contemporary EVM implementations should greatly drive the adoption of L2 performance and ultimately produce decentralized blockchains capable of handling real-world use cases!
While some may view MegaETH as a competitor to the Ethereum ecosystem uninterested in the base layer, the optimizations achieved by MegaETH are entirely through its ability to outsource security and censorship resistance to existing decentralized networks (such as Ethereum and EigenLayer).
