Hyperscaling Ethereum without fragmentation

We’re megaexcited to announce our investment into Mega Labs building MegaETH, the most performant real-time EVM-compatible Ethereum L2. By employing heavy optimizations to the execution client and leveraging EigenDA, MegaETH builds sub-milisecond latency and over 100k TPS L2. This achieves web2 UX while preserving web3 values of ownership, verifiability and eventual censorship resistance.

The evolution of scaling blockchains

Blockchain scaling has been the most debated topic since the beginning of this industry. There are now L1s, appchains, hyperchains, sidechains, parachains, subnets, rollups, L2s, L3s, and yet noone has been able to solve blockchain scaling and offer consistently good UX. Chains still get expensive to use, transactions fail, bridges get hacked.

There is, however, a consensus on what the endgame of blockchains looks like. All paths lead to the same destination of centralized block production, decentralized validation with data availability sampling and state transition verifying light clients, and anti-censorship protection techniques such as forced inclusions.

All paths lead to a single destination. Vitalik Buterin
All paths lead to a single destination. Vitalik Buterin

The only way how to currently achieve web2-like level pre-confirmation latency, high-troughput and negligible cost while preserving web3 values such as verifiability and eventual censorship resistance, is by building an L2 with a centralized sequencer. Today, however, even the most performant L2 is getting max daily TPS of 62 with peak performances around 187 TPS and all existing L2s still get congested and expensive to use.

All existing L2s still have low throughput. Source: L2beat
All existing L2s still have low throughput. Source: L2beat

There are two main reasons why that is the case:

  • 1, Lack of a high-bandwidth DA layer that is also secure

  • 2, Slow EVM implementations

Even after protodanksharding, Ethereum cannot support high-throughput rollups yet. Assuming that average optimistic rollup tx consumes 200 bytes, Ethereum today cannot support more than 240 TPS in aggregate for all rollups, because the Ethereum blobspace bandwidth is currently only about 50KB/s. It is going to improve a lot after danksharding, but that is going to take at least another 2 years.

Danksharding is going to scale Ethereum DA massively. In the meantime, EigenDA provides an alternative.
Danksharding is going to scale Ethereum DA massively. In the meantime, EigenDA provides an alternative.

EigenDA solves this in the meantime. It is the only secure DA layer that scales horizontally with the number of operators, and provides over 100MB/s at launch this month which can support 500k+ TPS L2s. The L2 scaling bottleneck now shifts from DA capacity to the execution client implementation.

Speeding up the EVM

MegaETH becomes the first real-time blockchain by heavily optimizing across the entire stack:

  • 1, Efficient concurrency control protocols for parallel EVM

  • 2, Novel state trie design that is memory & IO-efficient

  • 3, High-throughput transaction broadcasting protocol

  • 4, Better databases for storing blockchain states & history data

By introducing these optimizations, MegaETH is able to achieve 100k+ TPS with sub milisecond pre-confirmation latency on a single mega-L2.

MegaETH achieves the endgame of EVM and blockchain scalability. It is going to enable the next generation of web3 applications while not introducing unnecessary liquidity and state fragmentation.

MegaETH is the endgame. I don't make the rules.
MegaETH is the endgame. I don't make the rules.

MegaETH is the most likely L2 to scale and dominate as per Vitalik’s Endgame.