 |
The hype machine building up around Ethereum layer-2 MegaETH claims it’s the answer to Ethereum’s scaling and sentiment problems.
Prioritizing speed over decentralization, the new L2 will apparently feel as seamless as using a Web2 app or a centralized exchange, and run an order of magnitude faster than all the other rollups and Solana combined. In short, the aim is to “Make Ethereum Great Again.”
It’d be easy to dismiss this as pure hyperbole, but MegaETH is backed by some of the biggest names in the industry, including Ethereum co-founders Vitalik Buterin and Joe Lubin, EigenLayer’s Sreeram Kannan, OG influencer Cobie and even Solana maxi Mert Mumtaz.
Due to the launch of its public testnet in the coming weeks, the EVM-compatible chain has reportedly hit 15,000 transactions per second (TPS) in private testing while using 60x more gas (1.5 gigagas) than existing Ethereum rollups like Base. That gas figure is set to double when the proprietary parallelization engine is switched on, and the aim is to scale the L2 all the way up to 100,000 TPS and 10 gigagas.
MegaETH also claims execution times of 10 milliseconds or less, which is well under the 250ms of the fastest existing rollup, Arbitrum.
Centralized scaling: MegaETH’s controversial approach
To achieve such heroic levels of scaling, MegaETH has cut a lot of corners — the team prefers the term “optimizations” — by removing some of the key components that make blockchains, blockchains.
Unlike other L2s and rollups, it has no plans to ever decentralize the sequencer, which allows it to use a bunch of high-end equipment to ramp up throughput. Data storage has been outsourced to the less secure and as yet non-battle-tested EigenDA rather than Ethereum, and the L2 is missing plans for a crucial component of the “don’t trust, verify” ethos, which is consensus (where all the participants in the network agree on the state of the blockchain). All L1s require this, but L2s have the ability to opt-out.
Delphi Digital researcher Prasad Mahadik explains that decentralized consensus slows blockchains down. “The speed of light acts as an ultimate bottleneck and the more nodes there are dispersed, the more time it takes to get to consensus,” he says.
“MegaETH, being an L2 has only one sequencer, and doesn’t have constraints of hardware requirements,” Mahadik says.
MegaETH vs. Solana
It’s hard to see MegaETH as anything else but Ethereum’s L2 answer to Solana, which has been eating Ethereum’s lunch for the past two years with its faster, cheaper chain that uses expensive equipment in data centers rather than Ethereum’s maximally decentralized network, which is designed to be able to run on Raspberry Pis.
Crypto influencer Bread, the project’s public face and head of growth, says that Ethereum’s focus on decentralization above all has left L2 leaders like Base’s Jesse Pollak desperate to scale faster.
“He’s literally pleading with Ethereum developers and saying, guys like, we’re getting our ass kicked here by Solana,” says Bread, who says SOL’s prioritization of scalability was influential in MegaETH’s development.
“We have made design decisions that allow us to scale vertically, which to me is kind of similar to Solana,” he says. “We want to be super big, we want to be super fast, and because we’re doing it as an L2 construction as opposed to doing it as an L1, it allows us to make even more opinionated decisions on how to enable some of those upper bounds of performance.”
Being an L2 arguably allows it to make even more sacrifices to the scaling gods than Solana can, as it still retains the security of Ethereum’s base layer.
“We make some key trade-offs to allow us to be even better than Solana — purely better than all L1s in some regards.”
But has MegaETH got the balance right? And if these tradeoffs are able to attract a critical mass of users, what will that mean for competing L2s like Abritrum, Base, Optimism and Starknet that are currently constrained by Ethereum’s data blob capacity? Will they be forced to abandon their own plans to decentralize, or to use alternative data availability (alt DA) in order to compete with an L2 running 500 times faster?
Arbitrum and Starknet both declined to comment for this story.
How MegaETH began
The project was founded by Stanford computer scientist Yilong Li three years ago after he read Vitalik Buterin’s 2021 blog post about how a hypothetical 10,000 transaction-per-second chain with centralized block production could become “acceptably trustless and censorship-resistant.” He was also inspired by a 2022 post from Plaintext Capital about using data availability layers like Celestia to scale.
“It became clear to me that, from an architectural perspective, it was theoretically possible to improve blockchain performance by orders of magnitude,” Li tells Magazine. “However, it seemed that no one was pushing this idea hard enough at the time.”
He teamed up with Lei Yang, who’d collaborated with Sreeram Kannan on some of the early work that led to EigenLayer, and MegaRollup, later MegaETH was born.
The project raised $20 million in seed funding mid-last year from those big names mentioned earlier along with Dragonfly Capital, Figment Capital and Robot Ventures. It raised another $10 million from retail on Cobie’s Echo platform in December, and $30 million more from the recent Fluffle sale.
EigenDA turbocharges MegaETH
Most Ethereum rollups are limited to some extent by the fact they post data to Ethereum “blobs” and plan to decentralize their sequencers.
Unfortunately, Ethereum’s blobs are already full, and even after they double capacity in the Pectra hard fork, there will be little room for the L2s to scale aggressively. Even after Peer Data Availability Sampling (PeerDAS) is enabled with Fusaka, the rollups won’t be anywhere near tens of thousands of TPS.
“Such high bandwidth is unavailable on any DA platform other than EigenDA,” says co-founder Yang. “Plus, EigenDA provides strong security backed by EigenLayer restaking and the forking EIGEN token, along with a clear roadmap to even higher bandwidth.”
Bread, however, concedes that EigenLayer’s security has never been properly tested in an economic attack and that using it increases trust assumptions.
Is MegaETH just a big data center then?
Given it doesn’t need a network for consensus, it might sound as if MegaETH is just a very large data center.
And while there definitely is a big data center involved, there are also three peer-to-peer networks checking the data. There’s a network of full nodes that update their states according to the sequencer and a network of provers to double-check validity to help keep everything on the straight and narrow. They don’t all validate everything, however.
There are also replica nodes that receive state diffs (stripped-back info rather than all the transactions) from the sequencer and update their states. As they don’t need to re-execute everything like normal, the hardware requirements for the nodes aren’t as high.
“The combination of those two things, working in conjunction, means everyone can participate with lower hardware requirements, and they can decentrally validate that the sequencer is telling the truth.”
The sequencer also has to post collateral, which can be slashed if it acts maliciously and there’s a back up sequencer to keep things running if the first one fails.
MegaETH for high-frequency trading
Mahadik says a huge advantage of using a single sequencer is it enables market makers and high frequency algorithmic traders to colocate their gear there.
“This gives it two orders of magnitude lower latency compared to L1s and there’s a huge market that is not served by DeFi currently.”
Mahadik believes this will unlock DeFi and decentralized exchanges (DEXs) for institutional players, especially when combined with a favorable regulatory environment under the new administration.
“This points towards MegaETH’s benefits of having high chances of capturing this market.”
tldr: watch this trading speed simulation showing solana vs megaeth pic.twitter.com/ehKaHZlYzC
— macki (@_mackinac) February 10, 2025
MegaMafia built cool apps for MegaETH
MegaETH is EVM compatible, meaning that existing apps like Uniswap and Aave can port over. However, the plan is to bootstrap the ecosystem with new apps that take advantage of the high speed and throughput.
Collectively these builders are called the MegaMafia.
There are high-speed decentralized exchanges like GTE and perps from Valhalla, but some of the more novel apps include Euphoria, which gamifies trading by allowing users to tap on a chart to place bets against its movement. Sweep is a live streaming platform like Twitch where users can bet on what might happen, while Autonomous World Engine (AWE) offers up a 3D Roblox world where you can buy coins by bumping into something akin to a loot box.
Bread argues that giving users cool new stuff to play with will help support the price of ETH, in the same way the price of SOL went up as people bought it to swap for memecoins.
“Now that you enable unique applications that are going to be denominated in Ethereum again … it gives people a reason to acquire Ethereum.”
Future of Ethereum L2s: Decentralization vs. performance
MegaETH isn’t the only chain prepared to sacrifice the ideal of decentralization on the altar of scaling. Based on the Move language, Sui and Aptos are also targeting TPS in the hundreds of thousands per second.
BNB Chain last week announced an upgrade that will see it scale to 100 million transactions per day and Hyperliquid promises 100,000 trades per second or higher.
Eclipse uses a version of the Solana Virtual Machine to operate a high throughput L2. CEO Vijay Chetty says the chain hit a daily high of 269 million transactions, sustained an average of 3,142 TPS, and over two days processed more txs than Optimism has in its entire history.
Chetty concedes that “Eclipse makes similar tradeoffs to MegaETH.”
“MegaETH has one sequencer that enables it to minimize latency, whereas Eclipse chooses to have a more decentralized sequencer set that ensures short-term censorship resistance. Both chains have highly performant sequencers built for performance, allowing the network to be more performant than L1s.”
Will other L2s be forced to sacrifice decentralization too?
The Ethereum ecosystem has maintained a laser-like focus on maximizing decentralization, and Ethereum-aligned L2s have been under pressure to decentralize their sequencers and, more recently, become based or native rollups to help make the ecosystem more interoperable.
Ideally of course, MegaETH will just be one L2 choice offered on the spectrum of decentralization.
But the potential danger with MegaETH is that if the chain attracts enough users with high throughput and becomes the most dominant L2, it’s conceivable other L2s could take a commercial decision to abandon decentralization plans so they can compete.
“I do think this might happen in the shorter term similar to how optimistic rollups were embraced compared to ZK even though ZK was superior,” says Mahadik. However, he points out that ZK tech is now catching up and is likely to be integrated by OP rollups, and he expects similar developments in decentralized scaling.
“Similarly, L2s might use a single sequencer until improvements on decentralization with performance come closer,” he says.
That may take a while, of course, but “in the longer term, decentralization would be very important.”
Andrew Fenton
Off The Grid’s success shows ‘invisible’ blockchain is the winning play
Gaming firms are split on whether blockchain should be the main character in Web3 games.
Read more
Web3 gaming won’t exist in 5 years, $656K for best crypto game pitch: Web3 Gamer
Crypto VCs reveal why last cycle’s games sucked, Mega Cricket League review, plus $656K up for grabs in crypto game Shark Tank: Web3 Gamer.
Read moreRipple says SEC lawsuit ‘over,’ Trump at DAS, and more: Hodler’s Digest, March 16 – 22