Lei Yang: MegaETH achieves 55,000 transactions per second, Ethereum’s scaling strategy pivots back to layer one, and the challenges of layer two security

MegaETH leverages Ethereum for its superior blockchain execution environment. A stress test on MegaETH achieved 55,000 transactions per second. Layer two solutions that replicate layer one services face security challenges.

Key takeaways

MegaETH leverages Ethereum for its superior blockchain execution environment.
A stress test on MegaETH achieved 55,000 transactions per second.
Layer two solutions that replicate layer one services face security challenges.
Ethereum’s strength is crucial for the entire smart contract ecosystem.
Ethereum’s scaling strategy is shifting from layer twos to enhancing layer one.
Recent changes in Ethereum’s roadmap represent a logical pivot.
Regulatory pressures are pushing some rollups towards centralization.
Ethereum may return to higher transaction costs as activity shifts back to layer one.
Current low transaction costs are due to activity migration to layer twos.
The energy spent on layer two development is essential for Ethereum’s evolution.
Layer twos evolved from sharding, allowing for diversity and experimentation.
The zk-based optimistic fraud proof system ensures transaction correctness.

Guest intro

Lei Yang is cofounder and CTO of MegaLabs, the team building MegaETH, Ethereum’s high-performance Layer 2 with sub-10 millisecond block times and 100,000 transactions per second. He earned his PhD in Computer Science from MIT in 2024, with research on blockchain consensus and networking that achieved an end-to-end system at 80,000 transactions per second from 2018 to 2019.

Why MegaETH chooses Ethereum

Mega uses Ethereum because it allows for the most performant blockchain execution environment.

— Lei Yang
The reason why mega uses ethereum is not because of like you know fanboyism but because it allowed us to build the most performant possible blockchain.

— Lei Yang
MegaETH achieved 55,000 transactions per second during a mainnet stress test.
We did a stress test on mainnet so everyone was able to do it where we did 55,000 transactions per second.

— Lei Yang
The stress test demonstrated MegaETH’s capacity for handling high transaction volumes.
We saw it and that was really awesome… this trains really fast and has unbelievable compute and it can handle crazy levels of activity.

— Lei Yang
MegaETH’s performance metrics are reliable due to the mainnet environment closely emulating actual launch conditions.
If we flip the switch it’s exactly the thing we’re gonna run-in mainnet it’s like the same server same ip address same configuration same admin keys exactly the same.

— Lei Yang

The evolution of Ethereum’s scaling strategy

Ethereum’s scaling strategy is shifting from relying on layer twos to enhancing the capabilities of layer one.
The original version vision is is no longer working ethereum itself needs to scale in the original version basically we were offloading a lot of this to l twos and that’s kind of no longer necessary because the l one is scaling.

— Lei Yang
The recent changes in Ethereum’s roadmap represent a logical pivot rather than a values pivot.
I think of this as a pivot not a values pivot but let me use the p word a pivot in terms of the road map itself in terms of what was originally envisioned with the rollup centric road map and where we are now.

— Lei Yang
The shift towards more centralized solutions by some rollups is a disappointing but understandable response to regulatory pressures.
I think it’s a sad realization of like moloch won almost in some ways right we have to be honest everyone has different incentives… there were stage one roll ups that told me they did not want to become stage two point blank and the reason is because it was the regulatory environment they prefer to be more centralized.

— Lei Yang
Ethereum will eventually see a return to higher transaction costs as activity shifts back to layer one.
I think it is a bit of an overreaction to say that ethereum will no longer need l twos… we will very quickly get back to a world where like there’s $200 transactions on e file one.

— Lei Yang

The role of layer two solutions

Building a layer two that replicates layer one services is technically uninteresting and poses security challenges.
It’s pretty uninteresting to try to build and replicate exact primitives provided by the layer one… it’s actually very hard to build a secure layer two and if all you want from a layer two is kind of just adopt ethereum security as is then I think you are better off like building out layer one.

— Lei Yang
The energy spent on developing the layer two roadmap is not wasted but essential for Ethereum’s evolution.
I really don’t think like the energy we spend on this substrate is is is wasted or something because I think mega is only possible because ethereum was chasing this layer two centric road map for a few years.

— Lei Yang
Layer twos evolved from the concept of sharding, allowing for greater diversity and experimentation in Ethereum’s ecosystem.
The original idea is that ethereum would run run and operate all the shards… why don’t we just delegate it to a few teams so that we get some diversity some competition some experimentation.

— Lei Yang
Layer two solutions have built-in mechanisms to ensure censorship resistance and user exit options.
The guarantees you have first the layer two cannot censor you in the sense that if you are if your transactions are being maliciously excluded from the layer two… you can always go back to the layer one and submit your connection there and the the layer two sequence will be forced to include your connection.

— Lei Yang

Economic sustainability in blockchain

Chain fees are not a viable business model for Mega ETH.
I don’t think like it’s cool that you guys can do 11,000,000,000 transactions per second I don’t think you guys are getting to a sustained level of organic 11,000,000,000 transactions per second anytime soon.

— Lei Yang
Economic sustainability for blockchain systems should prioritize user accessibility over short-term profit from fees.
I think that’s actually short term thinking because the way mega grows is by letting everyone come and use my id for as cheap as humanly possible.

— Lei Yang
The current approach to blockchain revenue generation needs to evolve beyond simply launching tokens.
You kinda need to build some sort of like real business model and we kinda came down to stablecoins before at least that’s that’s what we have now.

— Lei Yang
The mechanism of earning yield from stablecoins on the Mageve platform allows for reinvestment into the chain without harming end users.
What we basically said is hey we’re gonna have like this native stable coin usdm and when applications use usdm we earn the tibo yield for that right that goes into the maggie eve balance sheet…

— Lei Yang

The future of rollups and governance

Stage two rollups require a commitment to immutability in governance logic, which poses significant risks.
Stage two basically says get rid of the security council code is to govern you like for life immutability… you are basically saying one piece of the design of the roll up of the layer two is going to be fixed forever for the lifetime of the universe for the lifetime of ethereum.

— Lei Yang
Reaching stage two in rollups may take longer than previously expected, potentially beyond two years.
I don’t think it’s responsible to claim that we will reach stage two within a year or maybe even within two years.

— Lei Yang
Trusting the security council in stage one can lead to potential misuse of power.
You also have to trust that the security council is not going to misuse your power because yeah within a year we… do see ourselves in stage one but actually it’s not probably not good for for us to claim ourselves to to label like a mega eth as a stage one.

— Lei Yang
AI could play a crucial role in verifying software correctness, which is essential for reaching stage two.
I think the thing that AI does the best is the stuff that is hard to produce but easy to verify… proofs of software code.

— Lei Yang

The impact of AI on blockchain

The next billion users of blockchain technology might actually be AI agents.
One other possibility here that has more recently risen to my attention here is that the next like billion users might actually be ai agents.

— Lei Yang
The poor user experience in crypto for humans may actually benefit software agents.
If you believe that some of the reason we haven’t had more crypto adoption has been the ux is terrible… all the ux problems we have are actually benefits for little software agents.

— Lei Yang
We should start prioritizing agent users in blockchain experiences.
Maybe we should start really prioritizing agent users in some of our blockchains experience.

— Lei Yang
Agents can operate with unlimited energy to experiment with different transaction paths, unlike humans who may give up after a few attempts.
I think to help agents you need… you need the block space to be sufficiently cheap so that they can try and error… agents have immense amount of energy… human do not are just try and errors.

— Lei Yang

The role of proximity markets in blockchain

Microscopic auctions in Ethereum’s block ordering are impractical for systems with very low block intervals.
What it fundamentally breaks for us is because our block interval is so small because in our case the blocking interval is ten milliseconds… running auctions at this kind of fine granularity completely breaks down.

— Lei Yang
We believe that a proximity market model is more effective than microscopic auctions for transaction ordering.
What we’re trying to achieve… we call it the proximity market… you run those options at this kind of interval and decide a bunch of people that.

— Lei Yang
Collocating with the sequencer allows for optimized trading algorithms to operate with minimal latency.
Will have the the seat to collocate with the sequencer and once you do that then they can take whatever algorithm they want to run and run them on these seats which are basically virtual machines that are right next to the sequencer so the quotes we get from our cloud provider is down to one millisecond…

— Lei Yang
The incentive structure should encourage high-frequency traders to physically locate near the sequencer to enhance market liquidity.
The reason for you to have a ten millisecond block interval is for you to build efficient liquid markets… they should come to the sequencer so this I think is both an incentive structure and also I think a more practical way for us to have like efficient priority allocation when you have real time trading.

— Lei Yang

MegaETH’s strategy for fostering innovation

Mega ETH’s approach to fostering its own app ecosystem is essential to avoid redundancy across chains.
We cannot end up in a situation where we just have a bunch of repeat applications that exist on every other chain… what is the point of us spending the past three years of our lives trying to build this thing if there’s literally nothing new that can exist on this train.

— Lei Yang
Mega ETH is actively seeking founders to build unique applications to enhance its ecosystem.
What we did is we basically just started trying to find founders and convince founders to… build cool shit to be frank.

— Lei Yang
There is a significant brain drain in crypto, with many potential founders leaving for AI due to poor user experience.
I think a lot of those founders that could have come and built awesome apps couldn’t do it last cycle because the ux was garbage and this cycle they just left to ai.

— Lei Yang
If we remain credibly neutral and do not actively encourage development, we risk ending up with no new applications.
My bigger fear is the developers are gone… you actually can be incredibly neutral because if you do that we just don’t end up with any more apps.

— Lei Yang

The future of token distribution and ownership

The current approach to token distribution in the market is flawed and inequitable.
I think that was a a problem that we we saw and we thought that it was just not ideal because you know you’d like crypto is unique because a lot…

— Lei Yang
Price discovery has shifted from public to private markets, leading to unequal access for public participants.
I think it was pretty awesome… a lot of price discovery moved to private markets and the end result was like you know public participants weren’t necessarily getting equal access to to assets.

— Lei Yang
The traditional model of ownership in tech doesn’t apply to crypto, where users want to be owners of the applications they use.
I think it’s totally fine that’s not exactly the same case in crypto where people have historically been like I wanna be an owner in the applications I use it’s just it felt very unfair.

— Lei Yang
The approach of allowing users to participate in ownership has proven successful, as evidenced by high participation rates in funding rounds.
We basically just doubled down with that thesis… 80% of echo users tried to participate… ended up you know oversubscribing by like 25x.

— Lei Yang