EIP-7732: ePBS, Stefan Bratanov, Consensys | ETHSofia 2025

[Music] So I'm going to talk about EIP7732 or EPBS and it has two names. One is enshrined proposal builder separation and it's also known as execution payload block separation. So I hope by the end of this presentation you have a better idea what this actually means. Uh yeah. So I'm going I'm just going to go quickly through the things that I'll cover.

I'll just say a few words about me. I'll go about what is MEV. I'll mention how Ethereum blocks are built today and how Ethereum blocks will be built after EPBS. I'll go through the benefits of EPBS, the current status, and uh hopefully there are some questions at the end. Uh yeah, so my name is Stefan and I'm a senior blockchain protocol engineer at Consensus and I work on one of the consensus layer clients, TEU.

So if you run a note yourself then you're probably aware of the consensus clients, execution clients. So I work on the Java implementation which is TEU and also as part of that I'm considered to be an Ethereum core developer which is around 180 people and it basically comprises of client teams the execution teams g bezel net and the consensus layer clients like prism tech lighthouse the researchers devops testers and basically everyone works on the layer one protocol So I'm going to start with a quick uh overview of what is MEV. So, MV stands for maximum extractable value and it basically refers to the maximum amount of value that can be extracted while block production and this means that uh changing the order of transactions uh excluding transactions, including transactions and some examples are sandwich attacks, front running, back running and basically anything that can be exploited like um like decentralized exchanges. Uh like one example is when there's like a memecoin or an NFT there's a lot of traffic there could be a lot of opportunity for me and validators basically can use the MV boost software to connect to a marketplace of block builders and they basically choose the block with the highest bit and around 90% of the blocks on Ethereum are built using MV boost and that's quite a big number but because it makes sense like there's a lot of financial incentive to run MV boost. That's why every validator is running it pretty much.

And one figure is the MV extracted on Ethereum from September 2022 to June 2024 was 526,27 Ethereum which is around $2.4 billion dollar. Uh and here you can see at the bottom right uh there was for example one validator got lucky and he received 200 Ethereum. So this is doesn't happen that often but uh it's just one of the incentives why validators run MV boost. So a quick introduction how block building works today in Ethereum.

So you have the consensus layer client and the execution layer client which are connected to the meu which is basically user transactions and um every time a validator needs to propose a block if they have chosen to use MV boost which we saw that it's over 90% they connect to MV boost which uh basically connects to the builders via a relay and the relay is like a centralized entity uh which basically acts like a intermediar between the builders and validators. And this is required because um relays and builders they're off protocol. So you need some sort of u uh middleman to do the communication between builders and validators. One analogy in uh life is like brokers who users connect to in order to connect to foreign exchanges. Um so you have builders uh searchers which basically uh submit uh transaction bundles to builders and then builders they submit their bundles to relays and relays and through MV boost validators connect to relays and they choose the highest bits.

Uh so that's how it works at the moment and we'll see in the next few slides how EVBBS changes that. Uh but the important thing here is to remember that relays and builders they are not part of the protocol. So also I'm going to go over a little bit of like how slots in Ethereum work. So if you don't know every 12 seconds there is an Ethereum block and uh the first 4 seconds of the block is currently preparing the block which means that the validator needs to uh code the execution client get the execution block which also means that MV boost needs to be called um and then the relays need to be called as well. So and also the consensus block which contains the validator rewards any withdrawals all of this is uh comprised in these 4 seconds that need to be delivered.

Um usually it happens in the first second because there needs to be voting. So the voting happens in the next 4 seconds. In the next 4 seconds basically validators vote on the blocks that they have received over those four seconds. Um and then because there are millions of validators, you need some sort of aggregation. So the aggregation happens at the 8 second.

Um so that the next block can contain the votes in an aggregated matter. Um yeah, so this is basically how it works now. And you can see that there's a bit of inefficiency because a lot of the work happens in the first 4 seconds where the consens layer client and execution client they need to do a lot of stuff and um it's not very scalable. So yeah I will go now uh how would Ethereum blocks will be built after EPBS. So you can see now that basically relays and MV boost would be obsolete in the design and builders essentially they become part of their protocol and essentially they will be staked entities as well and uh all the payments between the builders and validators it will be done trust trustlessly.

So they can um they now they can trust each other. So they don't require that middleman of the relay uh which is also removes one other centralization part of Ethereum. Uh otherwise everything else would work the same. Searchers they will search the meo and they will submit bundles to the builder. Uh and then all the payments then will be done on the protocol rather than off protocol like how it works right now.

So this is the 12 seconds after EPBS. So you have the first 3 seconds and these times they can be modified. Right now we're doing some research what are the best optimizations like this 3 seconds they could be cut down to really 1 second even because right now the uh all all the validators needs to do is just to create a consensus block which is very fast. Um and it and there will be basically the builders they will submit bits at the beginning of the slot to the validator and the and the consensus block would contain the bit but it wouldn't require any transaction to be run. This is delayed until later in the slot.

So basically the block production is very fast. Uh so that station deadline changes to the to the third second and that could be changed. Um and then you have the aggregation happening in the second six and then also then the builders they need to broadcast the execution. Uh so that means that at the six seconds the builders they should have um they should basically broadcast the payload that the block bit is for. Um so that basically allows uh these two things like the consensus block and the execution block to be done async.

So this allows for a lot more scalability. Uh and at the 9th second there will be uh another voting period where uh basically validators uh vote for the payload timeliness that they make sure that the builder has done its job. Um and and basically at that point the validators they need to run the transactions the execution transactions. So it gives them a lot more time to to do that. Uh yeah.

So I'll go over the benefits of EPBS. Uh so the one so one benefit of EPBS is the the decoupled responsibilities. So it separates the consensus duties the beacon blocks from and the blobs from the execution payload logic. So that means that these things can be done in parallel and uh basically it allows scalability. Uh one other benefit is is a trustless payload market.

So uh validators uh are able to sell their rights payload rights to without relying on any intermediaries relays and the payment guarantees they will be done on the protocol. Uh and this basically allows for more decentralized permissionless MV ecosystem since anyone can take part now being a builder. They don't have to rely on relays. Um I mean I still imagine the builders in the market today would still be the majority but I think there will be there's a more opportunity for more builders to get involved without relying on uh without that reliance on relays. uh also there in there there is an improved network livveness and stability.

So by separating the consensus from the execution means that uh consensus blocks can still finalize even if there is a problem with u execution payloads. Uh I mean no transactions will be executed but at least the consensus layer could continue and finalize. Uh and basically it also means that it the validator windows are extended and the bandwidth usage is optimized. Uh and that allows basically for layer to scaling which is uh one of the goals of Ethereum uh is to basically EPBS provides additional time for blob data dissemination across the network. So in instead of the right now you have like 4 seconds you have like 9 seconds and that means that there could be more blobs across the network and yeah it's a critical improvement for roll up throughput and data availability guarantees.

Um so these were the benefits. So the current status right now of EPBS. So EPBS has been selected for inclusion in the Amsterdam fork which is the fork after Fusaka fork which will happen towards the end of the year. So Amsterdam fork is scheduled for it's not yet scheduled but it's going to happen at some point next year. Um we had a devet with prism which basically uses some of the eps specification and we were able to finalize uh we saw here that there have been some eps stress testing then that means that the gas could be increased um and that shows like how good it is for scalability.

uh and I've included some links to the EIP itself. Uh we have a branch on tech for APBS and the specification is already there for the fork. Um yeah, that's that's pretty much it. [Music] Heat