Justin Drake

And so the question is, can the ZKVM be performant enough to crunch through one gigabyte of data per second?

And this is still yet to be determined based on future optimizations.

But what we do know for sure is that Ethereum will have the DA to have the one gigagas per second for the L1, plus a handful of other L2s.

Yeah, so I have two parts to this, to the answer here.

The first part is actually, it's even more ambitious than the way you framed it.

And the reason is that the change to the cryptography is so invasive that it's essentially almost a rewrite of the consensus layer, at least.

And so if we're going to rewrite the consensus layer, we might as well properly rewrite it and put all of the goodies and clean up all the technical debt.

And does that remind you of anything?

That's the Lean Consensus Project, where we're basically bundling together multiple rewrites, including the single-start finality with the upgrade to post-quantum.

So yes, it is a very ambitious project.

In some sense, we're starting from a clean slate.

and building something amazingly beautiful and simple and efficient and provably secure and all of the good things.

The good news is that in many ways, starting from scratch is simpler.

Because you don't have all of this technical debt and we can rewrite the spec to be as minimal and simple as possible.

And this is where the terminology lean comes from.

We want to have maximum simplicity where we want to have the whole state transition function basically be a thousand lines of Python code that some sort of smart high schooler can just read and

Right now we have testnets, sorry, devnets for Lean Consensus.

And the specs are so easy to ingest that we've seen about 10 teams all implement them, join, start joining the devnets and do so without even contacting the Ethereum Foundation.

So the barrier to entry is relatively low.

And we're in this crazy world where AI development means that you can basically just to a large extent, vibe code your clients.