Austin Federa - Rebuilding the internet - TOKEN2049 Dubai 2025

So uh we are going to talk about some of the problems with the internet and what double zero actually is. So we are building a high performance distributed systems network layer. So what does this mean? This is new base layer infrastructure and so we are building a global physical fiber network for high performance distributed systems. Um you can think of this as an alternative pathway for validators, RPCs, state transitions to occur over uh a system other than the public internet.

And so this is one of those things where people don't realize that one of the main limiting factors today in blockchain performance. It's not actually the validator clients, it's not the software, it's the physical fiber links in the ground that things run on. Because the public internet is not really designed for high performance. It's designed for incredible reach. But what it lacks is the ability uh to actually move data efficiently around the world in the ways that are needed for distributed systems.

And today pretty much every large tech company and every large traditional financial firm does not use the public internet. They use private dedicated fiber. Whether that is uh Flash Boys uh if you've read that book back in the day or whether it is uh Meta being one of the world's largest investors in undersea fiber cables, they're not wiring up the Philippines out of the goodness of their heart. They're wiring up the Philippines with internet because they want to sell people ads. Um and every software company gets to a size where they have to start thinking about hardware.

Facebook started by building its own data centers. Now it's building its own undersea cables. So, DZero and Malbec Labs, these are the two groups involved. These are my two co-founders. They have uh tons of experience building high performance systems.

Uh Andrew built these systems uh at Jump Trading where he built out their entire trading and execution network. Matteo uh has owned and operated a carrier and actually worked for Global Crossing back in the day, which was the first company to run fiber around the world and then worked on microwave towers and shortwave transition after that. So we're really bringing like a level of infrastructure that crypto hasn't seen before into this this industry which is dedicated fiber. So why do we need dedicated fiber? Uh you can look at this graph up here and this shows you a measurement of ping time uh for different links around the world.

So this is all roundtrip time from Singapore to LA over various different types of internet connections. And so what you can see here is there's a huge amount of variability. So we start out in the beginning of the morning here where with where we see sort of somewhere between 250 and maybe 340 milliseconds of latency and what that variance is called jitter and jitter is a real killer for high performance systems. It means you have to have bigger cues means data arrives out of order especially if you're waiting for data propagating from all around the world. You move over and you can see here when the public internet actually routed us over a different cable system underneath the Pacific Ocean.

That system is slower as you can see. So we started off with pretty good latency. We jump up to pretty bad latency. We get good for a minute again and then at the end of the day it comes back down. What you see from this though is a huge amount of variability.

So you see these weird spikes all the way up here. Public routers are not designed to handle this, especially when you're talking about high performance blockchain systems. Even companies like Zoom have have private fiber and data lines all over the world just to make sure that video conferencing, which is trivial compared to moving uh data for high performance distributed systems are able to keep up. You might notice this blue line on the bottom and this is0ero. What we provide is a much more consistent high throughput, highquality service.

And so what you can basically see is there's no jitter and a very consistent latency. Now there are moments where the public internet actually beats the zero network, but the goal is not necessarily to always be the fastest possible. It's to provide a consistent experience. And that consistent experience is really what's needed to help validators go from, you know, the maybe 5,000 to 6,000 transactions per second we see on the highest performing networks today up to millions of transactions per second because most validator systems can do this today. Salana has built a validator client that can do a million transactions per second.

A bunch of L2s if you run them in private testing environments can do huge amounts of transactions per second. But you bring them out into the real world and all of this goes away. And why does this go away? Well, partially mainet load is different. But the biggest limiting factor today is not compute, it's network connectivity.

So what are we building? Uh this is the OSI model of the internet. uh and from computer science if you're familiar with this from your commite 101 days uh pretty much everything that's ever occurred in blockchain is layer 5 and above uh we don't really think about data transport much in blockchain we certainly don't think about the network links and the data links and the physical links but this is what zero is focused on we are a blockchain project we are a token project but it's an OSI layer 1 2 and three network so we're going far below the stack of anything that you see in traditional blockchain systems today because our core core use case, our core value ad is fiber connectivity, low latency, high capacity, zero jitter networks. And so you ask like, oh, is is this a new layer 1? Is this a new layer 2?

The answer is no. Uh we call the thing an N1 mostly to not be competitive with anyone. But these are systems way below blockchains where blockchains run today. So a validator network or a centralized sequencer or anything like that can operate on top of the double zero network the same way they may run out of a data center today. instead of talking out over the public internet, they talk through the zero network.

So, it's comprised of two main systems. The first we talked a little bit about is a global fiber network. That's a physical undersea cable link is, you know, this is literally what they look like. It's pieces of big cables on the bottom of the ocean. Uh, and the second is an FPGA system that does filtration and transaction dduplication to provide a defense layer for validator systems.

So there's networks today like Salana that have experienced 80 gigabit per second of DOS attacks before and in a normal system like if you look at like a Google system or something like that if one piece of the network is targeted the others pick up the slack you can't do that in a blockchain right there's only one leader at a time there's only one block builder at a time so if you're boxing out that one connection you can actually disrupt the propagation of these networks and so we have built FPGA systems that can do uh 1 million uh Sorry, 100 gigabits per second of transaction dduplication and signature verification. And so we're throwing out spam. I don't mean spam like low economic value transactions, transactions that the key pairs do not match, right, for signatures or exact duplicate packets coming in. And this can offload a bunch from uh validator networks. So here's a quick overview of how this works.

As you can see there's many sort of different components but at its core data comes in from the internet uh it is goes through these FPGA devices and networking switches and then it can reach validators inside of that network and we are fully compatible with validator sets that are inside and outside the network as well. Um but at its core this is really meant to allow blockchains to hit a million transactions per second or more. So this is the fiber network as it exists today. Uh we are in testn net at the moment and we have a global fiber network that's built out. Uh we it's run by six independent contributors which is one of the key unlocks of this.

As I mentioned private companies have been using dedicated fiber for years but what we haven't been able to do is actually get an amalgamation of different contributors together running private fiber links that appear as one homogeneous network. And this is really key because if someone says, "Hey, we're going to build a one global fiber network for all blockchain traffic," that sounds like a point of censorship, right? And so you need checks and balances on this. You need independent contributors with multiple links coming into multiple countries uh to be able to build out something like this. But really what this is all about is this is making blockchain actually able to compete with Tradfi because we talk a lot about that in the industry.

But the network connectivity layer is too slow. The validator sets are too slow. The not just the transactions and throughput, but the lack of determinism in the network makes it extremely hard for Trifi to actually look at these things correctly. This is why market spreads are larger in crypto. You don't have deterministic time and deterministic latency.

So another highle overview of basically how this works. um we are running through you know J tunnels and BGP routes to link these devices together. I know this is not necessarily the most technical panel but this is the slide deck I've got. Um and so we go through different components of how this thing works. It's orchestrated on top with a0ero ledger which is a blockchain.

Now it's not a blockchain that runs smart contracts. is not a blockchain users deploy anything to, but it is a blockchain that does things like manage routes, data transparency, IP address, reward rate calculations because this is a token project. And so what we've basically been able to do is build uh a system that's normally centralized in the concept of a software controller which runs a network and we've been able to decentralize that component so many people can run those controllers and you can have asurances about how data is moving through the internet. One of the things that we don't think about much in blockchain is you actually have no asurances from the internet service providers that they're not holding back your packets, doing deep packet inspection on them, trying to decrypt them, anything like that. Right?

When we just sort of throw data out into the internet and say, "Well, the internet's magic. It'll get it there. Don't worry about it." This is why 98% of an Ethereum block is just reserved for networking. 98% of that 30 seconds is just for state propagation.

It's not for execution. It's not for anything else with the transactions. It's just to get the data everywhere and then get the votes back in. And so we can move these systems much more quickly if we rethink this this structure. I'm going to go through this a little quickly and skip over a few because I only have about four more minutes.

Um, but FPGA is one of the main uses uh of technology on this network. And so we can offload common tasks from validators. One FPGA system can support many downstream validators because it can do things like sig verify and ddup at 100 gigabits per second. We can also implement QoS prioritization to make sure that votes and state propagation if there's ever uh you know congestion on a network always go through ahead of transactions. Um and it can operate as a bump in the wire design which really means it introduces basically no latency into a system to incorporate it into it.

Uh and multiccast is one of the coolest technologies we are able to do on d0ero. This is sort of like um FM radio for data. Like if you think about it, a a good question is why did it take 15 years for us to figure out how to stream the Super Bowl or live sporting events, right? Well, we've been able to broadcast them on television forever. The difference is every time you're sending your someone's connecting in via like a streaming device to watch a live sporting event, a new copy of the data has to be sent out.

It's not one copy that you're just picking the stream off of like it is with FM radio. Multiccast kind of replicates that experience on uh private networks and you can't run multiccast on the public internet. But what it does is it allows the validator to send a packet once and then the network automatically replicates that data and gets it to everywhere it needs to go. This is an alternative to what you see as Turbine on Salana or Raptorcast on Monad. It's an alternative way to get state propagation moving throughout the world and it's significantly faster especially when it's running over dedicated fiber.

Uh we talked a little bit about the DC ledger, so I'll move past this. But one of the cool pieces that we're going to be able to do starting uh probably middle of next year or so is prioritized routing, which brings the same competitive uh economics of high frequency trading to blockchain. Right? If you think about high frequency trading, what they're doing a lot of the times is they're trying to get the fastest data links possible between two different centralized exchanges, let's say the New York Stock Exchange and the Chicago Merkantile Exchange, so they can make trades between those two. And if you can be smart or fast in Tradfi, you should always pick fast because smart people get things wrong, but it's very hard to get something wrong when you're fast.

Um, and so we're going to bring these types of technologies to blockchain where RPCs and traders and others will be able to operate on the fastest links available between any two different points in the zero network. And that's really one of the core uh principles here is we have to increase bandwidth and reduce latency. And that's the way we get blockchain systems operating from the tens of thousands of low tens of thousands of TPS they can do today up to millions of transactions per second into the future. Um, as we mentioned, today's internet doesn't actually offer security or censorship resistant guarantees. We think it does.

We all think the internet is this like great freedom uncensorable place, but packet loss happens all the time. If you uh, you know, countries have entire filtration systems on fiber coming into uh, their country, right? Right? And so when we think about this in the long run, we have to be able to say to a user, what path did your packet take? And we can guarantee that it actually got from point A to point B.

We can't include you in the block that every validator is a dictator, right? That's where MEV comes in and priority fees and all sorts of things like that. But right now, the connectivity layer, there's really no determinism around that. So this is uh this is what we're building. Uh we have a test net up and running now.

We're launching with Salana support first. We have 30 Salana mainet validators running on the network today. Uh we'll be launching the full product later this year. We're expanding out to uh build many more support for other blockchains and other high performance distributed systems. Uh docs.

mmalbecklabs.com is a bunch of the engineering information how the network works. And you can find the white paper and a bunch more info at.xyz. Thank you.