Powering high-performance distributed systems with DoubleZero ft. Austin Federa

GM, GM. Welcome to what's going to be another fantastic episode of DevNTell. So if you didn't know, DevNTell is a 30-minute podcast held every week, allowing founders, hackers, and anyone in between an opportunity to come on the show and showcase what they've built. And today, I'm ecstatic to welcome Austin Federa, who's the co-founder of DoubleZero. So if you didn't know, DoubleZero is a network of dedicated high-performance links engineered to bring low-latency networking to everyone, everywhere. So if you stick around for today's episode, you'll get to meet Austin, you'll learn all about DoubleZero, and maybe even a little bit more. All right, let's get into it.

GM, GM. Welcome to the show, Austin. I'm ecstatic to have you on today, man. Hey, thanks for having me. My pleasure. I know you are a very busy man, so thank you so much for taking the time out of your day to come chat with us. I guess before we get into all of the great things related to DoubleZero, perhaps would you like to give a brief introduction about yourself?

Yeah, certainly. So, as you mentioned, I'm one of the co-founders of DoubleZero. Before that, I spent the last four years or so at Solana Foundation, running strategy for the organization. Prior to that, I worked for Bison Trails, which was acquired by Coinbase and became Coinbase Cloud. But if you go back far enough, I was undergrad political science and environmental science, and lots of research work on transboundary water management treaties and that sort of thing. So, nothing at all crypto-related, but in retrospect, it's kind of a blockchain degree because it's all like messy governance where you can't force anyone to do anything.

Yeah, I mean, governance is a very interesting sub-topic of this space, that's for sure. But yes, that's very interesting, man. I didn't know you had that type of background. I guess, was it something in particular that kind of inspired you to take the leap into the tech side of the world?

Sort of. I mean, I think I really liked the undergraduate work I was doing, but there wasn't much impact on the world. There was this kind of famous study on NBER papers, the National Bureau of Economic Relations. It's this nice clearinghouse website where basically all of the economic papers get posted. And the most meta thing an economist did was an analysis of the NBER papers themselves. And they basically found that the average paper has four authors and is downloaded six times. And you're like, 'Oh man.' So every author downloads it, and then not even their mom is reading it. And it was just like, 'Oh, that's pretty rough.'

And so for me, I wanted to do stuff that was going to have more of a positive impact in the world. And I actually spent the first two and a half years of my career working in journalism for National Public Radio. And then I sort of got to a place where I was seeing people that were much older than me—like, I was 22 at the time and there were people who were like 32—that were getting promoted. I was up against them for jobs, and they were obviously getting the job because they had 10 years more experience, even though the job doesn't require it. And I was like, 'All right, this is a dead end.' Back in 2014, like Kara Swisher and Walt Mossberg and the ReCode conference and All Things Digital, that was—they were the best interviewers at the time around this stuff. This is before Acquired launched, which now is like the default best technology business podcast by far. But like Peter Kafka was doing stuff back then too. And I just sort of was like, 'Oh, this feels very interesting, this idea of technology and tech startups.' And so I started working for a robotics company in Cambridge, Massachusetts, was there for about 18 months before making the jump into financial technologies. Started working for a company that turned out to be on the verge of bankruptcy, which I did not know when I joined them. But their last attempt to build something was a borrow-lend token on Ethereum in 2017. And this is before we even had the word DeFi, basically. And it failed, didn't work. Company kind of did a hard pivot and laid everyone off, but I got hooked on crypto at that point. Then went to work for Republic, launched Republic Crypto with them, and a bunch of stuff like that.

Wow, that's certainly quite the journey. And yeah, I guess all those experiences led you to the Solana ecosystem. And you were leading strategy there, if I'm not mistaken, right? Yes. Would you want to just double-click into that experience for a bit? Really curious to hear about it.

Yeah, so Bison Trails got acquired by Coinbase. Coinbase was not going to be the right place for me. And I was starting to interview with a bunch of L1s and L2s. And at the time, this was the end of 2020, the prevailing wisdom at the time was that sharding was going to solve everything. This is back when Ethereum was going to have sharding on the actual L1 itself. And you know, whether it was Polkadot or Near, everything was sharded. And I sort of had this thing in the back of my head, because I'd been working on ETH 2 staking infrastructure at Bison Trails, about like, how the heck is DeFi going to work in a sharded ecosystem? Isn't the whole point of the Ethereum L1 composability? And aren't we fundamentally breaking composability with sharding? And everyone sort of would be like, 'You're not an engineer, don't worry about it. This is a solved problem. Bridges are going to solve this. We'll have state proofs.' And look, it's 2025 and we don't even have that stuff now.

But I was interviewing with a bunch of folks and I had offers from some major L1s and L2s. And this guy I used to work with, Ben Sparango, had just left Multicoin to join Solana because Multicoin had done a big investment in there and they kind of placed him into that organization. And he was like, 'You should come talk to Solana.' And I was like, 'I don't know, I worked with 50 protocols at Bison Trails, we never talked to anyone at Solana even though we technically supported them.' He's like, 'No, no, no, come talk to Raj and Toly.' And I remember sort of going into this being like, 'I don't really know much about Solana, not sure I really want this job,' all that sort of thing. And my first question to Toly was like, 'So I've been working on a bunch of ETH 2 stuff and a lot of really smart people tell me sharding's not going to be a problem, but what do you think?' And he sort of laughs and he goes, 'Sharding's never going to work. You need one global state machine with competitive markets. Ethereum was right.' Like he literally was like, 'Ethereum was totally right. You want everything in one state. I don't know what they're doing now.' And you know, this was this moment where I'd love to say I was like, 'Oh, Solana is going to be a $120 billion dollar project someday.' But really the entire thesis for me was like, I'm going to learn way more at this company taking an orthogonal bet to the market than I am going to work for a company building the exact same thing as everyone else and trying to fight it out in the trenches. And I joined Solana, it was like $1.25. Ten months later, it was $250. Ten months after that, it was $8. It was a pretty crazy time to be at the organization. But I started off just working on product and marketing. Some of the first stuff I worked on was the stake pool program for Solana, one of the six people that came up with Metaplex as a project, a bunch of that type of stuff before moving into the head of comms role after Breakpoint, which was sort of like an accidental role. It wasn't what I intended to be. I had no formal comms training or anything like that, but it was just what we needed at the time. And then in 2023, moved back into the strategy side of things.

Wow, that sounds so exciting and quite the experience. I bet, I mean, it's really great to see Solana continue to take these steps and evolve, and I think it's become pretty much a staple in the DeFi ecosystem today. And I think it's just going to keep growing. There's always chat of who's going to eat whose market cap, but we're not here to chat prices, we're here to chat cool tech, right?

Yeah, it's interesting too because I think we've seen a lot of really interesting application layer innovation. And as much as like Hyperliquid and Lyra are like L2s or L3s or their own L1 depending on who you ask, they're really single-application environments. And the Cosmos app-chain thesis is alive and well, it's just not on Cosmos. And I think it's really interesting that like this idea of application-specific L2s seems like maybe there's something there to it. What we haven't seen is any non-application specific L2 or L1 or L3—again, what is Hyperliquid, an L1 or an L3? Depends if you're looking at it from the perspective of USDC or if you're looking at it from the perspective of state—but we haven't really seen any general-purpose L2 get any significant amount of traction in performance. There's obviously stuff happening on Base, but none of it's performant. And so I'm really interested to see how MegaETH and Monad, if they're able to break into that market, because there's actually nothing competing with Solana today on speed and performance in a global sense. It's all on specific applications. And so I obviously still believe very strongly in the Solana vision of the world, but I think it's about to get much more interesting, and probably very good for every consumer out there, because what you want is competition, and there really hasn't been serious competition for Solana yet.

Yeah, competition brings out the best in everyone. I'm with you, I'm keen to see how some of these new entrants into the ecosystem perform. Application-specific chains have always been something in my mind. I always thought, you know, it's very hard to do a one-size-fits-all thing, right? So we'll see how it plays out, but I have hope. Segueing now, let's get into all things DoubleZero. So, what is DoubleZero? What's your elevator pitch of it?

Yeah, DoubleZero is a new type of project. There's nothing quite like it in crypto. What we are building is functionally a new internet designed for high-performance distributed systems like blockchain. And what that means is, all of blockchain today, all of blockchain that's ever existed, runs on the public internet. And the public internet is a great, awesome resource, we all know and love and use it, but real companies don't use the public internet. And that's kind of like a weird thing to say, but you think about like all these companies that are large public internet companies, whether it's Facebook or Meta or Apple or Google, YouTube, trading firms, financial technology firms, even eBay. They don't use the public internet. They all run private dedicated fiber networks because the public internet is not fast enough to suit their needs. And that's kind of a crazy thing when you think about it: that we basically have built a whole bunch of parallel corporate internets that are designed for one use. Meta is actually the largest investor in subsea fiber anywhere in the world. And so the thesis of DoubleZero was, if blockchain is truly going to compete with centralized systems, not just TradFi, but especially TradFi, we really need to be rethinking the fundamental foundation. And the fundamental foundation is not like Ethereum or Solana or Bitcoin, the fundamental foundation is like, what physical cables are you running on? Let's get down to OSI layer one and two here, let's not be playing around with layer five and six. And from that perspective, crypto has always been playing with a major handicap. That the trading firms will always outrun you. This is why so much volume happens on centralized exchanges—it's because these financial firms are able to use the same types of predictable technologies that they use on Wall Street, on Binance or Coinbase or Bybit or Upbit or OKX. And so our goal is to accelerate blockchain by a decade in the next two years. And this is really about saying the validator clients now are fast enough. You go back to 2020, you ran Geth locally, you couldn't get it that fast, the software was not built for performance. And we're in a different place now. Solana has validator clients that can do near a million transactions per second. I think Aptos and Sui have similar synthetic benchmarks. MegaETH and Monad are coming out. We're in a very different situation now where the limiting factor is the connectivity layer, it's not the quality of the software anymore.

So you would say, if I understand correctly, DoubleZero is kind of like a turbo boost for any protocol looking to basically speed up their system.

Yeah, I mean, look, there's all sorts of opinions in this space on analogies. But you can think about it sort of as the difference between the highway system and county roads, where highways have a little bit more limited egress and ingress points, they have entry ramps, but once you're on it, you can go—if you're in Germany, you can go 150 miles an hour—and you can't do that on a country road, even in Germany. And so really what we're dealing with here are the biggest problems for high-performance systems on the public internet, which are latency and jitter. So latency is just, if you grew up playing video games, it's your ping time, it's your connection time to a server. Jitter is a little bit more complicated, but jitter is basically the standard deviation of latency. If you take a bunch of measurements, how much those measurements change over the course of a measurement period. I'm in New York, sometimes it takes me 45 minutes to get to JFK, sometimes it takes me two hours and 45 minutes to get to JFK. That variance is jitter. And jitter is a real killer for high-performance systems because you have to cache things. Data arrives out of order, you have to throw it in a buffer, you have to basically wait and then go grab it again. And in modern computing systems, data flow for high-performance architecture is really the most important component of throughput nowadays. And every time you're caching things, every time you have to throw something into RAM—let alone, people don't realize how slow RAM is. It is astronomically slow compared to L1 or L2 cache. And L2 cache is astronomically slow compared to L1 cache, and cache is astronomically slow compared to just piping the data in a serialized format directly through the CPU. So the Firedancer team would do some profiling of a Solana transaction and they would say basically like, it takes X number of nanoseconds to process this transaction. What's actually going on? Turned out 92% of the time was just waiting for data from the RAM. And so you look at these systems and you say, okay, like on this very micro level, I can make this system astronomically more performant by just saying, you know, okay, maybe we need to think about hot state and cold state differently. If we're going out to NVMe, everyone was like, 'Actually, maybe we've done something bad here, because now if you're fetching data from NVMe, it's so much slower. The bottleneck is like the PCIe bus.' And we're getting pretty deep on this stuff, but this thing scales up perfectly fine to the network layer, too. Which is like, if all your optimizations can come from optimizing the flow of data between the NVMe, the RAM, and the CPU, well, we can a million-fold that performance increase just by looking at the networking layer and saying, 'Okay, if we can link everything with high bandwidth—high-capacity pipes that are also low latency and very predictable—we can now design systems that are able to move much more quickly because they have to buffer less data further up in the stack,' for lack of a better term.

Interesting. And yeah, don't be afraid to get technical on the show, we are developer heavy. So all good there. And I guess from a consumer perspective, who's able to utilize the DoubleZero tech? Is it chain agnostic or is it specific to Solana?

Yeah, so the core technology is chain agnostic. The system that allows you to join the network, we basically have to build some software to talk to each network. So for example, what we do on Solana is there's an on-chain program called Passport which basically queries the leader schedule and says, 'Are you a validator that's been asserted in Gossip? Are you staked? All these things. If so, great, you're permissioned to join the network.' And what we basically have is validators join the network via GRE tunnels—like short-haul GRE tunnels that are usually like 50 meters, right? You're usually just using the tunnel to get from your server to the DoubleZero device in the same data center as you. But you know, the way this basically works is they pay a percentage of their earnings to operate on the DoubleZero network and there's sort of an on-chain balance. So there's a few of those components that we need to build for new chain support, but DoubleZero is very much a neutral base layer infrastructure that's agnostic for whatever people want to build on top of it. And so we're in conversations with a bunch of L2s, a bunch of non-Solana projects too. But the reason we started with Solana is it's the hardest problem. It's the highest number of nodes and the highest data rate of any protocol. And so if we can get the DoubleZero network working well and prove out value on Solana, it's a much easier pitch to go to everyone else and say, 'Look, we did this for the most complicated protocol. Like, we can definitely help Hyperliquid, we can help Monad, we can help MegaETH, all these other types of protocols too.' And so, there's some people that are sort of like, 'Oh, it's a Solana project.' It's not a Solana project, Solana just has the biggest problem at the moment with state propagation and data movement.

We also can do things like multicast. And multicast is this really awesome technology that, again, trading firms have used for decades but is not available on the public internet. And multicast is hardware acceleration for packet replication. So you send the data once or twice because you want a redundant feed, and the switches in the network basically replicate the data as close to the source as possible. So if you have 99 validators in London and one validator in New York, on a unicast network like the public internet, there's 99 copies of the data that that validator has to send across the Atlantic. On a multicast network, they just send two, and then the switch at the end replicates that data on to all the validators on the other side. So it's a huge efficiency savings from a time perspective, because it takes time to send all that data out, and it's a huge bandwidth savings.

Wow, yeah, that's amazing. And yeah, you're right, that problem is pretty much underlying in most of the networks today, and it's awesome to see something like DoubleZero out here trying to tackle arguably one of the hardest problems out there. So good on you guys. Yeah, just in the interest of time, I just want to bounce to speaking around the 2Z token itself. There was some big news around that from the SEC. Like, what is the utility of the 2Z token? What is its purpose on the network?

Yeah, so the 2Z token is really integral to the functioning of the network. What it does is it manages payments on the network, which is a really big component obviously of any blockchain network. And the way the token works is validators pay in a percentage of their earnings to operate on the network. That's expressed in the 2Z token. About 50% of those rewards are burned. About 50% are distributed out to the network contributors. We have 11 entities that contribute fiber to the network today. DoubleZero Foundation, Malbec Labs, we do not run any of the fiber, just like the Ethereum Foundation does not run any of the Ethereum validators—at least I don't think they do, definitely a tiny amount if they do run any. And so that is basically the way that this token works. There's no inflation on the token. And so that's like on the validator side. RPCs and other kind of Web2 servers that might want to run on it, they have more of a pay-per-request model on the network. And they also have the ability to pay priority fees—not today, we're building this out—but priority fees basically to say, 'If I have an arbitrage event and I really want to make sure my data moves over the fastest path possible...' because we have about 70 different links. Maybe this is a good moment—if you go to doublezero.xyz/dashboard, you can see in real-time all of the fiber links that are contributed to the DoubleZero network. And so you can kind of see this giant mesh of the world. And we'll have resilient fiber between different locations, but those speeds are not necessarily the same. Maybe one path across the Atlantic is 81 milliseconds, one is 64 milliseconds, one is 88 milliseconds. And if you're a trader and you're like, 'Look, I've got a $100,000 arbitrage opportunity,' the same way that you'll add a MEV boost fee or a Jito tip to your transaction, you're going to add it to the network layer because most of the latency in these systems today is not intrinsic to the system. The latency is actually getting from your trade server to wherever the block is being built in the world.

Beauty. And yeah, you're free to share your screen if you want to showcase that dashboard. Sure, let me. Yeah. So it's pretty interesting, I think what we've...

All right, is that loading for you? Great. Yes, sir. So yeah, this is doublezero.xyz/dashboard. We've got about 30% of Solana running on this today. You can actually go in and you can click on a specific fiber link and you can see, London to New York City. Okay, this is a 10-gigabit link, here's the latency, here's the jitter, here's its utilization. If we go back and we kind of look at some of the telemetry metrics relative to the public internet, you can see there's areas where we beat the internet by 70%, 80% on some of these routes. Obviously, the longer the route, the better the improvement. Then there's some areas where like Dublin to London, the public internet's pretty great, we're only beating it by half a percent. But then you look at some of these other routes like New York City to Toronto at 17%. And so you can kind of go through this and you can look and say, 'Okay, where is DoubleZero providing the most improvement?' and a lot of times it's over long-haul routes.

Wow, that's quite impressive. The gain in speed... and you can see all the validators that are operating on the network too. We've got some big names on it as well. So it's pretty interesting to see. I think this map really helps people contextualize what actually even is this network. And you can kind of zoom in and you can actually click on this and say, 'Okay, like, how many validators are actually running in Frankfurt?' and these are the different data centers. We have a level of granularity of where a validator's actually run that you don't really get from any other type of dashboard. This is like a massive problem in Ethereum where no one even knows how many machines are running Ethereum because the Gossip protocol just doesn't support pinging that many different devices and stuff. Because things are connected to DoubleZero, we can see in a different way what that looks like.

That's really cool, man. And yeah, folks who are watching or listening today, we will link that dashboard for you in the description so you can take a browse around. And just coming back to the 2Z token, I want to get into this because I think it's really important. I believe in late September or early October, there was some big news that kind of came out of the SEC regarding the token and how it's not a security. Just want to get your thoughts on that and why you think it will be so important for setting a standard for basically any other DePIN network or similar network in the future.

Yeah, so for a lot of us crypto-native folks, the idea that something is or isn't a security feels like, 'Okay, we've been through this, we also sort of accept that there's some risk.' That's not the case for traditional companies. And so for DoubleZero, where our network contributors, they're not people in their basement with a Raspberry Pi, they're companies that have access to global fiber networks. And what that means is they usually don't have sophisticated securities lawyers on staff, let alone anyone familiar with crypto. And so the no-action letter that we got from the SEC covered two major components of the protocol, but the most important one for us was the reward payments out to network contributors. The SEC said those payments are not securities transactions. And so what that means is we can go to an AT&T or an NTT or something like that, and we can say, 'Hey look, you're going to get rewarded in tokens. Yes, tokens are weird and volatile and you have to figure all of that out, but they're commodities.' And you know exactly how to deal with this, this is the same as any other type of commodity you might have on your balance sheet like capital equipment or something like that. And from a tax treatment, from a securities treatment, they know exactly what they're doing at this point. And that was a really big unlock for the protocol. And so we were really—it was awesome to work with the SEC on this. We found them to be not pro-crypto, not anti-crypto, just very clearly trying to evaluate, 'Hey, does this thing run afoul of current securities law interpretation?' which was really great to see because that was not the mood of the SEC a few years ago. I was going to say, if you had told me in 2022 that in 2025 we would receive a letter from the SEC and the commissioner would issue a statement about my project, I would have been very concerned.

Yeah, it's very interesting times we live in, but I think it's for the better that that gray area that used to exist in cloud—this part of the tech world—is starting to kind of lift, and there's more clarity around what you can do, what you can't do. And I think it's for the better. Maybe some people would argue against that, but I think knowing exactly which lane you can stay in is very helpful from a business standpoint.

Yes, absolutely. There's still a huge gray zone, but the gray zone is a lot smaller than it used to be. Exactly. Hopefully one day it will just disappear, but still some work to do on that yet. But yeah, huge congratulations on that, and I think it sets a really great standard for anyone in a similar situation going forward. I will say, I really do hope that this is a framework that others can look at and say, 'Hey, if we build something using a similar economic framework to this, we're much more likely...' I mean, look, no-action letters are not case law, and they're certainly not acts of Congress, but they're the best we can get nowadays. And so hopefully this provides more guidance to folks on how to build compliant products.

Well said, well said. And yeah, I see we're coming up on time. I just got three more questions for you before we let you go. I know you're super busy. Firstly, how can people get started with DoubleZero? For folks who are watching today, they want to kind of dive in, play around with the protocol, what's the best way for them to do that?

So doublezero.xyz has all of our info there. If you are looking to contribute fiber, there's a whole button to get in touch with us on that. If you're a validator on Solana, there's an easy connect-in portal, it's all self-service just like anything in blockchain. But if you're a protocol developer working at another protocol, interested in kind of learning 'Is this interesting for us? Could we integrate with this?', we'd love to talk. DMs on Twitter are a great way to reach me or anyone else on the team. Tom Warner runs blockchain BD for us and is also a great person to reach out to. So yeah, I think we are at a place now where we are basically ready to start adding additional protocols to the DoubleZero network. And so if you feel inspired, we'd love to talk.

Exciting, awesome man, love to hear it. And I guess in a similar vein, are you guys hiring? Are you looking for any particular roles? We have a developer-heavy audience here, so...

Yeah, so there's a careers page at doublezero.xyz as well, it's way down in the footer where all good job links are buried at the bottom. There's a few roles we have open at the moment. We're always hiring network engineers at Malbec Labs, so that's a really big component now. We're about to open a job for a smart contract engineer to join our team. We've got a few people working on that already, but I think it's a pretty fun, interesting job because it will have to engage with multiple different types of networks. And so it's not like, 'Oh, this is just a Solana job' or 'just an EVM job.' Like, you're going to have to support Aptos and Sui and Solana and RISC-V architecture and all sorts of interesting stuff. So it's a nice interdisciplinary dev role that also really focuses on low-level performance optimization.

I mean, folks, Developer DAO is chain-agnostic, protocol-agnostic. So if that sounds like something up your alley, definitely apply at DoubleZero. And I guess in terms of the roadmap for DoubleZero, is there anything that you can share in terms of what you guys are thinking of doing in 2026? Yeah, I mean, it's really expanding out the fiber network, adding support for new networks and new network systems, both on-chain systems and off-chain systems. We've had a ton of inbound interest from all sorts of other types of protocols outside of crypto too, which is always really interesting—like AI systems and storage systems and these sorts of things. The prioritized data is a major one for us, though. I think that trader-focused type of product is something we'll definitely be rolling out in 2026.

Exciting, awesome man, love to hear it. And just to end on a bit of a fun note, I've seen a bunch of these IBRL black sweatshirts floating around. How do people get those? Do they have to go to conferences and find your swag station? Yeah, Eden is the keeper of the swag, so you can bug her on Twitter. I think we also do—I don't know if they're there—we've got some stuff at shop.doublezero.xyz. So there's some stuff there, I don't know if the IBRL hoodies are still there. We only made like 200 of them, 300 of them, something like that. So when they're gone, they're gone. So if it's interesting to you, bug Eden or there might still be a few available online. Awesome, good stuff man. Yeah, with that, unfortunately we've come to time. I think we could have probably gone another hour or so on a bunch of different things, which is great. But Austin, thank you so much for taking the time out of your day to come chat with us today. Really exciting to see all the work you guys have done with DoubleZero and all the great stuff that's going to come along in the future. Awesome, appreciate it. My pleasure. And with that, I just want to wish everybody a very happy Friday, happy weekend wherever you may be, and we will catch you back here next week for another great episode of DevNTell. Till then, have a good one folks. Cheers.