Wall Street Speed Meets Blockchain Scale: Pyth & D0ero with Austin Federa & Mike Cahill at KBW2025

Hey, hey, that was a great last chat. I'm looking forward to this one as well. Um, I was just up here a couple panel panels ago talking about ZK and future of L2s and then they put me on a panel with you guys and it's like totally on the other end of the spectrum. What

an upgrade?

I' I'd say so. And so, uh, you know, this one is kind of I'm I'm engaging a different part of my brain here. Um, and we don't usually start this way, but I'd actually like to start with a couple of intros because both projects here and both of you guys are really really deep, thoughtful. There's a lot going on here within each of the within each of these projects. So, maybe we could just start very quickly some intros and we'll we'll get into it.

Sure. Uh, Mike Cahill and I'm the CEO of Dora Labs and we are um the kind of the building team behind the Pith Network. Pith Network does effectively two things. So one thing is we bring data on chain in a secured uh fashion so that it can be used in uh blockchain applications. Um people call that in oracle.

Um so we bring lots of data. Historically it's been the data that's generated within network. So think prices of stuff from crypto markets to equity markets and FX. Um and a couple months ago we did a partnership with um this little group called the US Department of Commerce and we brought on chain the um the GDP. And then a week later we started bringing on a whole bunch of other data.

So um non-farm payrolls um unemployment data. So that all comes through a secured system which then makes it usable on a chain, right? So you need to have the cryptographic proof, you need to have some trust list um trust minimized uh network and you need have some transparency. The other thing we do is we generate firstparty data and this is something that we're just starting to roll out now. We call it Pith Pro.

I'm going to announce it later today. Don't tell anyone. Um, and it's effectively a subscription data business that's going to compete with the likes of the Bloombergs and NASDAQs of the world to be able to provide a lowcost but more comprehensive alternative to financial market data. It's a huge problem. Financial market data is a $50 billion revenue a year business.

And so we're basically dipping our toe into that. So that's sort of the background of uh of us.

Awesome. I'm Austin Federa. I'm one of the co-founders of D0ero. We are basically building a new internet. Um, which is a pitch you've heard some version of a lot in crypto.

Uh, but this time it's actually real. Uh, and what I mean by that is we're we're doing stuff at the lowest level possible. This is the OSI layer 1 2 and three part of the stack. This is physical fiber. These are physical cables under the oceans, over mountains that deliver internet uh to data centers all over the world.

And so what we have teamed up with is our contributor network of 12 different organizations and it's growing pretty rapidly which actually are contributing fiber to this new network. And why do we need a new internet? Because the current one is way too slow. It wasn't built for high performance distributed systems. And it's actually one of the main things holding back blockchains today from truly competing with traditional markets and traditional finance.

It prevents Salana from going to a million transactions per second. But it also means that every time you're trading on a crypto exchange or you're trading uh just as a user even on a centralized exchange, you are going over the public internet. And what does Wall Street do in traditional finance? They go on dedicated connections and dedicated lines. And that means they're significantly faster and really if we are going to better integrate uh the traditional finance world and the blockchain world, we need to be using the same tools and technologies as them.

Why has this not happened before? Well, those systems that exist are highly centralized. It is Jump building its own private network, DRW building its own private network, Jane Street building its own private network, Amazon building its own network. These are singleparty networks and they're anti-et effect networks. And so what we've done with double zero is contri uh multiple contributors come together to create one homogeneous network that is able to outreach and outperform these traditional private networks and offer significant speed and performance advantages over the public internet.

Yeah. Which speaks directly to the name of this panel which is IBRL. If if you're unfamiliar with what that means it's increase bandwidth reduce latency. And so I think at a very high level, we'll get into exactly how double zero may be doing that. But but first I'd like to understand from an Oracle's perspective, why is it so important that we increase bandwidth, reduce latency, what does this enable?

How are you able to then elevate and improve what you guys do onchain?

Yeah, so the the evolution of of Pith really starts in a gap where the previous oracles were were were kind of slow and they were slow for a few reasons. Um, one of them was design. And so basically you would only care about data being pushed on chain at certain intervals because you basically need to maintain a scarcity on around the block space because it was expensive and it was you know these are the old days right so this is in Ethereum land where it's 12 second blocks and so if you're spamming the network every time you know that that will become both consumptive and also expensive. So Piff Network actually first launched on Salana to be a you know high-speed Oracle and Salana was pushing the boundaries 400 milliseconds you know that's real blockchain fast um at least it was back in 2021 um and so the evolution of the network was Salana mainet pith at one point was Austin probably knows this number better than I do but it we were over 75% of the block space of Salana for many many months and um and and the intention was always to go crosschain right this was just the layer to be able to provide the transparency so that people were able to get comfortable with how the data was um was originated. And so we ended up creating our own version.

It was a fork of Salana is called Pithnet. Um and so still 400 milliseconds but now without any of the centralization risks and any of the um consum consumption charges that would come across with pushing thousands and thousands and millions of transactions um thousands of times per per second million transactions um per day. Um but that was not that good you know for for kind of the cutting edge of what people want to do on chain which is like more um of the high-speed perpetuals markets or derivatives markets. So earlier this year we created a new version um Pithnet v2 um which brings the update times down to 1 milliseconds. We rearchitected the entire blockchain from scratch to be able to cut out all the things we don't need.

And that's really important because if we're going to envision a world where we're bringing things into a decentralized sphere, you can't be trading off technology too much. It's just never been the case that people are willing to put up with so much of a trade-off, right? Like it's always that the incumbent technology wins until the um the the the new technology is as good at that but then also does something else, right? like you needed to have Amazon to be as good as turning a page in the Sears catalog and also does all this other stuff. People weren't willing to do that back when it was like you know loading a loading 10 images over a minute.

So I think that for us it's the most important thing is be able to continue to bring the speed down so that the applications that are possible to go take on the NASDAQs or the Binancees or whoever it is that they want to build forever whatever opportunity has been lacking um in the centralized traditional finance space like that needs to happen.

Yeah. Very cool. So I I didn't realize Pith was its own chain as well. So um that that explains why this is really really important. Awesome.

I'd love to get into how we make this a reality. Um like I said couple panels ago it was L2s and its replication and they're really investing heavily into ZK as a result of uh or as a way to kind of decrease the the replication of those transactions. You guys are taking a fundamentally different approach going to the deepest levels of our internet infrastructure. this is deepin at the most fundamental layer. So give us kind of like the the speedrun of of generally how you're you're going about creating this new internet.

Yeah. So there's a lot of really interesting exciting new technologies. Um but at the end of the day, everything that's happening with zero knowledge is not really going to improve market efficiency or market speed. It makes it much easier to prove that something is true after that thing has occurred, but that's not really what we're focused on for the most part here. Um, if you think about what's going on in a traditional matching system, whether that's NIS or NASDAQ or Binance, these are systems where you have multiple bits of data coming from multiple locations and really most of the uh, if you think about it, money that's made in the traditional market by high frequency trading firms is made on time delta arbitrage.

it is do I know where the market's going to move either because it did move in Chicago and I'm reflecting the trade in New York or I have some analysis that tells me where I think the market's going to move and it let's say we have a confidence that in 35 milliseconds the market's going to move in a specific way and you want to trade well if your execution time on that trade and by by execution time like crypto focuses on the wrong things all the time we have all this focus on what's the latency of you know trade confirmation that is a infantestimally small portion of the latency involved in trading. Um that that model works for a centralized exchange. It's like how quickly does it take when my trade server sends data to Binance for Binance to say confirmed. Well, if my trade server is sitting right next to the Binance server, the time in flight is very small. That's not the case in blockchain.

Whether it's Salana or Apdos or Sooie or pretty much any network out there, including a centralized L2, that data has to rush from wherever it is right now to wherever the block is being built. These are actually markets where the time in flight is significantly longer than the confirmation time for the blockchain now. And so if you're saying I'm trading from New York and the the node is building a block in soul at the moment, well that can be 600 milliseconds on the public internet on something like 0ero you're cutting that time by 2/3 when you're looking at a system like this and it's much more dependable in terms of those trades actually happening. And so the problem we're seeing in blockchain now is actually quite similar to one that traditional finance solved three decades ago. And this is sort of if you go back to spread networks and the rise of high frequency trading.

What folks realized was if you can be smart or fast, you should pick fast. It's very hard to be wrong when you're fast. Smart people still get things wrong. And so as we push the bounds of uh these systems and get them closer to traditional finance performance, the execution layer does still matter. There are meaningful differences in different execution environments, but one of the main differences now is how quickly can you get data around the world?

How quickly can you propagate those changes out? Because even in a centralized sequencer model, everyone is so focused on how quickly can we get data into the sequencer. And that's great, but once a block is produced at the sequencer, whether it's in a centralized system or whether it's something like Salana, that data, that market state change has to propagate out from that core. And that is your limiting factor in how fast markets can move. If I don't know what happened in the last block, I can't trade on the next block.

Um, and this is where centralized exchanges are very good. Like New York Stock Exchange and NASDAQ, they have been using technologies like multiccast for years to distribute price feeds and distribute new updates of information. And this is where something like, you know, Pith for a long time has come in and sort of said, okay, we can do uh a version of this that hits globally. And a lot of the new architecture that you guys have been doing is around shortening those times of how how long does it take us once we get a pin data in to be able to send it out. But even there, you know, if you're sending data from Pith and you happen to be receiving it in an area with bad internet or simply there's a lot of cables that are cut at the moment, the ISP is going to be the limiting factor in how long it takes you to get that data updated.

And so even when we're on like cutting edge Oracle systems, there's still a real role for increasing the capacity and reducing the latency of the base layer technology because today like that is the limiting factor in how fast we can push these systems.

Yeah. Um congrats on really an ambitious mission and congrats on an ambitious partnership uh with the US government because this has now the capacity to push I I I see this as a very slippery slope in a very good way, right? Like now we're publishing GDP. You mentioned non-farmms payroll. Like there's a whole swath of data here that has the potential to really impact financial markets on chain and and you know we always see at least at least you know when we're looking at the markets oh they revised these numbers and usually it's down in in you know a negative way because they they estimate on on a more positive note.

But if we have this data onchain and things are more interconnected and we have blockchains that almost are mechanisms to regulate these things and normalize them in real time, especially if real time onchain happens a lot quicker, then we have the capacity for greater financial products really a lot of creativity in these markets. And so as you sort of look at what is possible with reduced latency, I'm curious what kind of financial applications do you think with a greater swath of data as well as the reduced latency, what kind of financial products can you dream up that are currently not possible on chain? Yeah, I I think that um I like to put us in a position of like the past the 20 uh uh the 2000 um like IPO pop or like internet um uh bubble where all the stuff that everyone wanted to do on the internet but wasn't really possible was exposed like the pets.com only to then be revealed 10 years later that you actually are buying your you know your pet food on Amazon and you just didn't know like cuz you didn't have the infrastructure to be able to do that. So um we are very focused on bringing institutional level um trading on chain um institutional level of like finance on chain and I think that that's like a table stakes and the way that both of our protocols are constructed is you actually see like it's the biggest firms like the jumps and DRWs that are constructing their private networks.

It's like the you know the the the uber sort of um elite kind of deepin model and same with the pith network. It's like you go and get the Jane Street and Susahan and all their data and they they contribute it. Um that's that's a wild thing, right? Like that's never done been done before. But now we're able to put those tools to place to be able to do kind of like for like um for like versus the centralized um traditional markets.

But I think the thing that's exciting is you go back the 10 years from now and you look at the stuff that people are excited about like oh we can go do like you know houses on blockchains or whatever it is like mortgages on blockchains or things that are like reading a smart contract that's linked to Fed funds rate and that automatically updates so you don't now now need to go to a lawyer to do this or um update all these like paper contracts. That's the stuff that I think that the GDP first represents and is like the harbinger to um having us you know create because you need to have willing institutional participation. You need to have a willing regulatory landscape that endorses it. And having the US government do this was effectively a symbol to say, hey, we're here to do this and we're not doing it in a way where most government entities have done something before, which is a highly permissioned, you know, safety rail guards blockchain like the Ohio DMV on a private network. Like that's not it.

It's like these are public blockchains that the US government has told people go use. go use these to do whatever you like. Here's data from us and we're gonna put more of it on there because we now have the the the way and the means to do it and we're happy to let you do all these things. So, I think that's the stuff that's the most exciting to me is like structured products or you can you can add liquidity to things that have never really had liquidity before that I think we're going to get really more excited about over the next couple years.

Um, Austin, I I a little bit of a different question just to get a little bit more into the technicals of double zero. I'm curious what the the the model of incentivization is when you're thinking about coordinating across each of the partners that you mentioned. There's about 12 of these and how you incentivize them to contribute to this shared network. You know, it it's it's such an undertaking for even just one entity to go about doing this. How do you coordinate uh friendly compatible behavior across all 12 of them?

Uh Senator, we use a token. Uh no, but in all seriousness, the the the way that you build alignment across multiple parties that don't necessarily have direct alignment is through tokens, right? These things are at the end of the day giant digital cooperatives and consortiums that uh achieve something more than any one firm has an incentive or the ability to do on its own. And this is, you know, we we have companies contributing fiber to double zero that like compete with each other on trades and they compete with each other on markets and they make markets for and against one another constantly, but the base layer infrastructure capacity here is something that benefits all of them. So if you have and you know zero it is not a smart contract blockchain, right?

This is not a system where we're we're building our own trading systems or we're saying come deploy your smart contract on us. We're saying, "Hey, Sauna, Apptos, Sooie, Pith, anyone who wants to locate and build infrastructure on this, you should do that because you can benefit from a more high performant global fiber system. The way that works is folks pay to access that network capacity and that system is paid in tokens and then network contributors are rewarded in tokens and about 50% of the the rewards taken in are burned on the network as well. And so you have this sort of relationship that's very similar to how other token models have been built. The difference is we don't work on a proofofstake system because we're not actually securing data on the network in perpetuity.

This is what you know was called a few years ago a stateless network, a network that doesn't hold any state. We're not securing $4 billion of USDC or something along those lines. It's just the data that's in flight. And so we can have this very competitive system between contributors um called proof of utility where you're rewarded for the incremental value that your network link adds to the system. And so what that means is just because one contributor is the best today doesn't mean they're always going to make the most rewards.

In fact, if someone else comes along and says I can get a faster connection between here and here and contribute it to D0ero, they're actually going to be able to come in and basically steal the rewards from someone else going forward. And so, you know, this is the thing that tokens do best. They they weaponi weaponize capitalism and greed to produce win-win outcomes. Um, and that is like the genius of crypto. That is the point of Bitcoin originally.

And we've kind of forgotten about that in a lot of modern incentive design. You want competition between contributors and you want an incentive to outperform your competitor. And too much of crypto nowadays is is has forgotten that core incentive and motivation.

How would you guys leverage something like D0ero? like what is the integration process here and and generally when I came to this panel I was like why are these two guys on the panel together and I was like oh like one of them's pushing data the other one's helping pushing it faster what yeah what's the integration from both here

yeah so so um pit that v2 is actually in one of the data centers that's um uh where where double zero operates a um a lot of their network um and then we are going to have different points for different data sets um so a network like0 is you know kind of the the the hand and glove complement to a network like Pith where we're trying to make sure that we've got the best people in traditional finance and we're li limiting the end to end or tick to trade latency um for for our network. So for us it's a you know it's a it's a perfect type of partnership.

Yeah. I mean not to get too low level on this but like there is there are fiber running all over the world. It's like a giant spiderweb of of glass basically. And if you are in a a space where like you've got a like a Pith price being generated, Pith can talk out two different basically ports. One way it talks out to the public internet and any subscribers that are pulling in data from the public internet can receive it that way.

But at the same time, it can also transit through the zero network and then if you have let's say a trade server that's ingesting Pith data and then wants to take action on that, the fastest path most likely will be through 0ero. But if the fastest path is through the public internet, the server will automatically pick that choice as well. And so what's really nice about these systems is it's not all or one. It's not like you have to operate on zero or you have to operate exclusively on the public internet. These systems can coexist and the systems will automatically choose whatever the fastest path is.

And so, you know, if we have fiber directly connected into the Binance exchange point and you're trying to execute a trade between Jupiter and between the Binance central order book and you're pulling in a price data feed from Pith and saying, "Okay, like it's time to make that trade. I'm sending data out two different directions." Those both may go over the zero network, one may go over the zero network. It just depends on on the physical geography of where in the world that block that leader is being built for if you're trying to get to Jupiter, for example. um and all of these types of systems like this is the way the internet works today.

It picks the optimal routes when possible and D0ero just provides significantly more optimal routes to reach the uh to reach these systems. Austin, last question for you. Um my mother is fed up with AT&T Ubers. She like she when can retail get a zero internet subscription?

So it's uh you know we have no plans to get out to residential internet. um that is actually both a highly regulated business and a very hard one to to get into. Um what we're really primarily trying to service are servers in data centers running blockchain and non-blockchain systems where latency and bandwidth uh are material for the economics of of these systems. Um we have uh you know about 20% of Salana stake running on this today about 20% of Salana validators uh and we are launching our mainet uh in the next few weeks. If your if your mom is upset about her Bloomberg bill though, you can have her give us a call.

Yeah.

Thank you guys so much. Uh give it up for our panelists. Thank you.