Why Aptos is the FASTEST Growing Layer 1 Blockchain! with Avery Ching

end licensed. They are available in seventy plus countries and if you sign up using my code or link which will be in a description, you can get fifteen dollars in bitcoin when you trade your first one hundred dollars. So if you'd like to learn more about Gemini, visit the link in the description. Welcome into the Thinking Crypto Podcast. I'm your host, Tony Edward, and my guest today is Avery Chang, who's the co founder and CTO at app Toss Labs. Avery, great to have you on.

back then. I was there for over a decade working on data infrastructure, things like do graft processing, large scale scheduling, and then ultimately the DM project. I was the Web three technical project and at the end of twenty twenty one I realized, and you know, it wasn't a big surprise, I guess that we were unable to launch the blockchain.

We built amazing technology within Meta. We did all the things right, We open sourced it under the Apache two license any would use it, and then with my co found Remo, we'd launched aft toss Labs, designed to take a lot of the work that we had done within the DM project and bring it to a much larger audience, developed technology stack for a for all five billion interne users and support a much wider variety of use cases besides just remittances and payments.

distributed systems junkie for a long time. That was also my heir of my PhD, and just kind of reading through and learning about the economics of what bitcoin was bringing to the world and how it could build a secure system through economics, I think was a very fascinating concept and also a product that instead of going kind of being something that embedded within a company like a Meta where I worked in data infrastructure, which was then kind of used as a back end service to power

products in the front end. It was amazing to see a system that people would directly track with a database, anyone could use directly as opposed to something through an interface. Through an interface, it finally gets to kind of back an infrastructure. So that I think was what got me really excited about this space, about the possibilities of it, and then kind of thinking about how does that apply.

You know, now that we have this public infrastructure, how do we kind of build it at scale to support the scale that Metage used to the billions of users and the you know, billions of transactions and dataflow that happened within an environment like that.

definitely early early days of what possibilis could be. They built out a search engine, they built a bunch of different products, explored the space deeply obviously meta now doubling down on social and taking it to kind of its ultimate destination. And then you know, with Web three you know, going breaking down those borders of a kind of a closed environment and saying, look, anybody can build, anybody can interact.

You can build a protocol that does social. I can build a protocol that is financial, and someone can go ahead and build a protocol that links the two of us together and provides capabilities that no one thought was possible because it doesn't require, you know, you are to give consent for someone to do that. Everyone has access to it. All the codes open source, all the codes available, all the services that you and I build are available

to anyone. And that's really, I think, just a novel thing that we've got to take more advantage within the web Web three.

you know, if they don't change in time? Kind of the blockbuster Netflix dynamic.

to my friends? How do I keep on track on what they're doing. Are they focused on the fact that a lot of this data and you know, the interactions happen within a single company, Probably probably not as much. I think what we've got to do with web three instead is build new things, things that you couldn't do because of the kind of restrictions that that you might have at a place like Meta or or or other

companies like that, like Google. We've got to build products that that are that are that are different, that take advantage of the capabilities of Web three. An example might be, you know, if we build a social product and you can actually support payments to every single person on the platform at any granularity. You know, that's something that can be done. Trees infrastructure, you know, today their rules come around.

They may not ruled, but there's definitely limitations around something like you know, Visa will only support you know, transactions of a certain amount. There's definitely fees that come with it. If you want to pay out to these customers. You know, it's complications, there's KYC that needs to be done, there's bank accounts involved. You can only support so many different

customers to pay out to. But in an infrastructure like blockchain, where anyone can have an address the transaction fees of something like aptos or deminimous maybe like eight thousands of a cent, you can actually have payments in many different granularities, even as low as a cent, and they can be very effective and efficient and settle immediately. And so taking advantage of infrastructure like that, I think is the key to really unlocking a lot of the value within the web. Three.

that product. I was on the Web three side, on the on the DM tech stack building out the technology core along with side an amazing team, and so our past kind of intersected as we started to think about partnerships like we used for the blockchain, thinking about how the adoption could be even different ways in which the bloxing could be used outside of just the payments and MNTS infrastructure. And as we started to discuss that, it kind of became clear that this technology could serve a

much larger purpose than just what DM was planned to do. Uh, And we would obviously have to make a lot of changes, we'd have to you know, develop technology a lot further. But it's something that I think really excited both of us. The idea that we could build a public, decentralized utility

that anyone can use. If we think about it the same as water or roads or electricity, you know, it should be something that everyone has access to, everyone can can transact on, and can be a you know, a global settlement ledger for many different types of applications.

a lot since then. We've added features like objects to it. We've had a kind of fuller support for the move prover. We actually have this new effort to call to build something we call move too, which is a new compiler, a new vm UH support for what we call block STM, a dynamic paralism, which is kind of the third iteration of paralism in the journey so far in this space

and supports things like a chain randomness. It is, we think, going to be the ultimate vessel for any kind of web to be developer going forward into the future, and that's an area that definitely needs a technology chef.

of hardware. So if you're a large company trying to build on top of cloud, you're you're thinking like, if my user base grows substantially, you know, doesn't mean that now infrastructure is going to limit me and how much how many users I can onboard? Is it's going to give them time? And that's of course for any business

a huge risk and consideration. I think in blockchain is the same thing, you know, thinking about for a large large company or someone who wants to build at scale, you know, support hundreds of millions of users, and they're thinking like, well, if I build this blockchain infrastructure is it going to you know, eventually hit a limit where I can't go any further. H that's going to hinder

my product growth. We felt scalability was one of those important things to solve because if you don't have scalability, you see what happens on ethereum where demand based pricing drives up fees. You know, if you have high fees, transaction rates you know, can't be supported by an application because customers don't want to pay those fees. Applications can't internally handle those fees at certain scale, it's it's not

going to work out. And so we we've worked really hard in the last couple of years and even before that. This project has been going on since twenty eighteen basically so to really build the scale that we thought could support you know, the five billion Internet users out there, and we've been really pleased to see our results. Last December we kind of announced our results of what we call premun it and preving it is a way to run kind of a mainit like environment with the same things.

One hundred plus validators running our next iteration version of the software the code, and we are able to demonstrate with reproducer results if you want, you want to run them yourself. We have the configurations in GitHub with the same commands. You can go ahead and kind of repeat all this. That we could see more than two billion transactions occurrent day, two billion kind of transactions, and that's an average of more than twenty five thousand transactions per second.

So just to put that at scale in twenty twenty three, I think Visa is something like seventy million transactions per day master guarded formge million transactions per day. Two billion transactions in a single day is actually more than every single credit card processor combined. Wow. And so this gave us a lot of comments to show and in a repeatable or producer environment that you know, aptos is able to support some of the largest workloads in the world

in one industry at least, that's amazing. And then back in August, you know, there was a product called tap Us Cat, which is a very simple kind of click to top game, but underneath the cover is actually doing some pretty complicated things around voting, around points accumulation, and tap Us had kind of achieved a lot of traction and drove up the transactions on main net to three to twenty six million transactions in one day. Wow. And so app to us now holds the records for daily

transactions in main nets. I think the top three records three to twenty six million, one hundred and fifty million, I think something like ninety six million, you know, demonstrating that not only this you know result can happen in the next version of code, but actually in the software that's deployed today. And so we felt that this kind of ability to show that you have amazing scalability, the

fee stayed low the whole time. There were fractions of ascent are really important for developing infrastructure that anyone can build on and can not have to worry about their use case kind of getting hit by the ceiling of what the infrastructure can provide so cost scale. And then latency is something that we've also worked really hard on. Our infrastructure has evolved a lot. We've made improvements to consensus,

made improvements to the pipelining aspects. Something that apps us does very uniquely here is it has a very pipeline design, more than five phases data dissemination, kind of consensus parle execution, parle of storage, and then finally kind of proof generation. And this to these five stages, we can generate both high through put and low latency and kind of drive each stage down in terms of the kind of cost

to the users. We've seen that we put up a dashboard recently with the kind of US versus other networks and showing the app toss is the fastest for it typically around seventy eight indred milliseconds out of Asia for

user in ten times. And then why that's important is because you have a website that takes about two and a f seconds load on average if it's a good website, and you know, being able to be kind of about seven or injurred milliseconds or in that range means that you can actually put a blockchain operation in the critical path of the website loading. So native integration into kind of web apps, into mobile applications, into anything you do

on the internet today. And so the combination of these three things kind of you know, shows shows us where we think that the technology needs to be in order for that masculine adoption. So we work on things that you know, I really take advantage of this, I through pit applications, anything requires little latency, anything that kind of requires that low cost. I think a good example of

that would be use cases around payments. So if you think about you know, for instance, I was going back to earlier kind of micropayments, small amounts of transactions, transacting ascent, transacting ten cents. That's not great for networks like ethereum.

It's really great for things like aptos, right because you can actually move money at at high speed at very very low cost, and so be able to support micro payments or even streaming payments where you know, you are a content creator might have a have a stream and you want your users to be able to kind of you know, commentate you if they think they're doing a job, and they can actually just kind of stream the payments as they're watching your show, and and then that can

you know, you can turn the stream off if they're dissatisfied. Those kinds of things enable new use cases that weren't possibly for new myth is a payment that weren't possible, or if you want to think about things like micro loans in emerging countries, those things are very hard to do in traditional world, but they're possible to do in products like optos. So focusing on those use cases, especially

in the you know, finance low to commerce. We think that there's amazing innovation in that area that can leverage the APTOS technology staff to its fullest.

I know the folks at Brave Browser are doing it in a different way, but they haven't really cracked a model of having the traditional publishers and so forth really buying in and having real time stream payments and things like that. Is that a use case that you guys are maybe working with different companies with FO.

add up. Those amounts of transfers are important, and we can support the use cases better than anybody in the world. So that's the kind of area that I think we really are are are designed for solving. Oh for sure.

a really important thing. In general. I think that the token un locks are are are probably industry leading most you know. I think the investors and the core contributors have a for the year unlock schedule, and the Foundation is a much larger ten lot on ten year unlock schedule for fridge tokens and for the community bucket and for those kind of things. I think, you know, you really want to take a long term perspective on building

a network. It is not going to be something that happens quickly over time, and as we've seen in the space, finding actual product market fit is difficult. So designing something for the long term, I think, which is the way the tokenomics of aptoss is structured, is really important and I think it's you know, it's actually set up quite well.

The other thing is, I would say when it comes to the the mechanisms around the token, they're extremely simple, and that's by design as well just being able to have anyone but understand there is a staking emission and there's also a burning for transaction fees, and that's pretty much for the most part. Yet when it comes to creation and deletion of tokens makes it easy for anyone to understand, where sometimes there's very complicated schemes of like

some parts are burned, some parts are saved. You know, there's all this other stuff, and like you know, it's very hard for people to get the grasp of like what what's actually happening within the network. The simplicity is really really important.

maybe ten years ago. I also worked on some other kind of supervised learning algorithms gamins, clustering, waiting to scent, and so I think for those it's it's really interesting to see that intersection now of like the work I've done before in blockchain, now a lot of it is to be around the data preparation and the kind of data lineage, and that's something also in data infrastructure we spend a lot of time on. So thinking through that a bit and then kind of finding the right partners

in the space has been key for us. We've announced our partnership with Overlay. Overlay is a group of content creators, some former naturally to fact photographers and others who are really concerned about the you know, how how their data is being used in l MS, how their DA is

being just used elsewhere as well. Uh, and you know, one of the ways we want to solve this problem is through through tracking that data, creating controls around it, creating standards and how that data can be used with uh, you know, the creator, creator of permission and you know

we're all creators. You're creating a podcast, you know, I'm working on a blockchain, and I think it's important for us to be able to kind of start to put a lot of that historical information on chain, you know, having that ability to kind of say, this is my content, I created it, I'm signing up with my private key,

and you know everyone knows about it. And now going forward, because this is an immutable record of history of what I created, is something that is going to be I think really necessary in this world of data being used in different ways and kind of without you know, maybe without permission, without access. And so that's where I think blockchain can really really help with the I space. And I think we've seen a lot of concerns from you know, people looking at llms and saying like, did you know

where'd you get the data source from? Yeah? Did you have permission access that? And you know, sometimes the answers are a little a little hazy. It's going to be a world where we need to have more attribution around what's being created and more controls around it. And we're very proud of work with companies like Overlay on this specific area. Yeah.

more AI into it. It's not surprising that as you have also transaction fees lowering and if you have three point going up, what we've seen in systems like Nasdaq is I think sixty to eighty percent of that is automated, you know, algorithmic trading, kind of trading by bots, some sense program bots, and you'll start to see the same thing happening in a world of crypto where anything can be traded twenty four to seven without restriction and that is inevitable and so being able to kind of build

out infrastructure that can support that at scale, I think is probably the next generation of where this technology is going to go.

Right Versus Facebook and these social platforms and web two they monetize it, but they keep the money, right, but you don't get a cut. They just monetize your activity. But in the future, because of data privacy but also blockchain, you can sell that to advertisers.

that exact route. You know, if they if in today's actions, you know, if I watch an ad or if I you know, you said share my data for your usage, there's no reason why you can't give a cutback to me. And with blockchain you can have you know we're talking about earlier. It might be hard to traditional rails to do that, but it's actually quite easy within the blockchain space.

You could even have like very low amounts of payments going out for for you know, less valuable data, yet make that extremely efficient with infrastructural like apt toss.

designed and even features. It was designed like, you know, what's the core we can build that actually prevents programmers from making mistakes and you know, allowing mismanagement of funds, those types of things. And so that's kind of one of the core tenets of what we're doing. That's important

from a security aspect. The other thing is we we you know, security is also somewhat of a you know, I don't want our head of security outset, but you know, a little bit of a boring hard work in some sense, it requires us to be just very tedious around how does code get into the rebuk. How much testing is there. There's unit tests, there's aggression tests, there's cluster tests, there's security tests we add in. There's fuzzing that goes on as part of the process. There is a dev net,

test net, then a main net. There's also something we call the prev net to test major features going out. So there's a lot of factors here. Also auditing that happens at different levels of the stack, so it's it's not something that's like a one size fits all type thing. It's really about putting together a process from ideation all the wage production that kind of checks as many many different angles as possible and tries to prevent as many

many issues as possible along that entire journey. And so it's something that you know, anticipated, really a mindset from an organizational viewpoint, something that everyone internalizes as important. And the trade off for security here is more about speed of time getting code into production, because if you're not doing those checks, you can actually push the code much faster. You kind of the features rolling out, but at a you know, at risk and our toilet risk tolerance is

definitely going to be much more. You know, we want to push both the feature set moving out quickly and security, and we find a pretty good sweet spot you know, around like you know, doing a lot of these checks and balances as a first step and building really strong infrastructure to support that as much automation as possible, and then deploying uh, you know, through the IP process extremely quickly. For decentralization, we've taken a different perspective, I think than

than others in the space. We really focus on like what is this centralation trying to accomplish here, and it's it's it's about kind of think a couple of different things. One is about the process itself. A desmilized process has a lot of different participants in it allows a lot

of opportunities for participation as well. So going back to the idea of production concept I brought up earlier, and when someone has an idea in the aptos's blockchain, they kind of create an AIP, which is our process for like EPs. Essentially, they submit a proposal, they talk about what they want to do, they wishments, uh, there's community feedback.

Then there's code that comes in. There's more community feedback on that as well kind of review around that, and then there's also that codet shipped validators valudors need to kind of accept that code or not. And then there's also another check which is on chain governance and tokenholders will then vote to make the actual change go live. And so this kind of plays into both security and decentralization.

You have many opportunities participants to to give their feedback to voice support or you know, concerns, and you have a process that's very open and so this is all really important for the centralization. The other thing that is is about kind of the metrics of the centralization. And so for this, it's like, you know, how easily can your system be compromised? For a Byzantine fault tolerance system like ours, we support up to kind of one third

of the stake being kind of malicious, uh. From guarantees of that, we can guarant you know, we can ensure safety and liveness of the protocol. And so there's something called the knockomotive coefficient. I don't know if you're familiar with that, but it's the idea like what's the minimum number of machines that you need to be compromised in the system in order for it to kind of fall apart,

lose liveness or lose safety. And so for for aptos, I think ours is about twenty two right now, which puts us in very much one of the highest among major l ones. It's higher than Salana, for example. And we do that with about one hundred and fifty notes. And so it's less about the number of nodes and more about like how make these nodes can be compromised in order for the system kind of fallover. You could

have thousands of nodes in the network. But you know, if you have all the stake in one node, those other nodes do nothing really right and you don't actually secure the network. They're great for a media sound bite, they're great for a slide deck, but in practice they do not actually help the network one bit. And so we really focus on the metrics that kind of matter for us. When you think about decentralization, it's really about kind of avoiding, you know, single sources of failure, even

a couple of sources of failure. What's the right balance there, and then from the process standpoint, really making it as open and transparent as possible with as many participants as can be added to the process.

abtass and amazing rails. But around them we have a funge Blassis token stand which is extremely powerful and rich to support these assets at high scale, very low fees, and very interoperability across the ecosystem.

I think as a broad population. Technologists do obviously because they power a lot of the back ends of the largest technicomies in the world. But there's a lot of function that goes above that stack before you build applications, and so we're pretty excited about some of the technology really on there alongside that. I think something we have announced in the past is apt us Build. Apt Us Build is a product from Aptos Labs to allow you to construct rich applications and be a one stop shop

for you as a blockchain developer. Let me give you a little example. So again, if you're building directly on top of a database like my squel or Postgress, there's a lot of seacle course you have to write, there's a lot of infratructure, the kind of glue you have to glue together multiple databases or add layers on top layers of abstraction to get APIs that would make sense for your application. App to us build kind of does a lot of that hard kind of glue work for you.

It provides APIs for directly minting NFTSD, APIs for acticing data on chain in different ways, kind of more in databasel like format or more kind of even like through the full note kind of APIs, if you will. And it's starting to add more infrastructure around gas station fee management, custody solutions. We have something called as to us connect which allows you to log in with your Google ID or your Apple IDA and have a wallet in the back end kind of secure with a zero knowledge proof

between the two. So there's some anonymity and privacy for you. And this kind of infrastructure adds up into being like use application developer, really focus on your app. What are you trying to solve? Are you trying to make you know, videos more accessible in different parts of the world. Are you trying to build a new game that takes advantage of the dnamics of web three? Focus on those things and those user use cases, and you know the glue is going to all be there for you to just

integrate a blockchain. You don't have to hire ten blockchain engineers and smart contract developers and all this other stuff. You can put a bulk or your resources into building out great engaging experiences for your audience as opposed to focus on the blockchain. And so I've talked Bill something we're extremely excited about. It's already kind of available in

the early form. There's many functionalities, much of the functionalities there, but there'll be much more coming in the throughout the rest of the year and at beyond.

We have Raptor, which is the new consensus protocol being rolled out throughout twenty four and beyond. Raptor is going maybe just a brief history of consensus protocols. There's PBFT that was kind of from nineteen ninety nine define kind of the optimal latency possible and consensus. We've achieved that earlier this year. We're really proud of that, obviously, but

it's not just about achieving the optimal latency. It's about how do we now make sure we shrink that entire user latency too as close as we can to get the optimal liency as possible, and we're going to be using Raptor to do that. It's going to actually support a lot of the unstructured world of the Internet much better to handle kind of downtime from nodes while achieving those high high throughputs and extreme liancy. There's major roleout's

going to take place with that. We expect to see aliency decrease, throughput go up and kind of pushing us even beyond where we are as class leading to the next stage of that. There's also the work on block stmv two blockasting kind of the original protocol is something that we're really proud launched with aptos from day one. It's something that really DEFINEDES and made us unique from

other networks. It is a form of dynamic parlism. So if you have the history of paralism in the space, you have kind of the sequential execution executing transactions one by one with ethereum. You have the static paralism which networks like Salona do, which you can defend defined dependees up front and kind of put more owners on the programmer, and the system uses the dependencies to then schedule transactions

as concurrently as it can. And you have kind of this third wave of parallelism, which we call dynamic paralism, which can use those techniques but additionally has a runtime component. That runtime component means that you don't have to specify anything well, system will try to figure out for you what's to run in parallel and take care of about on the back end. And the dynamic paraism allows more

concurrency in many cases. Uh. And so what we've seen and we've been excited about is people have been kind of quick to adopt even the blockst TM and their their technology stacks. We've seen it with Polygon, with dark Net, with with say and even I think Monad and many others. And so block Test in v two is going to be the next duration of that solving the ability to salve to scale to many many cores and machine and we see the trend hardware trend is going to be

many more cores on a validator. That's just the kind of way that More's law is progressing. And it's been going that way for a little while. So going from MADI machine having say thirty two cores today going into sixty four, one hundred twenty cores two and fifty six cores, we want to take advantage of all that hardware, and

blockist v two is designed specifically to do that. The last area that we're really excited about from infrasturcer standpoints moved to I think I talked to us some of the high there, but we we we want to build something that is kind of almost the endgame for Web three developers. It's definitely a challenge for what three developers today compared to the traditional programming they're used to. And if you code in Rust or if you code in C PUS plus or C or Java, all these kind

of traditional languages. You have so many tools available to you, you have so much functionality available to the language it self has evolved to support what programmers need in terms of data structures, in terms of program interfaces, and so move to is designed to kind of be that really accessible but yet safe and secure and performance what three development environment for you. It will include things like dynamic

dispatch that is safe. It will make sure we support the form of verification that ensure the initial there of safety on top of your protocols. We're building out a VM that will rival you know, all the modern you know vms that exist in the in the current Internet world, and all these things will add up to something I think that's a true delight for web developed Web three developers to code with.