Nicholas Chen - Jupnet: Our next-gen Omnichain VM - TOKEN2049 Singapore 2025

How's it going? So, I'm Nick. Uh, I'm from Jupiter. So, today I'm going to talk a bit about JubNet and our our nextg omni chain VM. So, basically I lead product in Jubnet.

So, what I mainly do is I mainly work with closely with the developers in the entire team to really focus on the core pillars of Jubnet. So here's a recap of what we've been doing thus far over the past eight months. So JupNet aims to be the omni chain liquidity hub and we want to power omni chain swaps from one account and have the max performant avenue venue for all assets and also be an interchain infer for all chains. So to make it possible, we are going to create the most functionally innovative VM on top of the most performant one. So how do we do that?

By doing so, we would have to perform a few focuses, right? So we want we want to have three core anchors. So we want to have a concise and advanced interchain stack, a decentralized authentication layer capable of simplifying and onboarding of users for omni chain asset management and of course to have toxic flow prevention capabilities to build even greater capital markets. So why do we want to build our own VM, right? And there's a really really common quote that we've always seen over the past um decade and it's that people who are really serious about their own software should make their own hardware.

And we are really really serious about a full stack from enabling the maximum level of control information and authority to you the user to enabling a constructive base for an efficient capital markets network to simplifying how an advanced omni chain layer is built. So to ultimately enable us to build a single source of truth for all omni all of the chains. So oh so here's a quick overview of the VM. So juke VM is actually a heavily modified version of SVM the Salana virtual machine and it encompasses three main frameworks. So it basically works on aggregating ledger states across interchains.

It's going to carry a built-in cross-chain consensus and be able to process an infinite number of signature standards and also be able to prioritize trade executions. So infinite signature standards mean that not just Solana's ED25519, you will also be able to process a ton more um cryptographic signing standards that are available today like in Ethereum. So here's a broad overview about droop VM's architecture and the architecture can be really focused on three main concepts. So the three main concepts are dove sik verify batch transactions and of course jukevm itself. So what droovm actually contains is a suite of programs that enable the omni chain ledger to maintain a single state for all the chains we cover.

So here are the key updates in jukevm over the past four months. So we have Duffs which carries the native omni chain stack aggregated decentralized identity which is dupnet's builtin native decentralized authentication framework for users built in with ZK and of course the toxic prevention engine which prevents me and prioritizes performance leaders on the network. So here's the omni chain stack first. So in Jupet we don't just run validators we run doves. Doves are decentralized oracles that validate and execute user intents towards the other chains.

So there therefore we call them doves and doves focuses on four main concepts. So in consensus we also run the network for consensus but we also process omni chain consensus through observing related interchain transactions process interchain service validations and of course authorize the omni chain executions. They also produce and maintain a single state of truth across all chains. And because of this hub and spoke design, interchain state becomes exponentially easier to maintain over time as we adapt more and more chains. And this is also going to result in a much more minimalist and consistent design for security assumptions across varying finalities such as Ethereum, Solana, and Bitcoin.

So here's a quick look at how fast a first step on jukeet can can perform a purchase across multiple chains. So right here I'm going to make a multiple purchase across Salana base arbitum and more. And look it takes less than 10 seconds for me to purchase multiple assets from multiple chains. And of course, we are also excited to share that we have massive progress made over Bitcoin where we're going to closely work with Zeus network, Solana's native Bitcoin um framework team to actually work on implementing Bitcoin within Jupet. And that's the omni chain stack.

With the omni chain consensus, the omni chain ledger and omni chain services, we have a full integrated suite of core components native to jukevm to operate the omni chain liquidity network. Next, we have ADI. So, at the start of the year, we introduced a completely different way to access, manage, and recover assets in SVM with an onchain user account system. Today we now have ZK builtin and that means that your authentications are completely private. So the goals for ADI are just three simple things.

We want everything for authentication to be fully onchain and natively supporting all of the authentication schemes that you use today. Apart from that it should be universally accessible and be unified as much as possible and be secure and recover recoverable. So think about direct onchain pass keys and open ID login. Nothing centralized, nothing stored by anyone. Everyone onchain shares it together as part of the DA's network.

We also have MFA which is multiffactor authentication. A way for you to recover your key, I mean your account through recovery keys and of course have advanced user account managements as a default model for you to interact with the network. So, here's a quick look at how pass keys work in Jupet. Oh, wait. Uh, yeah.

Uh, O offers, sorry, with Google. So, you sign in, you click your account, you get in, and voila, you now have an account with multiple deposits across multiple networks in one place. The second is actually pass keys. This is never seen before. So all you have to do is just if you have a MacBook just tap your Touch ID and you're good to go.

You're into your Jupet account and that is ADI. Last but not least, we have the toxic prevention engine. The reason why we're building this is because we want to create an effective onchain capital market and that's why we want to introduce a toxic prevention engine framework to address a few key goals. So the few key goals actually are to remove as much toxicity as possible on chain. Today MEF operators do batching and sequencing of transactions to maximize the re revenue and rewards.

This is really really not aligned to users or apps or builders like yourselves. We also want to improve performance chain performance by enabling bigger capacity of trades per second. And ultimately we want better UX for our applications so that users get the best experience and use case and efficiency possible. So to address them we have three things. We have preferred leaders, batch transactions and prioritized sequencing.

So one very interesting concept is actually batch transactions. they actually get priority locking since they process a significantly larger number or count of user operations or what we call trade actions through a special batch transaction pipeline that's built within SVM. So this is a simple concept of how we can sequence transactions with batch transactions in Jupet. If you look at it in a different perspective, you have a ton of place and cancel orders. given enough place and cancel orders, they can actually cancel each other out.

So what's the point of pushing them onchain as a transaction? But what we can also do is that if we were to put them together, we can basically flatten them out and increase throughput. What we've seen thus far with our current implementation is that we've managed to reduce transaction size for about 300 trade operations by about 70%. So this is a brief overview of an explorer that we've built called Jup FM which actually which actually depicts and illustrates a transaction that contains 300 different trade actions in one place. You would not be able to see this in a single transaction and we are really really happy to share with you today.

And that's it. That's basically Jupet. So, JupNet focuses on these three key concepts across the entire Solana stack to build a heavily modified SVM to solve real problems that our users need. But before I end off, I would like to thank my team and my partners. They're really amazing people and without them, we really couldn't manage to build this.

We are about 20 people now. I mean close to 20 and uh we hope to grow more and we of course have valued partners like Instad who built Jupiter land with us, Swanden and Zeus of course and we need your help. So if you like to learn more about Jupet and you want to partner and work together with us to adapt more chains or even your chain feel free to come by me and uh look for me anytime.