Jake Salerno and Ming Wu - 0G - AI Without Limits - TOKEN2049 Singapore 2025

I love a good walk out song like that. Um, I hope everyone is enjoying their token 2049 and thanks for being here today to take um, you know, a little time out of your day to learn more about Zero Gravity and what we're building. Um, I'm Jake Serno, the head of business development at Zero Gravity. We're joined here by our CTO and co-founder, Ming Woo. And for the next 10 or so minutes, we'll be able to kind of talk through the vision of ZeroG, the kind of performance and key differentiators and also what's next because we all know that you're here for the alpha but also want to learn about why we're sitting on this stage today in general.

So, you know, for those new to zero G, can you share more about our vision and why this technology was built from the start? Yeah. So um yeah. So uh I think this is uh the very um so we have really very uh passionate about uh to building the AI with uh the blockchain technology and uh we think that um there are um the problems on both sides that can uh resolve this from each other and for example for the blockchain and uh because the web three uh interface is kind of different than the web two interface. So users need to manage the private keys and which is um uh quite not familiar for normal people and uh so user experience is a a big barrier for the users to adopt um the mass adoption of the web three.

So the AI um when AI comes in the AI definitely can help improve the a lot the user experience for users to um interact with the blockchain uh so that users don't need to worry about how to manage their keys and uh how to uh manage the different chains and how to move their assets between them and get benefit and um and also AI can also help uh the blockchain uh development for the audit of the smart contracts for example to make the system um buck free um and uh um reversely um we also see that AI has a lot of u problems that is mainly due to the centralization right so because in the centralized AI so people um they don't have the transparency so uh we don't know uh the entity who trained the model so what data they use and what target function they use and whether they will insert some poison um back door attack in there so that the AI can have them uh evil behavior when some in some critical scenarios. And so in order to achieve this, we believe that blockchain technology can help a lot because blockchain is a um decentralized ledger that can record everything um transparently and immutable and also traceable. So by applying this um um blockchain kind of stuff in every um critical stages in the entire AI workflow, we really hope that the uh blockchain can make AI more transparent and uh more um um yeah so so that we can guarantee the element of AI for the user intention and the value and also make AI more uh uh democratic. Yeah,

well said Mingan. Um when I first joined the team, the mission was clear and it's still clear today and that's to make AI a public good. But as Ming described, um it takes a lot of technical breakthroughs to allow AI to be a public good and have full providence, have all this transparency and be aligned, right? We want to align incentives and blockchains can certainly provide the ability to do so for AI. So, um, with our public mainet announcement on September 22nd, we're about a week into a public builder ecosystem.

And, you know, there's a lot of hype and excitement. There's a lot of builders on Zerog today, but, you know, perhaps you could talk us through some of the key differentiators of ZeroG versus just being another layer 1 blockchain.

Yeah, sure. So um I think when applying the blockchain or decentralized technology to the AI world there could be a lot of huge challenges. Um uh scalability is the first one and also the u verifiability and the privacy. Um so uh talk about the scalability. So because currently uh especially um the uh with advancement of the uh large language model the model is has a huge amount of uh parameters and it will consume a huge amount of data and um so this actually uh increase the um scalability issue like uh not only data and also the computation power and the storage and and even the user uh user scale.

So this uh this is already well known and um uh AI I think is a is a is a monster that um um really consumed a lot of computation resources. So but the blockchain is a system that um um care more about the security. So it used the uh technique of the replication uh replication based consensus to address the security issue. So which means that it uh natively it's um um lack some scalability um yeah originally. So basically so the key uh system design should be uh uh how to minimize the data that really uh the critical data that really need to be put on the blockchain and leave uh as much as uh datas on the separate um uh storage and the computer resources so that we can maximize the parallelism and the scalability of the of the data processing and computation.

Um so in uh zero G's design we actually introduced the our own decentralized storage and our own decentralized compute network all uh adopt this philosophy and um um so we we we we put the normal work on the offchain um resources but the key thing is that every uh job uh including the storage and the uh computation need to provide uh enough strongest enough proof about its work about the the correctness and the security of the result and this proof is a very concise information that can be put on the chain. So the chain with the smart contract can verify this kind of thing and then um can uh we can design a reasonable way uh incentive mechanism to incentivize the decentralized uh providers can provide these kind of uh resources and services. So um and also uh in this um system um the key thing is that we need to make um make the system uh have the property of the horizontal scalability so that the system can easily scale out with the increasing number of uh resources and this technique actually we applied for all all our our three components the storage the computation and even our our layer chain. So uh in our layer one chain we actually adopt a novel uh we call the shared staking consensus to um scale out the chains um uh in different way. So um we actually allows the validators to um extend their computation by launching the multiple um consensus protocols um participate multiple consensus protocols in parallel and also um the staking status can be put on some anchor chain that uh can be shared by all the different uh consensus protocols so that the multiple consensus can share the same uh same amount of uh the same set of validators and the same um uh share the same uh status um staking status so that the security level of all the chains are same.

Uh so in this sense we actually essentially sharding the the the chains but guarantee um yeah but but never sacrifice the security and the decentralization.

You know it's compelling to see how much has gone into such a usable solution now. Um, that was something that we really worked on from coming out of stealth in March of 2024 is the user kind of friendliness of this for people to build onchain, fully onchain AI. Um, so what we wanted to create was something that was of course usable, secure, highly performant, and now we've actually been able to build some really exciting ways for people to get started, right? using AI to build with AI is uh the lowhanging fruit for sure. So we are announcing a lot of fun different frontends and experiences on the zerog ecosystem side.

We have over 350 partners that have deployed and started using those services for the betterment of their protocol. I see also a few of them in the crowd. So thanks for turning out. But um from kind of the BD perspective, right, you can think about the four core solutions that Ming was talking about. The actual layer one, the decentralized storage, the data availability layer, and then the compute network.

And we have a lot of DAPs and builders that have deployed natively on the Zerog chain, but we also have hundreds of projects that are on different blockchains that are consuming uh decentralized storage, the data availability and compute. So we're pretty agnostic to the entire space, right? Um and I and I think when it turn when it comes to the revenue generating mechanisms in web 3 um we're able to provide decentralized and secure alternatives to cloud. We're going after what we like to call internally cloud 3. And so we're able to provide these solutions that are secure and scalable that rival a lot of the traditional cloud environments but have more security and they actually are more costefficient um through the coordination of these resources.

So um it's been a really fun and inspiring endeavor going out to all the builders across a multi-chain ecosystem and saying hey we have decentralized storage we have DA and compute that can serve your needs. It's very secure. It's very cost um optimized. And so therefore the uh integration of zerog is is relatively um undisputed, right? So we're seeing a lot of traction in the ecosystem and um we of course want to invite you guys to build with our technology.

But looking ahead, what would you say Ming that these folks and the zero ecosystem have to uh be excited about for the next let's say three six months or even the next year?

Yeah. So um since we just launched our midnight and it's a very uh preliminary version and we are going to um keep improving its scalability and performance. Uh for the chain side, we will uh uh we just start one chain now and we will soon um maybe in the um next quarter and we will have the multiple chain uh gradually set up and then we will really have our scalability and we will enlarge um the um the ecosystem side and by on boarding more u utility agent on top of our system because I think agent is the future of the AI right and um um and also I really believe that um um the agent in the decentralized environment by combining this tokconomics can really um lead to the real uh AGI because the agent can be set up in a really um a real society and interact with other agent and humans with some economic um uh constraint and um uh I'm really excited about that and I think u um uh zerog's mission is to um uh continuously u make our system infrastructure stronger to support such kind of a large volume of the Yoda scenarios. Yeah.

Yeah. It's um going to be a very exciting ride with our ecosystem. And um for those who might be new to Zerog um you can learn more about us by going to zerog.ai on any sort of browser, following us on x0_labs. and we've got some really great documentation, but our main page actually links a few different initiatives that might be of interest to you.

We of course have an accelerator that allows us to provide more resources um for you know new and upcoming projects and we've also got um an incubation program called uh you know guild on ZeroG. We've set aside $88.8 million for investing into projects with this program. So you can learn more about that and if you're a builder in the space regardless of the stage we're here to support both Ming and I are happy to hop on calls but uh we've got a great team of 70 and we've got a bunch of initiatives and a bunch of different fun incentives to come and build on ZeroG. So um for now please do keep a lookout for ZeroG's uh shared success with the ecosystem over the coming weeks, months and years.

But you know, we thank you for tuning in today to learn more about us and we look forward to building with you. Thanks, B.

Thank you.