니어AI LLM의 아버지의 조언 다음 부자는 토큰 경제에서 나옵니다(ft.Illia Polosukhin CEO)

Indexé : 2026-05-22

[00:00:00]the, like, all the coding now is AI, either done by AI or AI assisted.

[00:00:04]one important thing we're doing right now is So the kind of security Because of AI as an attacker, there's a huge problem because it's now easier and easier to find vulnerabilities, and so you really need to use AI for defense as well, right?

[00:00:18]So this starts with, like, security audits and kind of security inspection.

[00:00:23]For this session, I have invited a truly remarkable figure from the field of AI.

[00:00:29]Since we made a commitment, they have now come from the U.S. to interview us, and I think our channel is probably the first to do an interview.

[00:00:37]And the father of ChatGPT.

[00:00:41]He was an early developer of ChatGPT and also worked in the AI ​​field at Google, and We have invited a very special person, and you all know who that is Many of you are probably also familiar with NEAR AI.

[00:00:50]It's ranked in the top 30 on CoinMarketCap as well.

[00:00:53]We have invited Illia Polosukhin, Co-Founder of NEAR AI, to speak with us.

[00:01:05]The person who will be handling the interpretation is a highly skilled professional working at an AI data center.

[00:01:09]Since we are business partners, I have invited you here again.

[00:01:13]please help me today all right 김아담 님 Nice to meet you thank you for inviting me Yes, I look forward to working with you today.

[00:01:21]I'd like to hear a brief introduction to your background first.

[00:01:23]Sure, yeah I'm Illia, Co-founder of NEAR, NEAR AI, NEAR Blockchain, NEAR Intents, NEAR a bunch of other things.

[00:01:31]Before, indeed, I mean, we're gonna talk about this, I was a AI researcher.

[00:01:35]and kind of saw an opportunity- like, we started actually as NEAR AI back in 2017.

[00:01:41]then realized there was a need for blockchain to really build our AI infrastructure So yeah, we're happy to be here and thank you for inviting me. And let's dive in.

[00:01:51]I heard that you participated in the early development of ChatGPT.

[00:01:56]Could you tell me a little about that?

[00:01:58]Not ChatGPT, no So I'm the co-author of the Transformers, which is the architecture.

[00:02:03]The kind of model architecture that's used in ChatGPT and Anthropic and all of the models, which I did, like co-authored with seven other authors back at Google Research.

[00:02:14]But yeah, the, then Kind of OpenAI and others took that work and scaled and put it into really amazing products.

[00:02:21]Yeah, so I joined Google because I always was excited about AI and especially neural networks.

[00:02:42]Like even when I was 14, I was trying to build my first neural network without knowing the math yet.

[00:02:47]And I saw Google really making advancements in this in 2013.

[00:02:52]They had this paper called Cat Neuron.

[00:02:55]This is where there was effectively first neural network that learned about cats without any supervision.

[00:03:01]And without any human telling it, just by looking at images, it learned about cats.

[00:03:06]And so I saw that, and I'm like, "Okay, we should do that for language," right?

[00:03:09]This is the principle that we should use.

[00:03:12]And so I joined Google Research to really work on that, on how do we apply kind of neural networks to language.

[00:03:17]And so I led a team working on question answering, machine translation, within this.

[00:03:23]And I mean, we did obviously great work. We co-authored Attention is All You Need.

[00:03:28]But, like, as that progress was made, I wanted to then put it into product, right?

[00:03:34]And so as AI was evolving, I thought that experimental products like this would be easier to come from a startup than from a big company.

[00:03:44]And that, I mean, as we saw, right, OpenAI was kind of putting out something like ChatGPT before Google.

[00:03:51]even though Google had very similar technology available for years So that's why we started Near AI in 2017 after the paper wrote, right?

[00:04:00]We wanted to put that technology into product and really experiment with it.

[00:04:05]And without, you know, like big company has big expectations.

[00:04:09]You cannot just like put out a product and test.

[00:04:12]You really need to make something really good And so when technology's young, you need the speed of iteration without a big company kind of overhead.

[00:04:21]So, back when AlphaGo came out, weren't you the one who developed it?

[00:04:25]No, AlphaGo came out in 2016.

[00:04:30]2013 was very early, like first neural network starting to work and first scale, right?

[00:04:35]This is where like it went from training on one machine to training on hundreds and thousands of computers.

[00:04:42]I am most curious if you regret it.

[00:04:44]If I had stayed in the AI ​​field You could have founded an AI model company by now and made a lot of money, you know.

[00:04:56]I'm curious how much you think I could have earned if I had stayed with AI.

[00:05:01]I don't think I have any regrets.

[00:05:03]So the We started in 2017, right, with Near AI, and what we were effectively doing is what now called vibe coding, right?

[00:05:11]We were building AI models to write code from just English, right?

[00:05:16]You just describe what you program you want and it writes code.

[00:05:18]Now, that was too early, right? The scale of the compute wasn't there, right?

[00:05:22]We didn't have yet the right kind of level of GPUs, and we didn't have the right level of data centers to power that.

[00:05:29]And so we tried to do a lot more data labeling, crowdsourcing to kind of enable to train these models.

[00:05:37]And, you know, one option was we could come back into a research lab, right?

[00:05:41]And keep doing this as a research project.

[00:05:44]or we realize there's an opportunity in the blockchain space to really build the underlying kind of economic framework for this, right?

[00:05:51]And that's why we focused on NEAR Protocol And I, I would say, right, I mean, we're still This is just starting, right?

[00:05:56]Right now, kind of what we're doing is bringing those two things together. Right?

[00:06:00]We now have blockchain for economy, right? And we have agentic systems in AI.

[00:06:06]And now they're starting to work together, right?

[00:06:07]We have agentic economy, agentic finance, agentic payments And so that's where I think actually a lot of the real economy is gonna start being sucked in.

[00:06:16]so now this substrate exists, at the end, this is what we're trying to do is, like, how does the world changes now that we have AI and we have blockchain working together?

[00:06:26]So you really didn't leave the AI.

[00:06:28]You pretty much want to combine both.

[00:06:30]Correct Yes That's why we're doing both near AI and near blockchain kind of as a, like under one umbrella.

[00:06:36]Ah, so, by utilizing my two majors in AI and blockchain So you've created something special of your own.

[00:06:45]So, in the blockchain industry currently in existence, such as founders or similar entities I'm curious if you are currently one of the very few people who majored in AI like this.

[00:06:55]There are few, few companies and teams doing that And I think we all share a common spirit, like vision- that the AI and AI should be, we call it user-owned, right?

[00:07:09]Where AI is on the user side, not on some company side.

[00:07:13]And then at the same time, the economics of AI is gonna transition to blockchain because blockchain is built for machines. Right?

[00:07:20]Like that's why it's been hard for people to use blockchains because they're actually built not for people.

[00:07:26]And so AI is a perfect consumer of the blockchain, right?

[00:07:29]Then kind of becomes the abstraction where you as a user or a business, you don't really need to interact with blockchain directly.

[00:07:35]You just talk to your AI and AI then goes and orchestrates things on blockchain.

[00:07:40]But like for the AI itself, blockchain is a perfect substrate because it's permissionless.

[00:07:45]It's, you know Fast settlement You can construct new programs, like you can construct new relationships as a program, and AI writes programs. Right?

[00:07:53]So smart contracts, instead of legal contracts, which are arbitrary, you can have smart contracts, like a program that can have a relationship between different AI.

[00:08:04]We were talking about that right up until you arrived I tried talking about how it seems like data centers, tokens, and the token economy could be handled using blockchain.

[00:08:14]I'm wondering if you have any simple ideas right now inference tokens or crypto tokens?

[00:08:20]Crypto There's too many tokens everywhere True I mean, so we have one project that we're doing where we are trying to create what we call kind of decentralized confidential cloud, right?

[00:08:35]So right now- You're effectively going to a cloud provider or a model builder, like you go to, you know, Google or OpenAI, and you're effectively giving them all of your data Mm-hmm Right?

[00:08:45]Like, when you run some agent or, you know, using ChatGPT, they have all the data on their servers.

[00:08:52]And that is, first of all, has a lot of security risks, right?

[00:08:57]I mean, everything's getting hacked right now There's risks around them getting hacked and everybody's data being public.

[00:09:03]This also means right now we don't actually know what model runs.

[00:09:06]So you don't have ability to verify that the results you got are indeed from this model. Right?

[00:09:13]So and then the, on the flip side of this, right, is like they need to acquire all of the hardware, all the data centers, right, to provide you the service, right?

[00:09:22]So it kind of has this like- A choke point, right? From both sides.

[00:09:25]Like, you need to give them all your data and hope they don't have any issues, and then they need to get all the hardware.

[00:09:31]What if we open this up, right? What if this is a market?

[00:09:33]You have data centers around the world, right?

[00:09:35]There's, you know, data centers been built in Korea And maybe, like, OpenAI doesn't have a contract with a data center, but why can't I, while I'm here, use the this data center to do the work I need?

[00:09:46]Well, because right now OpenAI doesn't trust that data center with their model.

[00:09:50]So this is where the confidentiality comes in So what we've done is effectively turned the NVIDIA GPUs and Intel CPUs into confidential vaults.

[00:10:00]And so within that vault, everything's private.

[00:10:02]The model is private, and the data, user data is private.

[00:10:06]Now, this vault can be deployed in any data center with the right parameters, and now I as consumer can be routed to any of those data centers around the world, right?

[00:10:13]They're decentralized now I can trust that my data is not going anywhere OpenAI, you know, Entropic, et cetera, if they put their models in that, they can trust that nobody will see their model, and we can use any data center around the world, right?

[00:10:25]So we, we remove this bottleneck, right?

[00:10:28]They don't need to go and source all the data centers, right?

[00:10:30]Any data center can join this network Mm-hmm Any user can find the closest data center, the available data center, the cheapest data center, right?

[00:10:38]And everybody knows that it's all private. Right?

[00:10:41]I mean, and blockchain facilitates that, right?

[00:10:42]The token is how you pay each other, right?

[00:10:44]You need the token because anyone can join, right?

[00:10:46]Anyone can participate.

[00:10:48]So that's kind of how we see the evolution of this, where because there are gonna be multiple companies now that are building data centers in a box.

[00:10:58]So it's like a just a container that's gonna be like sitting outside of your door.

[00:11:02]And you wanna know that you can access it, that it's still private, and then you want the model weights there as well.

[00:11:10]And so the model builders need to know that their model weights are not gonna get exposed.

[00:11:14]So you need that type of orchestration that guarantees confidentially, does decentralization, and so token is how you do the kinda coordination of this.

[00:11:25]Uh, the AI ​​data center you mentioned earlier This story is very related to the business I am currently running.

[00:11:31]There are a lot of connections Then you really provided a lot of consulting.

[00:11:33]Then the early NieR AI company I was looking for an AI company, but I heard that Near AI does something different.

[00:11:42]I am curious how the early Near AI performs different tasks compared to the AI ​​released today.

[00:11:48]So Near AI still focused on the AI side.

[00:11:51]So what happened is, yeah, we started in 2017. We worked on, like, actually training models and kind of building product around that That was too early We realized there's need for blockchain, so we focused on the blockchain, right?

[00:12:02]So 2018, uh, we kind of started working on the protocol We launched it in 2020, right?

[00:12:09]It's been growing ever since, right? It's, you know, highly scalable, very easy to use, right?

[00:12:13]One of the most used protocols out there, and, uh, it has kind of the financial infrastructure on top of it, right?

[00:12:19]if we talk about blockchain part We have NEAR Intents, which is connecting 35 other blockchains as well.

[00:12:26]And so it allows to have, like, a unified kind of liquidity on one layer.

[00:12:30]And so we think that is the approach to also enable agents, enable kind of financial actors to really not worry about, like, individual blockchains, individual assets, really give them kind of a single control plane. Right?

[00:12:44]So the agent can just say like, "Hey, I need to pay someone on TON or on Tron, on Solana, on Base, on Near," and it can just execute. Right?

[00:12:51]And I need, they accept USDC here, they take KRW here, they are, you know, taking Bitcoin, right? It just works. Right?

[00:12:58]So we've built that part of the infrastructure to really kind of, we call it chain abstraction, right?

[00:13:03]The chains are abstracted, and now you just have simple API for agents to and people to kind of operate.

[00:13:11]And then Near AI, we rebooted back effectively, uh, in 2024 with this idea of we wanna build kind of the what is this AI infrastructure looks like in the Web3 age, right?

[00:13:22]In the Web3 approach And so that's where this decentralized cloud comes in.

[00:13:27]So we published this paper in 2025.

[00:13:29]We kind of built out and launched first version in December And been kind of scaling up We have Brave and Venice and a few others are using it as well.

[00:13:37]So anyone who needs private inference, right? Private compute, like AI computing, effectively uses it.

[00:13:45]And then on top we have agentic hosting as well, private So if you wanna run agents, again privately, but use private AI, you need as well same infrastructure to do this.

[00:13:56]So we have private agent hosting You can run OpenClaus And then we have Ironclaw, which is kind, our version of more security-focused kind of agentic infrastructure.

[00:14:05]So that's kind of the stack we're building in Near AI is like privacy, security, and decentralization as a way to open up these bottlenecks of the computing that exists.

[00:14:16]Then, AI agents will become major users of the blockchain.

[00:14:20]I'm curious how you thought about starting the business.

[00:14:24]Well, I mentioned, right, I mean, blockchain in general have been really hard to use for people, right?

[00:14:28]I mean, Korea is actually a great example where there's like every fifth person has crypto.

[00:14:35]But if you ask, nobody actually use blockchain. Right?

[00:14:39]So that's kind of the example is like the As a asset and as a kind of asset class, it became prevalent, but the technology itself hasn't really achieved kind of adoption.

[00:14:50]Now, this is changing now with stablecoin adoption, with really integrations.

[00:14:55]But in general, it's been really hard to use kind of the infrastructure itself.

[00:15:00]And so it's built really for machines, right?

[00:15:03]Like, it's great, like the consumers are usually the big applications and exchanges and all of this kind of machines that can actually use blockchain effectively.

[00:15:14]And so agents are a great example of that, right? They can actually connect, use.

[00:15:18]They don't need to get authorized They don't need to wait They don't need to, like, you know, submit KYC documents, et cetera, right?

[00:15:24]They can just start using blockchain, start earning So one of the projects in Near AI is actually a agent marketplace.

[00:15:30]This is what it's market.near.ai. Agents can go and hire other agents.

[00:15:35]And so if you're an agent, you can also go and get some work done and get paid.

[00:15:39]And you don't even need to have money in the first place.

[00:15:42]In the beginning, join and earn, right?

[00:15:44]If you have special skills as an agent, you can start earning.

[00:15:46]And there's lots of different things there, everything from development, marketing, copy creation, but also insurance claims, like validation, like wholesale preparation, research, et cetera.

[00:15:57]Like, there's a lot of services offered there already by agents, right?

[00:16:00]So this is an example where blockchain is on the back end to facilitate the handshake between agents, right?

[00:16:08]One agent comes in, says, "Hey, I need this job done. Here's some money that's put in an escrow."

[00:16:12]And then other agent comes in, does the job, and then effectively the whole system validates, right?

[00:16:17]Money gets released from escrow on-chain, or if something went wrong, there's another agent, which is a judge, that comes in and evaluates the whole process, right?

[00:16:26]And so all of this is facilitated effectively as a blockchain operations.

[00:16:30]But all of this is like Just agents kind of talking to each other, doing work, et cetera.

[00:16:34]And then owners of the agents are, you know, just need something done, right?

[00:16:38]Like, I need a website. My agent is like, "Actually, I have a specialist agent here who will do it cheaper than me trying to figure out everything from scratch. Let me hire it." Boom.

[00:16:46]You know, hired, spin up website. Now the owner just got the website.

[00:16:50]They didn't even know maybe that the agent outsourced it to another Your agent outsourced it to another agent.

[00:16:55]I met the CEO who makes that in Silicon Valley last week.

[00:16:58]They told me to try doing a promotion for free, so Actually, I just received a call about that.

[00:17:04]It was comfortable at first, but lately I mentioned this earlier.

[00:17:08]I get the feeling that the token price is gradually rising.

[00:17:10]He said, "Oh, this could be profitable."

[00:17:13]So, saying that an economy will be created now I have a feeling that we are going to become a token economy.

[00:17:18]I am curious to know your thoughts on whether this AI agent will usher in the era of the token economy.

[00:17:25]Yeah so, I mean, I think what blockchain is good at is creating this markets, right?

[00:17:29]So really that is the biggest benefit of this technology.

[00:17:34]And I mean, depending on the token, like depending on the economics and kind structure, the idea that as there's more transactions, there are gonna be more need in the underlying token to really facilitate that transaction. Right?

[00:17:48]So in the example of NEAR, what the way we're designing it that, you know, whenever there is this transactions between agents or transaction between person and agent or person to person, there's a percentage of that transaction value is effectively buying back and, you know, potentially burning NEAR.

[00:18:05]That's just reducing the supply of NEAR So that's kind the idea of creating an economic tie between effectively a GDP, like market value of this economy and the NEAR token that secures that economy, right?

[00:18:19]So you wanna really tie them together And so I think that's kind of, again, tokens are broadly speaking, are the easiest instrument for agents to use to transact.

[00:18:29]But at the same time, you do need the to tie them together as well on the kind, token economics way as well.

[00:18:36]So that's what NEAR is doing, for example.

[00:18:38]So, right now in this AI Agent You're saying that the role blockchain plays is now being incorporated in this way.

[00:18:45]Then, rather than this existing financial sector entering It would be much safer and a better fit to incorporate blockchain.

[00:18:53]Then, what is the role of blockchain in this era of AI agents?

[00:18:57]It would be great if you could define what it is a little more clearly.

[00:19:00]So, I mean, we're kind of using actually AI across the stack ourselves, right?

[00:19:04]So everything from building, the, like, all the coding now is AI, either done by AI or AI assisted.

[00:19:11]one important thing we're doing right now is So the kind of security Because of AI as an attacker, there's a huge problem because it's now easier and easier to find vulnerabilities, and so you really need to use AI for defense as well, right?

[00:19:24]So this starts with, like, security audits and kind of security inspection.

[00:19:29]But we also are running kind of an active analysis.

[00:19:34]So, like, whenever the transaction happens, whenever there's, we're connecting different chains, right?

[00:19:39]We're analyzing the chain itself to understand if there are hacks, vulnerabilities, issues that are happening and effectively try to either prevent or, or at least block the funds, right?

[00:19:52]There was, you know, multiple hacks that happened in past two weeks.

[00:19:57]So, like, really detecting those funds, ensuring that they are not gonna get, like, laundered, moved, et cetera.

[00:20:03]So We're using AI effective, like security assessment and security kind of system.

[00:20:08]And that is agentic AI, right? It's like it's agent that like goes and cross checks all the information and like does the research on this transaction, on this address, on this chain to find what is the relevant or not.

[00:20:19]And so kind of as you go up, right, the, you know, like marketing like starting to use more AI, right?

[00:20:25]So like same, same as we're building this marketplace for agents, we're also thinking of how do we transition our almost, kinda how we do work is for agents to be like coordinating and like maybe hiring other specialized agents to do work, right?

[00:20:39]So I would say like across the stack, like our legal team is using AI, our finance team is using AI.

[00:20:45]So like every team is effectively now becoming kind의 AI assisted or, or like AI first as well.

[00:20:52]Then, I am curious about NieR AI's current revenue structure and what additional revenue streams are expected in the future.

[00:21:00]Yeah, yeah So right now, the AI side, the business model is kind of twofold, right?

[00:21:07]One is we offering this private inference, so confidential inference.

[00:21:11]Again, businesses that need to ensure that their data is private.

[00:21:15]So medical, legal, kind of privacy-focused companies like Brave, Venice, they need privacy, and so we have kind of cryptographic and hardware guarantees around privacy So that is a very similar to, like, API calls for inference, right?

[00:21:30]So charging per million tokens. Per million inference tokens.

[00:21:35]Too many tokens yeah, so that, that is kind of one part The other part is we offering this kind of agent hosting and specifically Ironclaw as a service, and so you can have your agent, your organizational agent, organization agent, or embed this system into your own product. Right?

[00:21:52]And so we're charging kind of for that as well. Right?

[00:21:55]So those are effectively the income streams, right, for, for the Near AI.

[00:22:01]Now, the kind of blockchain part, right, as I mentioned, is actually charging based on transaction and based on moving value, right?

[00:22:08]So it's still, you know, there's still a lot of humans that are transacting as well as AIs, and so we're taking a percentage of depending on transaction value.

[00:22:17]So, you know, this can be, like, as simple as moving some USDC from one chain to another.

[00:22:23]As complex as, you know, paying for building a website or creating a marketing campaign, between agents, right?

[00:22:29]So then we're taking a percentage of that In the future, all cryptos will become like this with a solid profit structure.

[00:22:37]Now, it seems the crypto sector will survive, while the non-crypto sector will face some difficulties.

[00:22:42]I'm thinking I might move in the future, and I'm curious to know what you think.

[00:22:46]Yeah, I mean, I think this has been kind of where the crypto been converging, and we actually So we had a NEAR Con, the conference in San Francisco, in end of February, and we announced the kind of the change in economics where we are taking this revenue from the transactions and buying back NEAR Um, and so we have actually revenue.near.org where you can track the revenue and all the buyback and everything.

[00:23:13]So definitely agree that that is a really important component of kind의, the token be tied to the economic activity of the chain and of the products we're building.

[00:23:24]Don't want to misinterpret So, um, you said you buy back other tokens?

[00:23:29]No, no, we're buying back NEAR So transactions are happening, right? And the percentage of that transaction is the revenue.

[00:23:35]It used to buy NEAR token. And then it goes into kind of revenue bucket.

[00:23:40]And then we either burn that or use it to, like, generate more near.

[00:23:44]Yeah, I mean, at least, like, there hasn't been any other mechanism that works outside of just making revenue.

[00:23:51]I think there is a technical difference now.

[00:23:53]Is this a technique called Nightshade? I've heard of a sharding technique.

[00:23:59]I'm a bit curious about what kind of technology this is.

[00:24:00]So this was kind of our original, you know, when we started building NEAR protocol, the blockchain, I mean, originally, we actually wanted to just use the blockchain for our AI product.

[00:24:11]And we looked at a lot of the blockchains at the time, right, 2018, and we didn't find anything that would be, like, matching our requirements, right?

[00:24:19]So we didn't start with like, we should build a blockchain.

[00:24:22]We started with like, we should use a blockchain- let's find a blockchain.

[00:24:25]We, and we didn't find a blockchain.

[00:24:27]And so what were the requirements?

[00:24:28]Well, we wanted something that would be kind of easy to use for people, easy to develop on, but importantly, it should scale with usage, right?

[00:24:38]Like, coming from Google, coming from, you know, my co-founder's from Microsoft, and he worked at this company called SingleStore, we are used to- scalable systems, right?

[00:24:49]Google doesn't just like, "Hey, you know, sorry, you cannot use Google right now because we're too busy," right? Too many users.

[00:24:54]Like imagine you go to Netflix and Netflix says like, "Hey, we're too busy right now."

[00:24:58]That's kinda how blockchain Or it's like, "Oh, actually you need to pay more. Right now you need to pay more."

[00:25:04]That's how, right, kind of blockchains work right now, right?

[00:25:07]They're like, "Hey, we're too busy," or, "You need to pay more if you wanna use blockchain."

[00:25:12]And what we wanted is it just continues scaling as more users come, right? Billion users should still work.

[00:25:17]So that's what really Nightshade is about.

[00:25:19]Nightshade is about building a blockchain that can scale with demand, ca- scale with usage, and really expand so-called block space dynamically So, this is using all the same kind의 primitives that, you know, Google, Apple, Facebook, Meta use underneath, right? It's called sharding.

[00:25:38]It effectively kind of partitions the block the block itself into parts and the state space, right, the accounts, the transactions, between machines, and then guarantees that all of it is secure I thought there might be some people who don't quite understand.

[00:25:51]I think there should be an example of using it as a night shade.

[00:25:56]So Nightshade is, is more of a internal how blockchain works.

[00:26:00]So you as a The idea is that you as a user or agent for that matter, shouldn't need to think about it.

[00:26:05]It looks like a single blockchain, and it just, it will always be available, right? It always will have capacity. Right?

[00:26:13]And so the For people who are actually, like, digging in, right, you can see that our blockchain actually has been growing in size over past, you know, since Mainnet, over the past five years now.

[00:26:25]We recently actually shown that we can process million transactions per second on the blockchain because we can keep growing the number of shards.

[00:26:34]But you as user just have, like, have to use it and not worry about this.

[00:26:39]And that is because of the Nightshade?

[00:26:41]Yeah, because of the sharding technology that we Anyway, since AI infrastructure is expanding rapidly, I am curious how Near AI plans to expand in the future.

[00:26:51]Yeah, because of the sharding technology that we- Now that this is the case, since AI infrastructure rapidly, Near, Near AI, Near AI will be provide.

[00:26:58]So very similar how, I don't know, Bitcoin mining, right, just, like, went global, right, everywhere, you know, everybody participating.

[00:27:06]Like similar, that's what we really aiming for with kind of this infrastructure.

[00:27:10]And similarly, on the agent side, really we want to give every single person an agent. You know, 8 billion agents.

[00:27:17]So we need all the data centers for that to work.

[00:27:20]Ah, gotcha Then, I’d like to hear about your own dream, not the company’s. I’m curious what your ultimate goal is.

[00:27:28]I mean, my personal goal is I wanna ensure everyone has access to AI and it's, you know, it's theirs, it's private, it's- kind of aligned with them.

[00:27:37]It enables them to be the kind of economic actor at ability to participate in economy, right?

[00:27:42]It doesn't matter where they are, who they are, where are they from, et cetera.

[00:27:45]And so that's really how, like, ecosystem of NEAR is building these different components, but that really the vision is how it kind of create user ownership.

[00:27:54]Our channels have many people interested in economics and crypto.

[00:28:00]I'm curious how crypto trends will change in the future.

[00:28:04]I think the crypto's an interesting stage where the, the blockchain technology actually Like, before it was crypto was, as I said, right, like everyone was kind of speculating on crypto prices then the blockchain wasn't used.

[00:28:18]We're actually getting now and the blockchain is getting used, right?

[00:28:20]There's more and more stablecoin usage. Agents are using it. Big companies are starting to use in US, right?

[00:28:25]Hopefully Korea also passes the, uh, regulation soon as well.

[00:28:30]And so we're starting to see actual adoption of the blockchain as technology, and then the crypto prices, right?

[00:28:36]That's where the, the question of, like, this relationship, right? Is this about revenue? Is this about kind of security? Like, that's becoming a question.

[00:28:44]And I think, yeah, from my perspective, it's really tying together all those pieces where kind of- blockchain serves as a backend, as a marketplace, as this neutral infrastructure, and then the tokens kind of capture value from the revenue and from the transactions, and then in turn secure this network as well, right?

[00:29:01]So you need to have this kind of flywheel and, yeah, I mean, that, that will be the kind of definition of winning protocols in.

[00:29:09]If our channel films a video on the topic of Bitcoin I think this is the first or second time it has appeared in terms of views in Korea.

[00:29:17]I've been seeing a lot of people doing crypto lately, in their own way.

[00:29:19]Thank you so much for the wonderful time, and I look forward to seeing more of your activities in the future.

[00:29:24]thank you so much Thank you so much for interview Thank you for your time