Why AI Agents Need Crypto Rails: Ella Zhang on Wallets, Identity, Permissions & Payments

追加: 2026-06-02

[00:00:00]Zainab, thank you for joining. I will hand over to you for for the session with Ella.

[00:00:05]Thank you so much, Jessica. A pleasure to be here. Hi, Ella. Nice to meet you.

[00:00:09]Nice to meet you. Thanks for having me.

[00:00:11]So, of course, it's a pleasure. And I'm really looking forward to this chat. You know, we're going to be talking a little bit about the convergence of Web3 AI and new technologies.

[00:00:21]And, you know, you've been around the block for quite some time.

[00:00:24]Uh since 2018, uh I got into crypto quite early as well.

[00:00:29]Today, I'm the CEO and co-founder of a company called The Block. We are an ecosystem for tokenized assets out of Dubai. Uh we're expanding to Abu Dhabi, and we have an office in Vienna, in Austria as well, and looking to expand to New York as well. And the reason why we're really betting hard on tokenization is because we're really seeing the convergence between TradFi and DeFi happening now.

[00:00:49]And, you know, you've been uh co-founding Binance Labs since 2018. I don't think you really need an introduction cuz everyone in the space should already know you. But, for those who don't, uh Ella co-founded Binance Labs back in 2018. And, you know, you guys had a lot of successful startups coming out of your cohorts like Polygon, Aptos, Layer Zero, uh and and many, many others.

[00:01:13]And if you look at if you look back at, you know, the Ella back in 2018 versus the Ella now, what do you think you would have gotten wrong about where where capital will be flowing back in in in 2026? Cuz cuz in crypto, this is like looking hundreds of years into the future.

[00:01:33]Wow, spicy question from the start. Uh what could uh something wrong get back in 2018?

[00:01:40]Um I don't So, I would say that if we uh look at where we are today and looking back, 2018, we started our mission um at Binance Labs to help um the blockchain crypto community and ecosystem to grow um to the grow to the global uh markets. So, but back then um our mindset is only crypto, crypto is the biggest thing ever. Um and we only invest in crypto um blockchain.

[00:02:16]Um I think that that mindset is nothing wrong, but then we uh if jump out today, we see other industries uh especially like AI is taking off with no doubt. Uh if we're um back then think more about the holistic way of how uh the new batch of the innovations happen historically, the um the biggest innovation and industrial change won't happen at the single domain. W- when there's a new innovation, it's happened at the converging always convergence um with other um readiness of the enablement infrastructure. For example, um steam uh steam engine will combine with the railway to make the industrial uh economy happen. And electricity combined with the assembly line really make the mass production happen, right? So, blockchain today actually and AI um are enabling each other to become more and more powerful. Uh we um back then if we can make more bets um visionary enough, I probably will invest in more Nvidia.

[00:03:33]Early AI bets, that's what my reflection.

[00:03:37]Nice. And and now at the YC, you know, you have a broader mandate. What is what is really exciting you about this convergence? Is is there any specific vertical within this Web3 AI convergence that you guys are specifically looking at?

[00:03:52]At this convergence, yes. Um, I think there are kind of uh two dimension on convergence. We first uh think of the TradFi and DeFi is really kind of um uh coming to meet each other.

[00:04:06]Previously, we think DeFi is disrupting TradFi. Um, and also E and CZ, we we um we kind of discuss a lot. There's really no no such thing happen, right? Um, it's TradFi and DeFi. They both have their advantages and the legacies, and they have their their properties and come into enable each other, where we can see the um traditional stock is going to on chain and the crypto uh digital um treasury asset are going to be uh adopted on the traditional uh market in the DAT in um those companies like a MicroStrategy.

[00:04:52]Um, so the people of uh who trade usually in the TradFi in the traditional uh stock market or people only trade on chain or buying asset both side. And we're seeing uh stablecoin is really going to thanks to uh genius act and going to mainstream. And especially in those um frontier markets, um we can see over 50% of the penetration uh in the country of the uh of the uh stablecoin uh adoption. And there's another layer of convergence that we see the technology like AI, biology, um and blockchain are uh con con con convert uh converge to enable each other.

[00:05:39]For example, um AI agents nowadays is getting more powerful, but it's not would the only purely with intelligence is not enough.

[00:05:49]It need to execute, you need to transact, need to own assets, and need to gain the trust. This is where blockchain came in to empower AI agentic AI era. At the same time, AI really help blockchain crypto to solve the usability um issue where to majority of the people crypto is still very vague and quite complex in terms of use. But AI turned the interaction of in interface with the with the protocol with the wallet into intent base.

[00:06:27]Instead, users and agencies just need to tell um AI that's what we want and AI handles execution. So, uh move over over time we can see blockchain gradually become the invisible layer to ensure a lot of identity, uh incentive layer, and ownership where versus AI become the interface layer to enable the usage. That's a amazing convert convergence we're seeing happening.

[00:06:56]So, I I know that Binance is really, you know, going big on on on the agentic AI and and how money interacts with these AI agents. Uh even CZ, he you know, he went out and said that he he essentially sees the BNB chain becoming the money for the AI agents. And uh he also said that, you know, crazy cars don't have an API, which I find quite uh it it was a brilliant statement. So, we do see this shift. Now, if if we kind of go back to what you said earlier that uh money is flowing into tokenized products or institutional money is now moving more into tokenized products like T-bills, equities, etc. Tokenized equities, then you also have another aspect, which is the DeFi uh element. Uh and also you mentioned the settlements. And now we also have the AI agents. Out of these four buckets, which one would you kind of pinpoint to say this one I really believe is going to be the next big thing for this year?

[00:07:52]Yeah, obviously agentic payment is taking off get the adoption. I think this really starts from you talk about how Binance and BNB chain ecosystem Easy Labs were adopting agentic AI as a holistic view.

[00:08:09]This started early last middle of last year. Easy Labs start to invest in AI biotech as they expanded Monday from the beginning of 2025 and everyone is moving very fast into learn the new technologies. And CZ is really really the visionary. I still remember early last year he already write down a mind map where he thinks AI going to disrupt going to change Binance and BNB chain and there are 20-30 items and I still remember that he said how people trade on crypto will be different. The UI UX of Binance and on chain apps DApps will be completely changed and this is happening now. We seeing the um how the willing base the intent base decision making by AI agent. You actually don't need any interface.

[00:09:08]That's where we seeing AI is so powerful to empower blockchain crypto. At the same time the payments how to make the payment permissionless and trustable how to make users really trust the agent. Blockchain come the X42 the the all those kind of the protocols are ensured that the AI have the identity, have the wallets, um have the ownership of piece of your assets, which enable the the agent can execute for you.

[00:09:47]And also, there's another layer where we see the ownership happen, and data is something really lack in making in advancing AI. No matter in LLM or physical AI or in uh biology There's a need a lot of data, personal data, to train the AI model to be next level, and it's very sensitive. And blockchain uh uh based the ZK knowledge proof those technology will ensure that your data be shared for the training, but not reveal your identity. At the same time, you probably will get piece of the network economy because of the data ownership. So, that incentive of blockchain enable ownership incentive and security verifiable identity make AI will empower AI to be more more powerful. That's the area we invest a lot of time seeking for the next generation innovation and and applications happen.

[00:10:56]You say that that's a very interesting point, and you sit on a quite an interesting position, you know, heading Easy Labs.

[00:11:05]You get access to a lot of these new technologies that you privileged to see before everyone else kind of get to see what's happening on the market. Now, you guys have also You guys have also made some some very interesting bets on on Bana, TensorPix, Video Tutor, where AI is converging with blockchain.

[00:11:25]Uh where would you say these kind of convergences is are are kind of noise and where where where do you see it's actually working these combinations and where do you see you know when in crypto we always say there's a lot of products with a solution looking for a problem.

[00:11:45]So if if you can distinguish between those two where do you see that the actual convergence is is meaningful and valuable for the products and companies that people are building.

[00:11:57]Yeah, good question. So where we see we actually mentioned quite some applications and usability of where we see it's actually working where is the incentive layer, the ownership layer, the identity layer where blockchain really kind of make it work at the convergence with AI. Where we see it's a little bit maybe a little bit further away or not completely working or not working so effectively is for example the AI training. The training people want to put it on chain in a decentralized way however we're seeing the whole world building the data center and training the model in the centralized way. We're still seeing that um that usability is not there yet and the training inferencing in the centralized way is still the mainstream for for today at least.

[00:12:59]Uh we we also discussed before you know tokenization and I know that this is something that recently has become quite a hot topic for Easy Labs also for CZ.

[00:13:12]He recently at Davos he mentioned that you know Binance is in talks with at least 12 different governments on tokenizing the national assets. Now you guys also took a bet on Plume which you know that was your first kind of real, uh, I believe RWA bet.

[00:13:30]What Do you really believe that 2026 is the year where tokenized assets will be taken off, or are we still a few years ahead, or is it still kind of a pipe dream?

[00:13:41]Oh, yeah. We we do We definitely see very clear signal that, um, tokenized real real world asset is, uh, happening now, and we believe this is a probably very important catalyst for a crypto to go mainstream. Um, you can see that the transaction volume of the tokenized the stock is already taken off, um, gained the traction on on the on Binance, um, some other exchanges, as well. Um, at the same time, uh, we see the the demand of pre- uh, IPO stock actually kind of going to be trade on chain in a decentralized matter matter, especially for people who don't have the access to buy pre- IPO it because it's a private market, not everyone have the privilege to access even to the secondary, uh, shares. So, that's where demand people will, uh, kind of want to buy tokenized, um, SpaceX, tokenized the OpenAI, those pre- IPO stock. And at the same time, so Plume is the first batch of the RWA portfolios in our incubation, we also incubated projects that focus on early startup um, ownership. They make their, um, equity tokenized the equity to be owned, um, by the community. So, we're really seeing those valuable assets is moving online. Um, not only treasury, um, as a long kind of we have where last year happened a long time, but also other type of, um, credit credits that the the treasure the sovereign credit credit credits the real estate title those really valuable assets are gradually moving on chain. This is definitely the trend we see.

[00:15:38]Amazing and and you know we uh we here at The Block we're also sitting in Dubai and and and our ecosystem is built around tokenization and helping companies and we we see a lot of these projects coming on uh that want to tokenize assets. What would you say is still the challenge or the problem that hasn't really been solved for in order for these assets to really take off?

[00:16:01]Yeah uh good question. I think one major challenge we see is the custody solution for the tokenized stock and for the equity for even for the real estate titles to be on put on chain how to build the the the um trustless um on custody solution for people to really understand they have a piece of something really valuable tied to that um that asset is is where we see is have some gap and a lot of entrepreneurs are trying to solve that problem and also liquidity because it's the asset are so diverse diverse right? You always in order to enable the trading you always need enough liquidity to make the transaction happen and how to make the liquidity pool effective enough to support the very diverse asset class >> [gasps] >> to be traded on chain is another problems we seeing founders are very interested in and trying to to solve um Um, and also probably um some kind of follow-up issues. For example, we're seeing issues like pre-IPO stock maybe trained and when they go to IPO, the private market ownership will be liquid will will be liquidated at the liquidation event and how to make that transaction happen when the pre-IPO stock can go to go to a bigger liquidity pool on Nasdaq or New York exchange. And that's kind of price gap will be smooth. So, those kind of some very interesting problems we're seeing.

[00:17:55]Never previously never thought thought about.

[00:17:59]Amazing Ella, we need to wrap up. We started to reach the end of this conversation. Just one final question if you can keep it really short. You know, what is one contrarian bet that you would make for this or next or the next two or three years that you think if you made that bet, crypto Twitter would be mocking you?

[00:18:17]Yeah.

[00:18:18]Um, I I've been thought about this. I think um, previously our mindset is everything going to be issued token and everything web three will be tight with tokenomics.

[00:18:31]And now we're we're seeing the convergence. We see blockchain is really powerful technology and it will enable a lot of other technology happen at the same time. Not necessary everything need to issue a token. Instead, blockchain crypto going to become the very powerful enablement layer plug into our daily life, institution adoption. Not My bet is not the very valuable crypto and blockchain company for the next not necessary have a token. They can have a token tokenize the equity. But then Um, they all they can only have equity and we will see more and more blockchain crypto company go to IPO in the traditional stock market, not necessary have to issue a token.

[00:19:23]We're seeing the convergence from TradFi to DeFi.

[00:19:27]Amazing. I think that's a perfect answer to finish this conversation. Thank you so much Ella for joining us today.

[00:19:34]It was a true pleasure and I hope to see you in Dubai.