[00:00:00] Jonathan DeYoung: Crypto is for everyone, not just rocket scientists, venture capitalists, and high-IQ developers. Welcome to The Agenda, a Cointelegraph podcast that explores the promises of crypto, blockchain and Web3, and how regular-ass people level up with technology.
[00:00:24] Jonathan DeYoung: AI is everywhere. It seems like it’s inescapable at this point, from AI-generated art everywhere on my timeline to all the major search engines and cell phone manufacturers integrating AI tooling into their operating systems in newer generation products. The average person is still divided, I think, on whether AI should be everywhere, but the reality is that it is here, and there is likely no turning back at this point.
[00:00:49] Ray Salmond: Now, if you follow The Agenda podcast, you’ll know by now that there are several Web3 projects that are specifically working to decentralize and democratize the AI space. We’ve interviewed a few of them on this show. Each of them is taking on various aspects of AI’s centralization, but this week’s guest is looking to shake up the cloud computing market.
[00:01:10] Jonathan DeYoung: So, today, we are joined by Wes Levitt, who is the head of strategy at Theta Labs, which is the core developer behind the Theta Network. Theta describes itself as the decentralized cloud for AI, media and entertainment, and Wes is here to tell us why the cloud needs to be decentralized and how exactly Theta Network fits into the picture. So, thank you for joining us, Wes, and hope your day is going well.
[00:01:36] Wes Levitt: Thanks, Jonathan, Ray. Glad to be here, and thanks for having me on the show.
[00:01:39] Jonathan DeYoung: Of course, it’s our pleasure. So, why don’t we kick things off from the high level for somebody who has zero clue what Theta Network is? Give us a high-level overview of what it is and where it fits in the Web3 space.
[00:01:52] Wes Levitt: Sure. Theta Network we describe as, easiest way is as a decentralized cloud service provider. And so, what that means in practice is a lot of things that enterprises or even individual users would normally go to for centralized cloud services, like your AWS or Azure, you can get via our decentralized network that’s comprised of nodes around the world that provide the services you’re looking for. So, whether it’s streaming video, compute for AI or encoding, transcoding, variety of things. For pretty much anything that can be done on a centralized cloud service, we can provide over a decentralized network. The biggest advantage of that is our network is primarily made up of users around the world that are providing excess capacity, whether it’s CPU, GPU, or streaming that’s effectively unused if they don’t provide it to our network. So, it can be done at more competitive prices and also spun up or down and scaled up and down easily compared to a centralized provider that does it via contract.
[00:02:47] Jonathan DeYoung: So, what is the experience like? If I’m a customer and I’m comparing AWS to Theta, what is my user experience or onboarding experience like at Theta Network? And how does that compare versus if I were to go to a centralized service provider?
[00:03:03] Wes Levitt: Yeah, that’s been a huge selling point we found with many of the academic customers we signed up this year is that it’s quite simple with Theta. You can start in a matter of minutes just by creating an account and then picking which devices or GPUs you want to use for your jobs, for example, and then getting to work. You can do so with, say, AWS, for example, but at almost extortionate prices compared to what you get for a long-term contract. With Theta, you’re able to get the same level of ease of use and onboarding in minutes, but with competitive pricing we can offer without a multiyear contract.
[00:03:37] Jonathan DeYoung: Got you. Got you. Looking at Theta’s website, I noticed, as I alluded to in the intro, there is a heavy focus now on artificial intelligence and providing the cloud computing capacity for AI systems so that they don’t have to go to an Amazon Web Services, as you just mentioned a moment ago. Theta’s been around for a while, as I understand it. So, have you always had in mind this idea of being a part of the AI ecosystem, or was this something, a new use case that has emerged in the light of the kind of mass commercialization of AI?
[00:04:13] Wes Levitt: Yeah, it’s a little bit of both. Maybe our backstory would kind of help answer that question. We actually started building Theta originally, we had an e-sports video platform that the company had developed, and basically looking for a way to defray the costs that we were incurring as a startup, paying 50% of our monthly expenses going to CDN. So, the original vision in 2017 was to basically create this decentralized network to provide streaming services. And then, along the way, once users are already creating nodes on the network to do this, it made sense for them to provide compute as well for the ingest, for the transcoding, encoding of video as well. But that was sort of the North Star of what we were doing, is building around this use case.
We had talked about powering AI use cases as well, even back in 2018, 2019, in various white papers and blogs. But to be frank, we thought this was a potential use case years and years in the future and were sort of caught off guard, like probably a lot of people were, when ChatGPT came out, and it felt like instantly the sentiment had changed to where this is where a huge amount of the demand for these cloud services are going to come from. So, it was somewhat on our radar, but in some ways, we were fortunate enough to have built out the network and had tens of thousands of nodes up and running and capable of doing this AI work. So, pretty quickly afterward, when we saw that being a huge new source of demand, we already started well-positioned to do that. And so, then we could go to different clients that we signed on to use Edge Cloud now and basically say like this has been tested over years. This is a working product. And really, it was just a new source of demand rather than having to change much about how Theta works.
[00:05:49] Jonathan DeYoung: You mentioned Edge Cloud. That, I assume, is the AI-focused part of the product?
[00:05:54] Wes Levitt: It’s really just the term we use to encompass the decentralized network of nodes that can do the productive work on it. So, when you think about the name, it’s users around the world on the edge that can respond quickly to work being pushed to the network, that needs to be done, but then also encompasses cloud service providers that we have added to the network as well. So, there are some large data center providers, those that we basically have backfilled to help add to meet the needs of some of these clients that have come on.
[00:06:21] Jonathan DeYoung: So, some of the other projects that we’ve interviewed, such as SingularityNET, OriginTrail, have a very like philosophical approach to decentralizing AI, and they see centralization of AI as a threat, in the case of SingularityNET, potentially to humanity. In the case of OriginTrail, to our privacy, safety, data integrity, things like that. So, does Theta Network approach the topic of AI with sort of the mission of decentralizing it for any particular goal or motivation, or is it more so the mission is decentralizing cloud computing for your mission on that, and then AI kind of just falls into that naturally?
[00:07:01] Wes Levitt: I think more the latter. We’re generally looking at decentralizing these cloud services to get away from having only a few points of failure and from having near-monopolies in the case of the major cloud providers, and what they can provide in pricing and other terms. I don’t know if it’s as philosophical for us. It’s just in practical terms, we’ve been on the other end of it, where we felt crushed under our options for using cloud services as a small company. So, we wanted to fix that problem for ourselves, and we’re happy to help other companies in the same situation. Where increasingly so for us, it’s been academic customers who, they don’t have the weight of a Fortune 500 company to be able to use cloud services, so they find that very beneficial that they can use a pay-as-you-go, scale-up-as-you-need, scale-down-as-you-need-to service, and at good pricing.
So, yeah, I think there is a positive mission in what we’re doing, but it’s not as much of thinking about Skynet coming down if we have centralized AI. That might be a risk, too, but it’s not what’s driving it.
[00:07:57] Jonathan DeYoung: Gotcha. Yeah, that makes sense.
[00:07:58] Ray Salmond: Okay, so I was wondering, how is business since the advent of and kind of commercialization of AI? I’m wondering if all the new players that have arrived to the space eat away at the client inquiries that a company like Theta might receive.
[00:08:14] Wes Levitt: Yeah, there’s an enormous amount of new clients in the sense that virtually every company and institution around the world now has some demand for AI services, or in the case of a lot of our users, training and testing models. So, that helps, too. Of course, a lot of new companies and protocols that spring up to meet that demand. But I think, on balance, it’s been positive. And the service that we provide and the connections we make have helped a lot in that. I mean, we work directly with the professors leading the AI groups at nearly a dozen universities now. And I think being able to work directly with them and get their inputs and improve Edge Cloud on the fly to what they need to add, say they need more persistent storage, or we learn how to improve it based on that customer feedback in real time. I think that is making a big difference. And they’re telling us that’s why they chose Edge Cloud in a lot of cases versus some of the other providers. So, there’s still room to stand out and be a leader in the market. But yeah, there’s no doubt. Everyone is trying to move into this both on the supply and demand side. So, there’s a lot of chaos in the market right now. Positive destruction or creative destruction, as they say.
[00:09:17] Ray Salmond: Yeah, that makes sense. And another question I had is why do you need a blockchain and a token to decentralize computing? It kind of seems like you could theoretically offer the same sort of service without it. So, what exactly gets stored on the blockchain?
[00:09:33] Wes Levitt: Yeah. So, on the demand side, for now, a lot of it is white glove service and bringing on new customers. That’s not really our end goal. We don’t want us to have to be doing that. Even now, any user can sign up and use Edge Cloud themselves without having any contact with Theta Labs. It does operate as a decentralized network. But it’s actually on the supply side of the compute resources it’s more important that we don’t want to or have the ability really to coordinate tens of thousands now, and hopefully millions in the future, of devices about how they’re contributing these excess resources to the network and how they’re being used on the other side. We really just want to build a protocol that’s basically going to matchmake and have the supply meet the demand, and we don’t intend to sit in the middle and play accountant and coordinator of that in real time as millions of devices have to communicate with each other and decide who pays what microtransaction to what. I’m not even sure if it would be possible, but certainly, that’s not what we want or we’re trying to build. We’re just trying to set down the rules basically on this network. It’s a marketplace, effectively, for these resources.
[00:10:35] Jonathan DeYoung: So, do the customers that use Theta Labs, do they have to like purchase Theta tokens to pay the providers of the cloud storage space? Is that how the transactions happen?
[00:10:47] Wes Levitt: They don’t have to, although they get better rates if they are to do so in the terms of Theta Fuel token. That’s something that comes up frequently, and I think it’s the right course, given so many of the customers dealing with are not necessarily crypto-native. And it’s very difficult talking to some of these customers to tell them you have to buy our token just to get in. One, because of the friction just in buying it to begin with, but then they don’t necessarily want to hold credits in something that’s going to change in value. Certainly, some already are, and to the ones that don’t, we simply just take it and buy the token. In effect, use it as a pass-through, which I think is pretty common, I’m finding, among other DePIN-style projects. But we do want to create an incentive for users to use TFUEL to purchase a product, because if more do, then that’s going to stabilize and have more consistent pricing over time. And it makes more sense for the providers of the resources to begin with, because that’s what they’re earning on the back end. So, we want them to see value in the network growing as well.
[00:11:44] Jonathan DeYoung: Do any of your customers have regulatory concerns about the idea of purchasing cryptocurrencies to pay for something when they could go to Amazon, which they know is like probably super cozy and comfortable with the government already?
[00:11:57] Wes Levitt: Yeah, we haven’t had as many of those. I felt like in the early days of protocol, there was more of it, but I think that’s shifted a bit. There’s still some that had... And there were a lot of accounting concerns, even more than regulatory. But just holding cryptocurrency on a balance sheet generally wasn’t a huge regulatory risk for a lot of them. Now, we haven’t dealt with any government clients, so that could be different in that case. But for the private sector, we haven’t really noticed that.
They do have a lot of concerns about data privacy. So, and it depends on the use case. If you’re training a new model and there’s no sensitive data, then they’re less concerned. There’s some healthcare-focused research being done, for example, at Korea University using us that they had to wrap their heads around what’s the actual flow of data to make sure nothing sensitive is. But they’re not using unencrypted data sets with users’ names and medical histories on them to begin with, and then it’s just a matter of like discussing security concerns and how that works as a network. But that’s a solvable problem. It’s just something that they have, you know, of course, they have to do their due diligence and understand that.
[00:13:00] Jonathan DeYoung: It’s actually a perfect pivot into another question I had, which was, how does the security and privacy work on Theta Network, both for the customer when you’re giving your data to people? Like, if I’m a node runner, can I read the data that I’m hosting? Is it like completely hashed? And then, as a node runner, what sort of security do I have to know that like somebody can’t hijack my node or something to run some sort of malicious code or something like that?
[00:13:31] Wes Levitt: Yeah. So, to the first part, nothing’s being said in plain text. I mean, I guess in theory, if you want to, as the end-user of it, if you could provide something to that, but there’s no way to unwrap a packet as you’re running an Edge Node and basically see the raw content of it. I can see the concern on the other side of being worried about what code is being pushed on you, but that’s also, it’s really just a matter of your own sort of security because Edge Nodes run locally. So, like if you have some intrusion into your system where someone can change the code of the client itself and allow somebody else into it, then maybe that’d be a risk. But I mean, I think that’s kind of on the user to do that. You have to protect your own device. But through the Edge Node client, no, there’s no way for something to leak out to then basically access your device.
[00:14:16] Jonathan DeYoung: Yeah, that makes sense. Theta Labs is obviously a company, but if you’re onboarding people directly through Theta Labs, if they reach out to you directly, an enterprise client or something, I’m assuming there’s some sort of like vetting process where you’re making sure they’re not using the storage space to host pirated movies or something like that. Tell me if that’s correct. And then, on the decentralized side, if someone were to just go directly to Theta Network without interfacing through you, is there some sort of community governance in place or something to be like, oh, this person is using the network for something malicious. We want to boot them off.
[00:14:54] Wes Levitt: Yeah, that’s kind of hit the nail on the head there. If we’re working with customers directly, then we’ll try to get a sense of what they’re doing. I mean, generally, with the customers we sign on, we’re not overly concerned about them. It certainly could happen, but major university clients in their AI labs, we’re generally not concerned about that. But no, it’s a valid question, though. I mean, you don’t know who they might be using. Or maybe someone that’s in there, they don’t know it themselves. Students at the dorm who are using the university services for something they’re not supposed to. But whether it’s them or just an anonymous user on the network that we don’t have any relationship with, we do have the ability to stop a user from using. We can’t access what they’re actually streaming without actually seeing on the end. We can’t see the data that they’re streaming in real time across the network, but we can see the end product if there’s enough circumstantial evidence or someone can point to like, hey, this is where pirated content is being streamed, and we can check it against the IP address, and realize that’s our client that’s doing that, we can blacklist them in that case.
I don’t think we’d be able to ever fully get rid of that. I’m not sure we’d want to, but, you know, there’s certain types of illegal content, and it’s just not a viable option as a US company to say we have no way of stopping someone from putting this on the network. That network will not survive very long. Put it that way. But I think that’s a risk to any streaming provider. I mean, a lot of illegal content gets streamed over AWS every day. It’s a bit of a whack-a-mole game, but when they find out, they shut it down. It’s not a via community governance, but that certainly could be in the future, just as long as it’s doing an effective job, since it’s pretty critical.
[00:16:29] Ray Salmond: So, do you foresee a future where a project like Theta is as big and widely used as, say, like an AWS or a Google Cloud?
[00:16:37] Wes Levitt: It depends a little bit on how the use cases evolve. And there’s always going to be some that, just like anything in crypto, you talk about this spectrum between centralization and decentralization. There’s some use cases that will be more effective in a decentralized environment. That doesn’t mean, or at least in our opinion at Theta Labs, we don’t have stickers on our laptops that say “decentralize everything” where, from a philosophical standpoint, even if a use case is better served by a centralized solution, we want to decentralize it anyway. That doesn’t make sense to us. So, I think there’s always going to be some use cases. Maybe it’s training certain models where it doesn’t help to have a federated learning model spread on nodes around the world. In this case, it needs to be on one cluster of nodes being done in one geographic location.
So, AWS is not going to be killed off, or the centralized cloud players aren’t going to cease to exist. There’s going to be centralized use cases. Depending on how things evolve and how a taste for decentralized AI develops, it could be just as large. I think that will kind of come down to some of these problems you alluded to that someone like SingularityNET talk about, which seem a little bit pie-in-the-sky now, could very well be a problem five years from now, or 10 years from now, where there is much more advanced model developed under centralized corporations, and there’s serious concerns about how powerful this is or what kind of knowledge it has, and should that be only controlled over via one company that doesn’t have to share any source code if they don’t want to. Then, I could see very quickly people moving to the centralized vision for that, especially if something goes wrong.
I mean, now we’re getting really into speculative territory. I don’t know what that involves. I don’t know if ChatGPT fires off a missile somewhere connected to the DoD grid. I don’t know what scares people enough, but something could, in more realistic terms, something that involves like a big problem in banking or communications because it was assigned to a closed-source AI model controlled by a single corporation. That kind of thing will scare people pretty quickly, I think. And then you could see use cases like Theta that could even rival those, because people will finally realize, like, this is not something that should be fully centralized.
[00:18:43] Ray Salmond: Right. What type of customers are using Theta Network right now?
[00:18:47] Wes Levitt: So, the strongest inroads we’ve made in the last six to 12 months has been in academia, actually, which wasn’t our first expectation, but we caught on with KAIST, a university in Korea, and pretty quickly we were able to expand to four out of the five top Korean universities just because we realized very quickly something is very advantageous about that, is that researchers in academic institutions collaborate constantly with folks at other universities. And word spread quickly, and we were able to make connections throughout. And we recently signed on University of Oregon. And talking now to a few of the US universities. And I think we can recreate that blueprint because there’s a lot of connections between different academic institutions. We signed on some private sector clients as well, like Liner, which is an AI search engine company, and then also Jam Coding, which does software development education. I think there’s definitely some interest there as well. But right now, the academic sector is where it feels like we really are gaining the momentum and found product-market fit.
[00:19:45] Jonathan DeYoung: Before you mentioned that earlier, that would not be something that I would have assumed would be your, I don’t know if it’s your biggest customer, but your biggest chunk of new customers. But it makes sense, right? I mean, a lot of the AI research is being done in academia, in universities. So, obviously, they’re going to need... And universities tend to be cost-conscious, unless you’re like at one of the premier universities.
[00:20:07] Wes Levitt: Sure. Yeah. There’s also some other interesting, idiosyncratic things we found about how they operate. Like they have some baseline research that they’re doing constantly, but then they have dates where academic papers are due before conferences. And it’s like a crunch, like if you’re familiar like in game development where it’s weeks at a time, 24/7, all hands on deck, where they need to scale up their usage very quickly. And then, the week after that, they may scale down a lot. It’s not a model that really fits for them to say, we’re going to sign a three-year contract with heavy commitments on it. They want to have flexibility and be able to use a lot when they want, and not pay for times when they’re not. So, not something we were fully conscious of, but that turns out it’s a great fit for this type of model.
[00:20:52] Jonathan DeYoung: When you’re selling an academic or really any non-Web3 customer on Theta Network, do you lead with, you know, we are a blockchain Web3 cryptocurrency project? Is that something they care about at all? Or is your conversation primarily about, oh, we have better uptime? It’s cheaper. And oh, by the way, if you’re curious, somewhere in the fine print with an asterisk, we also do crypto.
[00:21:20] Wes Levitt: Yeah, that’s right. The three most important things for them, at least generally, it’s price, ease of onboarding, and then ease of use generally. And then maybe throw in that they like that we’re very responsive as a team and work with them as partners to develop the product on an ongoing basis to what works for them. Occasionally, or 20% of the time, big into crypto as well, so maybe that’s a bonus for them. But I don’t think it’s core to the product on the demand side. They just want something that works at least as good as their Web2 experience.
On the supply side, it’s obviously very important to users because they don’t have a way without using a blockchain to contribute these resources so easily and be paid for them. So, I think it’s more key on the supply side that node operators care about that this is a crypto. But no, on the demand side, I think it might be a bonus to some people, but they want a product that works, and they want it to be at least as good or better than their Web2 alternatives.
[00:22:13] Jonathan DeYoung: So, one of the critiques or concerns around artificial intelligence, in particular, is the vast amount of energy and computing power that it takes. It’s been attributed to like contributing to chip shortages. And people are saying it’s already eclipsed the amount of energy that crypto consumes. So, and it’s seemingly only growing, getting bigger, and going to require more and more both energy and computing power. So, as the head of strategy at Theta Labs, what is Theta’s strategy to be able to continue to grow the network and the capacity to be able to keep up with the growing demands around AI?
[00:22:56] Wes Levitt: Yeah, I think that’s where we can, because we’re able to tap into these pockets of users that aren’t really on the radar for, let’s say, major data centers or for the major cloud providers, we can access compute power that isn’t necessarily on the network already. Sometimes that’s high-end GPUs, but in some cases, it’s a 3090 graphics card, which certainly there’s a lot of model training on our network that you can’t get done with that. But there is some that gets done. And for a more cost-conscious user, they want that flexibility to say, I’m going to use the A100 for this job. Now, this isn’t as time-sensitive. I can do it even on a gaming graphics card, and it’ll take longer, but I’ll get it done cheaper. And so, we’re accessing a lot of this compute. It’s not accessible by any other means. Like it’s not being aggregated by a major cloud provider. It certainly is not being operated by an AWS or an Azure.
More broadly, it’s going to be an even bigger concern. My wife actually works for one of the companies that’s a heavy demander of these services, and it’s something that they run into all the time, and all throughout society. It’s not even within tech dealing with different countries and governments. Big concerns about how much is being used for this. But these are generally still free markets. So, if it’s not provided by the general data center population, it will just be built by companies that need it. I don’t think there’s really a way to stop the demand from being met. Someone will build the data centers and build the GPUs to meet that demand unless there is some type of legislative reason why they wouldn’t do so.
I’ve even seen, I can’t recall the name, but there’s a startup in California that’s basically they’ve developed their own GPUs. I think sort of Frankenstein. And they’re not quite on Nvidia level, but they can meet some of the lower-grade AI use cases. And they’re sort of trying to fill that middle ground. Maybe Meta is not going to use it to train their latest model, but these thousands and thousands of companies and schools that are out there that need access to GPUs, I think they have found a way basically to meet that segment of the market. And I think that’s just going to keep happening. And it’s only been a year or two since this explosion. But if there’s a need for compute, people will find a way to provide it.
[00:25:03] Jonathan DeYoung: Do you think, though, that there’s a potential bottleneck in terms of the consumer devices being able to provide this compute that Theta taps into? Like in the future, might the Theta Network start tapping into larger, more commercialized... Maybe a better way to put it is, in the future, might a larger, more commercialized, more highly equipped computing center join the Theta Network, seeing that there’s a demand there? Or do you think that maybe that the rate of consumer, the increase in power in consumer devices’ computing power is enough to, because it does gradually increase over time, that it will compensate for the increasing demand?
[00:25:44] Wes Levitt: Yeah, we are seeing there’s definitely a trend up in the power of mobile devices. And we actually just launched a mobile version of the Edge Node about a month or two ago that can process some lower-tier jobs. Like it can’t do the core model training, of course, that needs Nvidia GPUs for it, but some things, like an AI model we demoed on it is object recognition videos. Or it could do text-to-video, video-to-text summaries, that kind of thing. So, it can do that today. In a few years, on mobile devices, maybe you can move up the food chain and do even more so. Obviously, some of the use cases are going to continue to grow in their GPU compute requirements. So, that will continue to outpace... Even if a mobile device makes some steps up, then it’s never going to do the highest-tier jobs.
But I think it’s, our approach is just be ready to take on any and all devices that can contribute to that network. So, we have talked to some data centers, that is on our mind as well, for ones that don’t want to necessarily work with basically one of your big three cloud providers or don’t want to sell the data center to them, that we can be another option where they can effectively lease GPU to them that is used on the network for these use cases. So, any and all, I think. And all of that’s possible from a technical standpoint. I think it’s really more just understanding does it make sense from a business standpoint, and can we fit them with the right jobs that are being demanded on our network.
[00:27:04] Ray Salmond: So, a little bit earlier you alluded to regulation and legislation. Do you think our current legislators and world leaders are capable of regulating AI effectively and also appropriately?
[00:27:17] Wes Levitt: I don’t know, that’s a tough question. I worry in some cases that these models and these networks they’re building are, sometimes there’s only a handful of people in the world that even truly understand it, and they’re a lot smarter than the politicians regulating it. And I don’t even mean that as a knock on the politicians. Like not saying they’re dumb. It’s just these are such advanced topics, and the implications are so broad, I think it’s going to be very tough to regulate it. I know here in Europe, where I’m based, they’re already starting to take a more heavy-handed approach. Which I get respecting the power and the danger of these AI models being developed, but that’s sort of a knee-jerk thing for Europe to do is just, you know, try to clamp down on it as soon as possible.
It’s a balancing act. And I also like that the US is, at least for now, letting it flourish more so we can advance as a technology. But I’m not sure I fully understand at the top level enough to know which is the right approach. We’ll know depending on if the Skynet scenario comes to pass. Maybe we should have regulated it more. It’s, I think, being fiercely debated, you know, probably among a lot of important circles right now.
[00:28:18] Ray Salmond: Yeah, I agree it’s a tough one. Everything’s like done after the fact or very heavy-handed up front, isn’t it?
[00:28:26] Wes Levitt: Yeah. Many things in history, we may have to have something go very wrong first before it’s kind of a wake-up call, right?
[00:28:33] Ray Salmond: So, if you could go to Capitol Hill and speak to, like, senators and other members of Congress about creating a framework for regulating AI, what would be kind of your three main points that you would emphasize?
[00:28:47] Wes Levitt: Well, the first thing that comes to mind is like rather than a regulatory body that’s sort of coming from an antagonistic point of view, you start with something... Like in the finance world, in the US, they have FINRA, which, if I recall correctly, is a self-regulatory organization amongst finance professionals, but it sort of interfaces with the government regulators, and they can come together and create some crafts, whether it’s a regulation or prevention of crises or reacting to market crises and working together on that. So, you have a balance of viewpoints of the private sector, which knows what’s happening, usually much more deeply and intimately day to day, and the regulators who have more of an interest in society at large and protecting the people in their country and their jurisdiction, rather than just protecting their own financial interests.
I’d like to see bodies like that come together where both viewpoints get heard. I don’t think the private sector should just run wild and do whatever the hell they want, but I also don’t want the government to take the approach like the US has done in crypto for the last four, eight years or however long it’s been, where they just want to smack it down every chance they get, and they don’t really care to hear much from industry leaders. I’d want the most important decision-makers on both sides to get to know each other, develop mutual respect and deeply understand the issues so they can work together on it instead of fighting each other.
[00:30:06] Ray Salmond: Does the fact that Theta is a decentralized network, or that computing power is sourced in a way that’s not centralized, does that raise concern for clients, or does that raise any sort of red flags from a regulatory standpoint?
[00:30:20] Wes Levitt: For clients, it hasn’t so far, so much on the crypto side, just the data concerns like we talked about. But that’s a solvable problem. In fact, we’ve actually beefed up how that’s compartmentalized and monitored and everything as a response to some customer concerns. So, it’s that back and forth that helps us improve the product that’s ongoing on that side as well.
I don’t know on a regulatory side. I haven’t heard any concerns about it. I think they’d be probably most concerned about what you guys are talking about, the type of work being done. And I don’t know if that’s a decentralized versus centralized issue. Maybe it’s just about what tools do we have in place. And I don’t think monitoring that and being able to prevent illegal type of work being done would be hampered by it being decentralized compute. But if it becomes large enough, certainly there will be more scrutiny on it, and it’ll probably be more things like that to work through. But we’ll do that as it comes.
[00:31:13] Jonathan DeYoung: One of the selling points of Theta Network, you were mentioning earlier, that the downtime of the network is going to be lower than on a centralized network, or the risk of downtime is lower than you might see on like an AWS where if it’s down, it’s down and for everybody, and it like completely disrupt the entire world, as we’ve seen from certain network downtimes or faulty patches being pushed to software or whatever the case may be. So, what are Theta’s customers concerned about when it comes to downtime for AI models? Or, I guess, like a streaming service, I understand that your customers don’t get to watch TV or whatever the case may be. For an AI developer, what is their concern when it comes to network uptime? Why is it important to them right now? And then, I can see in the future if like you’re interacting with an AI chatbot, like the city’s official AI chatbot, and you talk to it, and it directs you to 911, or it tells you to call the fire or the police or whatever. I can see why it might be important to have reliant uptime for something like that, but what are people’s concerns with that now? Is it those deadlines that the academics were talking about?
[00:32:31] Wes Levitt: Yeah, and it’s been less about small amounts of downtime but more about persistence. So, if they’re doing training of a model, and it’s getting a little bit beyond my knowledge of how actual AI model training goes, but I do know that they need to have the training data that’s consistently been training on over that period of weeks, or sometimes even months. They can’t just shut it down and start over. Otherwise, they really are starting over. So, it’s more about persistence via storage. That’s their biggest concern. If there’s major downtime, of course, any customer is going to be unhappy about that, especially if they have a deadline coming up. But that’s not as much of a concern in theory. I’d be most concerned about a software update or a patch or something like that that’s put through. Of course, we thoroughly test this like any other software developer, but things happen. And in that sense, almost any decentralized network is vulnerable to that, just like a centralized one. You’ve seen like how a blockchain upgrade can go wrong, and that’s, I guess, not centralized or decentralized, but just software has to be tested before it’s pushed to prod, and even sometimes it does, in the best of intentions, things go wrong.
I think redundancies are probably the answer to that. Like in the case of what we’ve worked on streaming video use cases, which is sort of the core of what we originally built it for, it was just about having multiple instances in place, and it often times even using a hybrid approach where the decentralized data network is the primary for streaming. If there is an outage, for some reason, it immediately can flip to a centralized CDN that sits basically as a backup, and it might be less efficient or more expensive. But in those cases, you’d rather have the more expensive CDN be turned on for a short period of time than have buffering and have the user pissed off and leave your site. Something similar to that, probably, as long as the persistent storage holds and has backups, then you can quickly flip to another service if you need to.
[00:34:18] Jonathan DeYoung: Yeah, that makes sense. I want to pivot here to a topic, as much as I love AI, perhaps closer to my heart, which is Theta Network’s role in the entertainment industry for creatives. So, can you give me a rundown of what sort of work you do with like entertainment and entertainers? And like on your website, you mentioned something about NFT-based digital rights management, so maybe you can touch on that as well.
[00:34:45] Wes Levitt: Yeah, yeah, AI’s obviously become the hot topic, and it’s generating most of the work being done on our network, but streaming is what we originally created, and a lot of our investors, like Creative Artists Agency and these, are from the entertainment industry. It’s where a lot of our DNA is, and it’s still something we think the Edge Cloud can be useful for, even if, at the moment, it feels like everyone is AI all the time. So, yeah, we worked with a number of customers that used the network to stream on their sites. For example, like World Poker Tour streamed some of their matches over our decentralized network, and they can use NFTs as basically a DRM system where if you buy access to it via this NFT, then you can watch this stream or this VOD, and it’s transferable. Obviously, it’s an NFT. So, it’s almost like an ownership pass that you can transfer to another user. So, less the model of Netflix where you just stream it but you don’t own anything, and more like you used to buy the DVD. And if you want, you can give it to your friend or sell it to someone.
So, I think that’s a super interesting use case. I like to see it expanded to ticketing, which we’ve had some startups demo that were building in our ecosystem. And I know it’s something we’ve talked to some folks at the major ticketing companies, and it’s something they’re interested in. I think they’re a little concerned about more value being pushed to the user, maybe a little zero-sum and might take some value away from them. So, I think that causes some skittishness. So, I don’t know if it’ll come from within, but I think someone should do it and disrupt that, because if you follow the lawsuits of some of these big ticketing companies, they’re pretty shady in their monopolistic practices. But they’re so super interesting use case that I think has, even if some of the AI is the shinier new ball, something interesting seem to work on.
And even on the AI side, some also on the entertainment side, or at least some more in that field. Like the Vegas Golden Knights hockey team in the NHL, they launched a chatbot that runs on Edge Cloud. You know, it’s a little bit more of an entertaining thing for them. They’re not doing cutting-edge research, but like their fans going there onto the Golden Knights site, and then they use it. They can ask about what’s going on with the upcoming schedule, or I want to know this about the player, or where do I find this merch from you guys that we really liked. So, besides, like the hard academic research, there’s still some fun customer-focused things that can be done. And a lot of these media partners we worked with in the past on the video side, they’re exploring all this too. They’d almost be not doing their jobs if they weren’t figuring out how they can use this tech in an entertainment context. So, I think more of that will be popping up, too.
[00:37:12] Jonathan DeYoung: Do you have any ideas for or examples of how the music industry could use this sort of technology? Obviously, you mentioned tickets, which obviously could also work for live concerts or live-streamed concerts. Taylor Swift could sell an NFT pass to her Eras Tour VR experience, or whatever the case may be. But yeah, any thoughts for how the music industry might be able to benefit from this?
[00:37:35] Wes Levitt: I think it’s not so much for the music itself, because we’ve talked about streaming use cases, like for music streaming, like a Spotify-type service. And it certainly works on a technical standpoint, but there’s not really a business use case for it just because it’s not very data-intensive. So, sure, you could save some money by doing it potentially over the network, but it’s not enough to move the needle. The folks like their major... Unlike video platforms where it’s very expensive to move high-definition video, for audio, it just doesn’t matter as much.
The things you’re talking about around that, like VR experiences or live-streamed concerts, that’s probably where it makes most sense. And from what I understand, the music industry, in general, the reason why concert tickets have gotten so insanely expensive is because they sort of shifted more to this experience model and live music because there’s no money in record sales anymore. Fractions of a penny on Spotify. And it’s just the whole industry, even if you aggregate it, the amount of money spent by consumers is far less on music than it used to be. They’d spend hundreds of years on CDs in the 90s, and Spotify is seven bucks a month or something like that. So, just like on a per-consumer basis, that’s not where the money’s getting made anymore.
So, that could work out good in the sense of for a new use case for a service like Edge Cloud. Because the more it goes to these experiences, the spectacle will go up. I mean, you’re seeing like a Taylor Swift concert, but I heard that you kind of get your money’s worth. Like the production values are a lot higher than the shows I might have gone to in the past, so I assume that means more people will want to watch those, and there’ll be more of a culture of watching those live experiences. Even if you can’t go to one at $2,000 bucks a pop.
[00:39:10] Jonathan DeYoung: Yeah. I forget who it was that we were talking to, Ray and I, with a previous guest. We were talking about the Apple Vision Pro, which had just come out. But we were talking about the Apple Vision Pro and whether or not like we could foresee a future where the VR experience becomes a truly marketable one, where like if you can’t see Taylor Swift in person, you could sit in the front row, you could sit on stage, you could perform on stage with her with the VR glasses, and it might really feel like you’re there. So, whether or not Theta Network is involved in that, I do think that is potentially a really interesting use case.
[00:39:49] Wes Levitt: Yeah. No, I agree. It feels like the experiences in VR are kind of like a stepwise function where, like five years ago, I tried out one of the first headsets, and it blows your mind for the time, but then it’s like once it sets in, you’re like, it’s not that great of an experience. And then the last generation, where it’s like two years ago, it was a lot better, I was impressed. It’s still not perfect. So, you’re not going to like, spend all day doing it, but it’s getting a lot better. There’s probably a tipping point where the experience is just going to get better and better, where it’s very light wearables. It’s not bothering you like wearing a headset, and maybe the resolution and the latency get so much better that you feel more and more like you’re there. I can totally see that being something that grows and something that if they have to move lots of data to do that, well, then they need a cloud provider to do so, whether it’s data or another one.
And we actually, in that initial iteration of the video platform we built, that was one of the things, interestingly enough, that we have patents for where we developed VR software for gaming, where if there’s a Counter-Strike match, then it renders in real time what it’d be like if you were on the map with the players, and you could walk, change your viewpoint and see what it was like, almost like being in the middle of the fight. This is 2015, 2016. So, compared to today, it was pretty primitive, considering how fast these things move. But with the changes you’re talking about, like if it gets a better and better experience, then I think those things could be super interesting. We got a good reception at the time, but 10 years ago, even compared to today, it was a little bit more novelty and not like, wow, I’m really in the experience. But it’ll get better.
[00:41:19] Jonathan DeYoung: And I guess it’d be Fortnite now or Minecraft or something. It’d be the one you’d be doing instead.
[00:41:24] Wes Levitt: Yeah, exactly. You got to change with the times.
[00:41:27] Jonathan DeYoung: Just to close things out. Your website mentions that the Theta team, I’m quoting here, is made up of experienced technologists and executives from Salesforce, Netflix, Amazon, Samsung, Vimeo and Google. And I’m always curious why people make the jump from Web2 to Web3. Everyone seems to have some sort of motivation or story behind it. So, why do you think so many people moved on to Theta? And what brought you to Theta?
[00:41:53] Wes Levitt: Yeah, as the odd man out that came from finance rather than tech, I can’t speak to them, although I think it’s probably pretty similar motivation. When I started looking into getting into blockchain in general, but then looking into actually changing roles into something, this felt like I was working on much more interesting problems. It moves fast. You’re doing things in a lot of cases that no one has done before. That’s not to say that doesn’t happen at Google or Amazon or Netflix either, but in some ways, crypto feels like that on steroids. I mean, you really do get to be on the ground floor of a very dynamic sector and move very quickly. And I found it very interesting and challenging in a way that prior roles hadn’t. So, for me, it was a very easy call, and I never looked back.
[00:42:33] Jonathan DeYoung: There you go. Well, thank you so much for joining us. I know you’ve got to get back to your children, and I don’t want to hold you up from that. I’m sure Ray understands as well.
[00:42:41] Ray Salmond: Yeah.
[00:42:42] Jonathan DeYoung: As he has a kid himself.
[00:42:43] Wes Levitt: Yeah. Well, thanks. It was great talking to you, Jonathan and Ray.
[00:42:46] Jonathan DeYoung: Of course. I learned a lot. I obviously wanted to do some research into the project. I don’t want to go in blind. But it’s sometimes fun if it’s something you’re not too familiar with to go in as kind of the average consumer, because that’s the person who’s going to be listening to the podcast. Somebody who’s also not very familiar with it. So, you definitely enlighten me on a ton of stuff, and I definitely appreciate that.
[00:43:05] Wes Levitt: Great. Glad to hear. I enjoyed it as well.
[00:43:14] Ray Salmond: The Agenda is hosted and produced by me, Ray Salmond.
[00:43:18] Jonathan DeYoung: And by me, Jonathan DeYoung. You can listen and subscribe to The Agenda at Cointelegraph.com/podcasts or on Spotify, Apple Podcasts, and wherever else podcasts are found.
[00:43:30] Ray Salmond: If you enjoyed what you heard, rate us and leave a review. You can find me on Twitter at @horushughes. H-O-R-U-S-H-U-G-H-E-S.
[00:43:39] Jonathan DeYoung: And I’m on Twitter, Instagram, and just about everywhere else at @maddopemadic. That’s M-A-D-D-O-P-E-M-A-D-I-C.
[00:43:50] Ray Salmond: Be sure to follow Cointelegraph on Twitter and Instagram at @Cointelegraph.
