Decentralized Compute Revolution: Tom Trowbridge on Fluence Labs and DePIN Token Economics

Decentralized compute revolution, Tom Trobridge on Fluence Labs and Deepen Token Economics. In this episode, Gemma sits down with Tom Trobridge, co-founder of Fluence Labs and former co-founder of Hedera Hashgraph, who shares how Fluence is building a decentralized compute platform, offering enterprise-grade services up to 70% cheaper than AWS and traditional cloud providers. Tom reveals why Deepen Token Economics must be grounded in real fundamentals to attract institutional capital and how focusing on specific customer segments like node providers is the key to scaling decentralized infrastructure. And the lessons learned from hosting 13 deep-in day events across nine cities. I'm your co-host, Anthony Pearl, and whether you're an investor or a startup looking for insights, it's time to get unblocked. Tom, it's so great to have you on unblock today. Welcome. Listen, it's great to be here. Nice to see you, Gemma. Likewise, Tom, for those that may not know you yet, could you start by telling us a bit about yourself and what you're building right now with Fluence Labs? Well, sure. Those are a little bit different questions, I guess, but I will start with Fluence is a decentralized compute platform. And so what that means is you think about what drives the internet, what drives applications and what drives kind of the whole virtual and internet economy to which we are all accustomed independent and it's computing power, right? And people think of computing power with regard to AI training and that's a huge part of it. That's GPUs is in the news all the time. But even before that became a real story in the last kind of 18 months, compute powered everything from your Uber application, figuring out how to price a car and how long it was going to take to get you in which card to get or your Amazon search or your Netflix search or any of these things, you know, your airline, these are all consumer oriented stuff, but also compute is what you use to sequence a genome to do drug discovery, like everything relies on compute power. And so what Fluence has been building is a decentralized platform that uses an open network to assemble enterprise grade compute, which is in top tier data centers around the world and is effectively stranded or unused or for a variety of reasons is very cheap and offers that for customers, businesses to use for whatever computing they need. And so we are effectively creating a global network of compute, which is up to 70% cheaper than the centralized clouds. And the network is live is launched. We have customers and we are scaling it. And before help joining Fluence as a co founder, I helped found today, or hash graph, which is a layer one competitor to Ethereum and Salana that that project. I think I joined that in 2017, basically helped found it, launched the public ledger that token launched in 2019 into top 20 now has been for a while. And before that, I had kind of a variety of roles in kind of intersection of technology and finance with time at some venture capital firms and Goldman Sachs, the well-known American bank and basically use a lot of those relationships and a lot of those kind of the skills I learned and was able to transfer them into the crypto world full time in 2017. And then in terms of just to go back to your kind of opening comments about decentralized compute and what Fluence is doing, how big is that space? Like how much is happening around that? And what is it about the way that you're doing decentralized compute that is particularly compelting? I guess, first of all, compute is huge. If you think about Amazon or the way I like to kind of frame it, is there are you take Amazon, Microsoft, Google and then a couple of the large Asian kind of compute, hyper scale or equivalence like Alibaba, Tencent, et cetera, there are about 300 to 350 data centers that have each on average 55 or 60,000 CPUs. All right. So that's a huge amount of compute and that's even before talking about any of the AI focused data centers, which have been in the news over, you know, the past kind of year or so with, you know, Elon and XII building one and open AI building another right. So just those are a vast amount of compute and that even existing CPU growth is scaling and obviously GPU growth and demand seems to be growing dramatically as well. That market is in the kind of hundreds of billions in aggregate and the decentralized compute market is also large. It's a little hard to nail down exactly how big it is because some groups are not super transparent. But I would say that I think in aggregate, we're talking about a couple hundred million of reported annual revenue in decentralized compute. And that's heavily skewed towards the GPU market now, which is done for AI training. And there's a couple of big projects like Aether, the reports 70 or 100 million and there's IO dot net which reports relatively significant numbers well. And so I think those are the large ones. I think what has changed and people have been trying to attack the decentralized compute problem and opportunity for quite a while historically projects. I think we're a little overly ambitious and attempted to aggregate retail machines and retail CPUs in order to put them into a common network and solve kind of significant enterprise grade computing problems and that is a very laudable goal. But it has both technical hurdles and also adoption hurdles. And so technically it's obviously more difficult to create a consistent kind of enterprise grade offering and then furthermore, you have a marketing problem of trying to convince customers who are working at top tier kind of data centers that they should go to a network that's running off of home PCs and that's also just a difficult hurdle for people to get over. And so ours influence starts from day one with the top tier data centers. And so we both have a less of a technical challenge in terms of promising and delivering reliable service and also a much easier story and pitch to tell when customers can see and know the level and quality of the data centers at which their compute is taking place. And so that is, I guess, a big evolution. And that is part of the differentiator that we have versus kind of earlier gen one projects that came before us and I'm actually had on the podcast true decentralized compute projects one, which I'm sure you're aware of Filecoin and other one, which was old mobile phones that are being read purpose to provide decentralized compute called Acres. If you heard of that project before. Yeah, so I know both those very well and protocol labs is investor influence, but they're really decentralized storage, not compute. And they have a compute they talk about compute, but they've been founded based on storage. Correct. Yeah. And they have kind of a sort of a layer that's kind of compute, but it's really designed to enable customers to use their storage more. And so I really put them in squarely in the storage bucket primarily. Got it. I just is doing some interesting and I've had them on, you know, my podcast they spoke on a deep and day events. And what's interesting about them is they're really focused on security and the ability to use is you're probably very well aware the security that is inherent in the iPhone, which is superior to what you have in kind of retail or other machines. And so they can promise and deliver a higher security compute, which is a terrific solution. It's for a number of very specific applications. And so I think that is certainly interesting. It's quite different than in terms of I think scale of compute in terms of what they can offer, but the security level is clearly a very attractive attribute to have. Got it. Thanks for just kind of distinguishing those segments in a more fine green way. I mean, you mentioned IO.net and a, they either main other projects that you would put in the same categories. Yes, both of those projects are in the decentralized compute space for sure. So in one sense, yes, they also are exclusively focused on GPUs. So we started off on the CPU side. We've since added support for GPUs. We know them well. They've spoken a deep and days. I've had both of them on the podcast as well. So like them both, I think big picture, yes, those are, I think the two kind of leading from a revenue perspective, decentralized GPU compute providers for sure, right? And apart from fluent slides, could you tell us what else you're working on? Obviously, you've got a podcast. Tell us about that and other things. Well, yeah, we've consolidated a couple things into something called deep in space. So deepenspace.co and what we've done, and this was a desire, I think, basically what happened was we launched Fluence at ETH Denver in February of 2024. And as part of that launch event, which is kind of an IRL launch event, which had kind of stopped during the COVID period by the time we got to 2024 people were back doing those. And we invited a number of other projects and investors to speak as well. And so we had an IRL day for the Fluence launch and we realized that there was a lot of interest in deep in and we had a lot of relationships in the space. Even though our launch is done, we can keep doing deep in focused events. And so we then created more of them. I think we've done as of now 13 deep in day events around the world. And I'd have to look but nine different cities or 10 different cities and then realize also what got these events. It's great to dig into more detail with a larger number of projects, some of whom may not even be able to make it to these events. So I host a podcast called Depend as well. And those are just all designed to raise awareness of the deep in space and of projects, investors and providers and thought leaders in the deep in space. And so it basically is something that Fluence subsidizes. We certainly there's only question of how much money we lose, not if we lose money. But it's just a way for us to kind of bring the ecosystem together. And what I find in deep in is that L ones, as an example, all know of the other competing L ones pretty much, right? But if you're in deep in the commonality is really the business model, having a token component to it, but you can have otherwise businesses that would have no recognition of each other, right? And so you could be accuracy, for example, you could exist for 20 years and not know of wing bits, which is tracking airplanes because why would you, right? And so I find that the sector benefits from a space for whether it's a virtual space or a physical space for people to come together because they're not coming into contact with other projects as regularly as you would in other sectors where you have kind of more natural overlapping kind of relationships. And there still is a commonality there of how do you scale networks? How do you buy and reward? How do you buy hardware design hardware for some and then how do you attract and send customers? There's still some key commonalities that make it worthwhile. But it has been, I think, really interesting and has really helped us as a project to get to know the broader ecosystem than if we hadn't done it. Yeah. So what I'm hearing from what you're saying there is it's partly just like just for generally understanding because you're interested to know what's going on. But there's also the potential for information sharing that's actually useful for your project and others through that process that you recognize as well. Yeah, I mean, that is definitely true. We have gotten some customers as a result of some deep in projects that are using Fluent's Compute, so that has happened. But I think it's also just been helpful overall to know the challenges that other projects face, the opportunities that they're facing as well and some of the successes. And I think that is also a real benefit to hosting these and I mean, I should have looked this up beforehand, but I know we've had over as probably close to 150 speakers at these different events. And I think I met like number 70 or 72 in terms of recorded deep in podcasts. So it's been a considerable effort. I think this earlier this year, just recently, we did a deep in day, a month for a couple like we were in Buenos Aires, it's been a big pace. I'm girls. It's not America. Buenos Aires was around DevCon. In terms of like token models, could you tell us about the token model for Fluent's and how it works if you've got one? Well, sure. Listen, I think that's a really important thing and that is I think a key thing that is important and deep in it. And also I'll just mention that I wrote a paper on deep in token economics, which you can find a deep in space that's CEO where I review probably 30 different projects in terms of their token economics and kind of tried to still down the key factors in the different types of deep in projects and how to attract and use it. But I think the basic point I have is that deep in is a space is will generate real revenue. And with that, we'll come real investors and with real investors comes real scrutiny. And it's not just meme coins in my, the comment I love to say is deep is not a meme. And so with regard to fluids, what we recognize is that in order for token to be successful over the long term, that token should bear relation direct economic relationship to the scale of the network. And so those have to be linked together. And so with regard to fluids, you know, with many, many projects, you know, you can have revenue that revenue can buy and burn and token and that I think is works for a number of deep in models. And for us, because we're a network, we connect providers and suppliers. So fluids itself doesn't generate the lion's share of revenue. But what does happen is we have to secure the compute on the platform. And so for every CPU that joins the fluids network, fluids tokens must be staked in order for that CPU to be active and provide service. And then people add CPUs because then they share in the revenue of those CPUs, right? And so as fluids scales, more and more fluids tokens need to be staked. And importantly, their fluids tokens, but the amount staked is dollar based. So if the token goes down and you add a new CPU, if that new CPU will acquire more tokens to be staked in the previous CPU, because the dollar amounts consistent, not the token amount. And now token goes up, right? The next CPU, the joins requires fewer tokens, but obviously it's higher at that point. So it is, I think, the dollar kind of denomination of that, I think, is very important. And I also think the relationship of that demand to the scale of the network and the thing I like to say is that if, or I should say, when we get the scale back to my comment about the scale of the compute ecosystem globally, 155,000 CPU data center, right? I mentioned that hyperscalers have about 350 of those. One of those would require about $300 million in fluency stake, which is many, many, many, you know, multiples of our entire market cap right now, right? So that gives you a sense as to, as we scale, what the demand possibility is for the fluency token. We're also looking at adding some other attributes to it as well, which will, I think, be related to compute units and helping set a value for compute credits in a way that you effectively create a real world asset for compute on fluids that can be traded. So they'll be a token that will then allow you to buy compute on fluids that token price will change based on that demand. So it's something else that we're, we're looking and we've, we've written about this. Fluence is an RWA real world assets relationship with regard to compute and we're making steps to actually launch that in the kind of coming quarter or so. Thanks for blushing that out for everyone's understanding. Your background is like in traditional finance. I think you worked in investment banking, private equity and hedge fund roles before getting into web three. Are there any assumptions from finance that you now challenge or that give you a unique lens on a being in crypto? Well, the main thing I am aware of coming from that space is I think regards the token economics piece, which is this memes, I've always sort of been skeptical of memes and that fundamentals ultimately do matter. And that's why I'm attracted to deep in overall. And that's also why I wrote this token economics report because I'm convinced that as the deep in space evolves, real investors will come and real investors will bring with them the traditional lens of evaluation, which crypto projects historically have not really had to deal with their bother with, but with retail, particularly now being so exhausted and hurt by the incredible volatility in the space, institutional investors are kind of the only hope the whole sector has for significant kind of value creation appreciation. And those investors are going to demand very different things than what retail has demanded. And so I think I bring that outside kind of traditional finance perspective to this, which was completely irrelevant and unnecessary for the first decade of crypto where people just cared about memes and kind of all kinds of unusual metrics or non-traditional metrics I should say that I think over time are going to become much less relevant as crypto becomes more mainstream. And with it becoming mainstream, huge amount of capital is potentially allocated to it, but that capital is going to be looking at things in the way that capital looks at things, which is not how the kind of crypto ecosystem has historically. Yeah, I mean, I was at Solana Breakpoint last week in Abu Dhabi and the opening presentation was about how crypto has produced text fastest growing companies to a hundred million revenue. They're most of the in the last 12 months, the companies in that category have come from the crypto space. And so I do think you're seeing an inflection point in what you're saying Tom, but I would just say that I don't think it's mutually exclusive like meme coins all this because I mean, it's very easy to dismiss them, but they're really about the attention economy and marketing. And I think that's a really key part of commerce. I heard also not at Breakpoint, but previously literally, the president of Solana talked about that meme coins are like the test net for the new operating system for financial markets in Web 3. I mean, that just kind of makes them sound more like a tenorant rather than anything of substance. But yeah, I don't think they're as trivial as the real stuff's happening over here. I think they're both significant fair enough, but I guess I'll make two comments. So those first that come and you said about the fastest to a hundred million revenue, how many of those were exchanges? I would think a lot were I'd love to see that list because I got to think a bunch of those were exchanges. Yeah. So fan and like are they ramp radium? I mean, there are Dixas in there, obviously, backpacks sky, also no flake, Jupiter. I'll send it to you. Yeah, but you know, swap, right? So the bunch of is pumped up. I mean, those are great, but a lot of those are companies, you know, swap a little bit different, but because I would think that a lot of the crypto success has been, I mean, look at tether, right? These are centralized tethers, like the biggest success in the space. And so I would love to see that, but I think a lot are going to be more traditional businesses in the crypto space. So it's a little different. Yeah. I mean, together's not on the list. I've just sent it to you on WhatsApp for what it's worth. So you can have a little glance if you wanted to. Yeah, let's see. And then the other point is that on memes, memes, let me be super clear, memes are not going away. And for sure they provide something in fascinating, interesting, and I don't think they go, but I also don't think that that's where pension funds, endowments and foundations, right? They're not going there. Now they may go into a pump.fund that enables that. Great. But people will always trade meme coins, no doubt in their meme stocks, right? So don't get me wrong. I don't think that goes away at all. But that's not where I think the institutional capital is going to go. That's all. I would agree with you on that, but having said that, you know, the institutional capital is going into the layer one, which is using that. So they are why are you doing that? But more in a secondary or even advert and white, let's say, yeah, they're monetizing it. It's a way to monetize that it's a use case, which is being monetized. It's like going into an exchange that trades where money is created, where the value is created by the people who are trading these meme coins. Same thing, which is, again, I don't think it goes away at all. But I would also be curious to see meme coin volume over time because I got to think it's down significantly, but I don't know. Well, there was a debate actually on stage about whether token burns good. And I think the case of pump.fun, the community wasn't happy about the company reinvesting to grow the market because they didn't believe there was a huge growth opportunity beyond the current status. So they were not, you know, because the debate was about should you do value distribution to token holders, should you burn tokens and how much of the company retain as well? So it's kind of like, yeah, and that's a fun philosophical debate because I get it. And if you compare it back to traditional world of stock buybacks, right? You can the company invest capital more effectively and grow, right? Or does it buy back the stock and effectively return money to shareholders? And so if you want to be in a company that has a better use for the capital, right? So that's the concept. However, the crypto concept, what I would say is that, I mean, for this way, you've another overlay in crypto, which is trust. And a lot of the projects I mentioned who claim a lot of revenue, there's no auditing. There's no, not even case studies, right? They're just, here's what I have, right? And so you have zero accountability. And I frankly think there's just a question of time before some of these projects that claim a bunch of revenue are found out to not have them or have it to be, you know, very overstated. And so by burn is even more verifiable than an audit. We've all seen audits that have a huge amount of number of companies in the traditional finance space, if the auditors have been completely fooled. So by burn done correctly, transparently, validates proves revenue in a way that even auditors can't or not as effective at doing. So that's 0.1. And then 0.2, if the company holds some treasury as many do, if you buy burn, you then have treasury, your treasury is more valuable. And then you can then spend some of that treasury on whatever you need to do in self some of that. Now you end up in some of a similar place, but you're at least fully aligned with the token holders at that point. And what I argue is that every divergence from 100% by burn is now diverging from your token holders. Now the name may not be much, may just be a little bit, but it is some divergence. As a token holder, I would like to be fully aligned with the management of the project and what they're doing. And so any divergence from 100% starts to creep that disalignment up and to the extent where they do zero by burning, keep all the revenue themselves, well, then why do I even hold the token? That doesn't make any sense whatsoever. Right. So you got to figure out where you are on that spectrum. But if I have a project, this generating revenue and there's zero by burn, and there's, I can't think of a reason to hold the token. Interesting. I mean, you can actually listen to the debate. It's on YouTube. You might find it interesting, Tom, because obviously you've thought about this quite a lot. And the analogy to, you know, traditional financial markets was made as well, and that with Tried Fire, they can just based apply for a permission to do a buyback, but then sit on it and not actually do it. Whereas in crypto, if that was stated that the company was operating like that, and then they didn't, it would be seen far more negatively than in Tried Fire. And it sounds like you would prefer a buy and burn, then like a staking model with value distribution to token holders. Is that right? I guess it depends a little bit on the project. I think it's a buy burn is just much, much simpler staking. It can be useful, but I guess unless there's a reason to stake, IE trust, then you add a layer of complexity, but you also add a layer of additional token demand, right? And you are then rewarding people who stake and lock up for longer periods. So I would say that if you have 100 million of revenue and you want to reward people that stake with that revenue, I'm not against that. It's a different mechanism, but it's what to be clear. That's a different mechanism of rewarding token holders that may take a little bit longer. They may not reward all token holders, right? But it also rewards activity that behavior that you want to reward. So I'm not against staking rewards at all, presuming they're done openly, transparently, but again, we're back to the same thing of that rewards token holders. And again, if it's 100%, the capital's return revenues return to staking rewards. That's a hard to argue with. It's more to me the question of how much is retained to run the business? And obviously you would think some needs to be retained to run the business. That comes back to how big is treasury, how align are you with the treasury or not? And that is, I think more of the question, whether it's a staking rewards or a buy burn. How do your clients buy that maybe aren't in Web 3 fine, having to buy and stake tokens in order to access the service? Well, they don't because they don't have to know about it at all. And so the important thing with fluency is we've separated the staking from the people who provide server. So for example, these are all different ecosystems. So if you are a data center and you want to contribute servers, investors can stake those tokens. So you don't have to do it. You just promise them a share of revenue. So they don't have to buy because that just effectively it just raises the cost for providers. And that's not their business to do that. So we have a model where individual investors can do that institutions can or their pools that do that as well. So you as an individual who don't have enough to stake to a particular machine can contribute money to a pool that pool then stake. So the provider is enforced to do that. And then if you're a compute user, you also can pay in Fiat. So you don't have to pay in tokens as well. So the idea is we sort of aren't abstract out as much as possible. The complexity of crypto. Doctor, I mean, you mentioned, I think early that that fluency has lower costs. I think you said earlier, something like 70% compared with traditional services. What are the trade-offs if yet there any around like things like performance or stability or latency or availability for AI or compute teams to take this lower cost route? I guess the main point on the pricing is that it's not like there's one price for compute. And so when I say that price, that is off of a list price at a data center. But then if you are going to have a longer-term commitment at a certain scale, then your pricing starts to go down. And so are pricing the discount at which it is available or rather the discount to which you can compare really differs based on the scale and duration of the compute you want to purchase. And so if you're talking about larger numbers for longer periods of time, then the fluence price is at less of a discount than the hyper-scalers. Now, in terms of trade-offs, the main point is that we're offering similar, if not the same machines, same kind of reliability and top tier data centers. You don't have the brand name, right? So that is one thing you don't have. We also are working to implement SLAs. I hope that comes soon. But the SLA is a key piece that we need to have that we don't, which is something that is, I think, will unlock a lot of use for us when we complete it, but we don't yet. But those, it's really brand name and kind of comfort in using it is really the two main things. But let's also be clear that we don't offer a full range of services at all. And compute sounds like it's one thing. Compute is a very complex heterogeneous set of services. And so our first customer base are a third party node providers. And so these are companies that provide and run nodes that support a wide number of layer one and other blockchains. And they run them for individuals and for companies. And so for those businesses, about half of their cost is compute. And so we can then save them. Again, these are people that buy its scale and longer duration, right? So we're not saving them 70 percent, but we can save them 20 or 30. And that's off of half of their cost base. And so that is compelling. And I think that is by itself a billion dollar market. And so that just gives you a little bit more detail in terms of what we're looking at and doing. And because it's selling a very basic compute service, the differentiation in terms of what's offered versus centralized clouds is very, very small because of the basic purpose and basic use of compute for that use case. I'm intrigued about the decentralized versus the centralized option. And is your target market the same market as people are you trying to convince some of the people that traditionally have now been using those more centralized services, AWS and the like, to start considering this as an option. Oh 100% and that's in the R. And I guess the main reason people do it, I think, is not that it's decentralized. That's sort of a nice to have at best. It is more that is the mechanism that allows pricing to stay cheap. And I'll explain it is that it's an open network, which means that if you're running compute on fluids and a provider wants to raise price, you can easily find the next provider that hasn't raised price or is cheaper. So your switching costs are effectively zero, which means price is always competed down to like the lowest level. There's no locking. Like if you want to leave Amazon as a headache and you know, it's a real arduous journey. And so the idea and fluids is that it's an open network. Anyone can contribute compute. And so that basically allows price discovery across the world of lowest price. And that works because it's decentralized and because it's an open network. So you're on it for that feature more than you're on it because it is decentralized. Now it being decentralized also means you're maybe more resilient in terms of failure. You've got heterogeneous hardware and you know, you're not on one network and you're not in one geography, right? So there's other attributes that come with decentralization, but that hasn't been the focus. And if you're now one, it's kind of crazy for all these L1, you know, decentralized projects to be running on Amazon or Google, which many are, right? But that's just where the nodes are running because of the cheapest and easiest. So the goal here at Florence is to help decentralize this whole ecosystem, but doing it not for the sake of doing it, doing it because the decentralized mechanism actually provides a value and a durable value to that ecosystem. In terms of security, though, I mean, that would be I guess a primary concern for people. Is there a security issue? Is there a security advantage perhaps in between one and the other? Well, there's two components. There's reliability and there's security. And so those are quite different. And so with regard to reliability, you can argue this is more reliable because you can shift providers. You're in different geographies, right? So you can be more reliable. Obviously, despite Amazon going down a bunch of times, it obviously is very reliable. It has gone down. It will go down again. We know that, but it's just so there are vulnerabilities, but and that also comes by the way of having a homogeneous software stack that has a lot of single points of failure where if Cloudflare goes down or what goes down, you have these ripple effects across the network, right? So you've seen that architecture that's built a specific way, it obviously has vulnerabilities in it. In a decentralized network, you may have other vulnerabilities, but you may not have that vulnerability. So I think it is as secure or that as resilience, if not more, but it's a conversation we can certainly have. I argue open source in general is more resilient, but that's a whole path. And then in terms of security and privacy, which is really the key piece here, that is a very complicated conversation because the privacy in data centers if you're using Amazon is effectively done via contract, right? And so it's trust. And so if you're doing it decentralized, you now no longer have that trust. You want to make a trust list. That gets difficult. And so the only way really execute that is trust execution. Varmus kind of chip that was very sort of chip specific. And that is a challenge. And so the privacy component of it is certainly an issue. And it's something that I don't think that we have solved the scale, but it's also why the use cases that we're working on right now are not privacy focused. Yeah. I mean, now there are lessons to be learned as well from what the move into Cloud was a big thing for a lot of people, right? There was a lot of resistance for a long period of time. There's probably still people out there that don't like the idea of things being on the cloud. Are there lessons to learn from how to market it and how to gain that trust in that market space that you can take forward for what you're doing? It's funny to mention that because I am in the long career that Gemma mentioned earlier in my venture capital days in 1998, we were investing in data centers that were effectively trying to, you know, we didn't have the term then create the cloud, right? That was in 1998. And we said, this is inevitable. This is happening. And you know, I don't think Amazon launched its service until 2000 or something along those lines, right? And there may be longer. It took a long time to scale. And so, you know, I'm quite familiar with the adoption challenges in getting companies to move to the cloud. It's obviously been enormously successful. But it's taken a while. I think every evolution in technology happens faster than the previous. So this will not. And we view decentralized compute as like the next and maybe final evolution. But it will co-exist with the cloud. But the cloud to be clear has spent decades and tens of billions of dollars on not only infrastructure, but products and services. And so that is a very difficult thing to compete with. And so what I think our lesson is is to focus on one segment at a time and really work to find a product and offering that resonates with one particular segment and then grow from that segment to one other segment. Because it is just impossible to try and compete with the cloud across all the products and services they offer. And I think in the beginning, if I think back to my data center investment days, you know, there was a wide number of different types of businesses that these, I can think of two we invested that we're trying to bring into the data centers. And we probably should have focused on customer segments more than geography at the time. And that would have, I think, helped our scale ability back then. So that's one way we look at it and one thing we've taken away. I mean, you just touched on the kind of challenges with adoption and uses. Many of the projects obviously in deep in struggle with the flywheel, you know, in terms of scaling users and nodes and value all simultaneously. What are the design principles that you think are really critical to overcoming this challenge and that you're implementing? Well, I mean, this and I think there's a question and answer for fluency. There's a question answer for other compute and question answer for kind of the sector overall and more what I call physical deep ends, which are not compute or storage or kind of bandwidth, I suppose, related. And I guess with fluency, particularly, I think that is just customer segment focus. So scaling within one customer base and using that, getting a customer and having that customer be a reference for the next one. And they know each other rights and ecosystem and building trust within that segment is a way to start that flywheel for us. And I think that's also true for other projects in the kind of what I call digital deep end space compute storage bandwidth, etc. And I think it's not that different than the physical side where, you know, if you're mapping or you're selling location services or whatever, is still understanding the different customer segments, how target I think is the only way really to go about it. So really got to have like a deep understanding of your core customer, understand what segment you're going to focus on and that there's like enough addressable market there to make it worthwhile and really deeply understand your customer and the product tailored for that 100% got it. I mean, it is akin to traditional markets. And I think also what you've said earlier, you know, you're putting blockchain is front and center, but for the customer experience, it might be more of a background thing to the extent that they're more web three native or inclined. Yeah, for sure. It's all background. Yeah, without a doubt. I mean, and that's how this industry goes. I'm pretty sure ultimately not just deep in but overall where we need to survive and thrive based on the products and services we offer and how we get to that may be blockchain related, maybe decentralized, but that's not no one is going to buy a product or service because of the infrastructure they're going to buy it because of the attributes. Now those attributes may be unique to the infrastructure is decentralized, but really it's about the value that project and service can offer, right? And the blockchain is just for some projects, it is the most effective infrastructure to deliver the product and service that they think is valuable. And so we're waiting to see and we'll see and by some projects are showing us the case, others are not, but that in some sectors will make sense others who won't. Yeah, absolutely. I mean, I'm aware of like blockchain project that might be merging with like a tried five project, and then the blockchain rails make the margins and EBITDA are much more compelling compared to the industry norms and you can see with like use cases like that that it's, you know, as you say, that the attributes, maybe I'll just turn to like what's on your mind and how are you thinking about 2026 in terms of the work that you're doing and your focus? You know, 2026 should be a exciting year from a revenue side. So we've got a pipeline that keeps growing and so we just need to kind of just pull these customers in and get it on and start to scale up. That's the real focus and the only focus really, really for fluency is doing that. And so that is a big thing that we're all kind of pulling together to make happen. And I'm excited to see that by the way, across the whole deep in space is revenue coming in across the deep in space overall. And I thought 25 would be a big year for that. I think we've obviously seen some terrific strides there, but I think 26 will be much more significant. What about slightly different topic, music? What's your favorite song at the moment, Tom? Ah, the favorite song is, and it's actually, my kids are annoyed because I was just playing it this morning here was lose your soul. I don't know if you know, lose your soul six months older. So it's by interplanetary affair. Oh, okay. I'm going to look it up. Thank you so much for that. Your kids are not fond of the song and getting. Well, it's because it says you're going to get up. I wake up to the beat of the drum. So in the morning, it's kind of a fun thing to try to get them going, you know, kind of like the Dolly Parton song working nine to five. Yes, but the other one I play is five minutes before school. There is a country song called you got five minutes. And so they're very familiar with the five minutes song as well. Oh, I'm going to use that for my kids as well. They're quite often telling me that I'm very cringe. And I'm sure this will help in that. Yes, exactly. Tom, it's really been a pleasure to have you on Unblock. Thank you so much for joining us for the conversation and helping us to understand more about deep in and also decentralized compute space and what you're working on. And it's not just about, you know, the work you're doing, you're really trying to support the ecosystem to understand itself, understand each other and through that kind of cross-pollination for it to kind of reach and fulfill its potential. So I just wanted to acknowledge you on that and really appreciate your time today. Well, appreciate it. It's a bunch of work on our side, but it's been fulfilling. And so we want to keep doing it. And we've got to get you on in person on one of our deep end days. So hopefully we can do that in 2026 as well. Thank you for the invite. And I accept I would be delighted. Thank you. That's all for this episode of Unblocked. Please check out the show notes for information on power ledger and other contact information. We welcome your comments and feedback and please hit subscribe wherever you are listening. This podcast was produced by podcast done for you. We look forward to your company next time on Unblocked.