3190: The Role of Decentralized GPUs in AI Innovation

[00:00:04] What does it take to democratize access to AI and level the playing field in a world dominated by big tech companies? And as we look at the demand for AI compute power surging, how can businesses, researchers and startups compete when GPU resources are costly and concentrated in the hands of a few? Could decentralization be the key to unlocking the next wave of innovation?

[00:00:32] Well, my guest today is from a company called NetMind AI and they are a London based AI solutions and infrastructure company that's dedicated to advancing artificial general intelligence or AGI. And he brings over 15 years of experience as a two time founder, investor, deep tech expert that also has a passion for democratizing AI access.

[00:00:58] So in my conversation today, we're going to explore how NetMind Power, how their GPU marketplace is disrupting big tech's dominance by tapping into idle GPUs from individuals and data centers. And dive into how agentic AI and its real world applications across industries are already reshaping business operations by enabling smarter decision making processes.

[00:01:24] So could decentralization be the answer to ethical concerns like AI misinformation? And how can smaller players use these technologies to compete with big tech? And what does the future hold for AI agents and decentralized innovation? So thank you for joining me on the show today. Can you just tell everyone listening a little about who you are and what you do at NetMind? Hi, Neil. Well, first of all, thanks for having me.

[00:01:53] So I'm Sina Rajal, the chief commercial officer at NetMind AI. And as chief commercial officer, my job really is to drive strategy, partnerships and growth of the company in a way that will ultimately lead to commercial success for us. And I know you've listened to the podcast before we've been talking, before we started recording today.

[00:02:14] And one of the things I always try and do on these podcasts is learn more about the origin story and the story behind the guests and their success that they're experiencing right now. So can you tell me a bit more about your career trajectory? Because having done a little research, I was reading that you founded two tech startups to becoming an investor, now leading NetMind strategy as CCO. I suspect there's a big story here. So can you tell me more about that?

[00:02:42] I guess this is a story about just looking for something all the time and being driven by something powerful inside that for all of us. I mean, that's, everyone has their own story. For me, I guess it's always been about, I've been fascinated about how technology can bring about groundbreaking change in the world and how we can to sort of use that saying, move the needle in this world.

[00:03:06] Be it for things to do with the greater good, be that sustainability, be that democratization. So I'm always after that purpose beyond the profit in whatever I'm doing. And I think it goes back to my time at Cambridge when I was doing my undergrad, my PhD there.

[00:03:25] And the work I was focused on, particularly my PhD was what role manufacturing plays in the economy, especially in the UK economy where we think we've gone beyond production and beyond manufacturing. I wanted to see to what extent that is true, especially in an increasingly digitized and modern economy. And so that was where my fascination with, I guess, technology and tech startups and all that began.

[00:03:51] And I was one of the early ones into a student led competition for entrepreneurship, very much, I guess, inspired by a big entrepreneurship competition at MIT in the US. And we brought that model to the UK. And at the time when Cambridge was very much still in its ivory towers days and the words commercial and the words business and entrepreneurship were all no-nos. You just didn't bring those up.

[00:04:20] It wasn't polite conversation. And we went ahead and did this anyway. And it was such a great grassroots movement. We very quickly became the largest student body, student organization in the university. And I think that then sparked the university itself to take action. Now, today, of course, that's led to many branches of activity in the university itself, of course. And Cambridge is very much the sort of Silicon Valley of the UK. But I'm happy. I'm excited. And I was there in the early days of that.

[00:04:48] And that really was the spark for me. And I couldn't get that out of my system. So when I graduated, my first act was to get on a plane and go to Silicon Valley proper to see what I can do out there. And having helped so many companies spin out of the university while being in that sort of initiative and that competition at university, I felt I'd very much like to do this myself now, having helped so many do it. But it was slightly disappointing time because it was very much the beginning of the social media frenzy. And so everyone wanted to create the next Facebook.

[00:05:17] And just having come out of a PhD in engineering, that to me just felt like a, this is very unfair, of course. And lots of people are going to sneer at the podcast hearing this now. But, you know, it just felt unsubstantive and not really what I wanted to dedicate my life to the next big social network. So, disappointedly, a while later, I got back on the plane to come back to London. And I picked up a magazine in the airport. I think it was Fast Company.

[00:05:45] And in it, they'd listed a number of companies that we should be looking out for the next 10 big companies to look out for. And one of them was interestingly called MFG.com, short for manufacturing.com. And its backstory was that this guy called Mitch Free in the US, a manufacturer himself with a, so not your typical tech background, had decided that it was just, he'd had enough of manufacturing going abroad. And so he created an online marketplace for manufacturers in the US. And it had just taken off massively.

[00:06:16] And I believe that the guys behind Jeff Bezos' Blue Origin spaceship program were really sort of keen users of this website. And so he'd had a very early injection of money from Jeff Bezos, which was a very big figure in those days. So anyway, I wrote to him and said, I would love to do something with you. And obviously, you don't know me from Adam, but just come hot off a manufacturing PhD and love tech. I'm sure we can do something.

[00:06:43] And he wrote back and said, fine, great, let's do it. So I joined him and his team at a very critical time in the company's development and worked on expanding and growing this rapidly moving company. But what happened there was I learned a lot about early stage startups and rapid growth, but also that technology was still trying to keep up with some of these interesting developments in industry.

[00:07:12] So these guys were literally, when people were requesting components to be made in this marketplace, they were faxing them in, which A, dates me, but also shows you how ironic the situation was. It was meant to be a web-based platform, but people were faxing in requests. And then these guys were faxing those requests back out to various suppliers around the world. And it obviously didn't make any sense.

[00:07:35] And a little later on, we'll come back to how that helped me launch one of my own startups, because I left with the understanding that the gap here is some technological solution that allows people to match the shapes of the components they want to make with the catalogs of companies around the world that have those shapes and geometries in them. And that was the basis for a company I created, which Forbes called the 3D Google, which was essentially a 3D search engine for engineering components.

[00:08:03] And that was my own sort of tech startup start, if you like. But coming back before that point, I actually went on and joined the Clinton Foundation, President Clinton's foundation for various sort of large-scale global initiatives. And particularly my division was the Clinton Climate Initiative, and I joined to head up technology in that group.

[00:08:25] So my job at the Clinton Climate Initiative, we were looking at near-term climate change mitigating projects that can be deployed at large scale very quickly. And so we were an interesting mix of people from politics, finance and technology who'd come together to expedite the deployment of these sustainable technologies.

[00:08:50] And we had a particular focus on a number of areas, but two that were particularly interesting were one was solar and the other one was carbon capture, both of which needed a big heave to get them going at the time. And my job in particular was to sort of scour the world for the leading technologies in this space and help bring them to market. And so do the analysis behind them and see which are real and which are sort of smoke and mirrors.

[00:09:16] And that was a fascinating time, and I got to meet a lot of really interesting companies and people around the world. And everyone flung their doors wide open because we were seen as an interesting mix of being both a sort of an NGO, but at the same time a very commercial NGO looking to help accelerate venture development and venture rollout. And yeah, so that was a fascinating period from which I then went on to join two really interesting startups.

[00:09:43] One was a company in Cambridge here in the UK called 819 that had a world-changing solar technology, flexible, organic-based sort of solar panels. And the business model around that didn't work at the time because there were much cheaper alternatives in the market, silicon-based alternatives, solid alternatives.

[00:10:05] And we then combined that with a very interesting business model in traditionally unexplored markets in the emerging world. So in Africa, for example, where there's a massive need for clean energy, but of course, the economics traditionally didn't make sense. We looked at combining, therefore, this new technology with pay-as-you-go units. And so we're able to unlock this enormous market.

[00:10:33] And so this concept of clean energy for all became our sort of clarion call. And that was a mission that was absolutely fascinating. So that was a great time. And then I joined a company called Global Thermostat. And in both, I was heading up strategy and business development. And for them, they were still very early stage. And so we had to pull up our sleeves and be involved in every part of it. But Global Thermostat was capturing carbon from the air.

[00:11:01] And this was a team led out of the US, particularly out of New York, with some of the folks behind the Kyoto Protocol and a number of Nobel laureate scientists. And it was, again, a very ambitious and sort of moonshot type of project, which, again, I found absolutely fascinating. And after those, I came back to start up that 3D search engine that I mentioned, which I moved forward for a number of years. We raised finance and funding and VC backing for.

[00:11:28] And then another company beyond that, which was focused on AI. So it was the beginning of the AI revolution, I guess. And we were early in that with a company that was developing visual learning, visual intelligence, essentially. So AI that was fully explainable. That was what we call deep visual reasoning. So it could reason about what it's seeing and explain what it's seeing.

[00:11:51] And it had a number of very high impact applications in dealing with toxic content online, supporting security and policing in large crowds. And that led us to a big win, an international AI prize in the US for AI developments, which was a big success for us in that year.

[00:12:13] Having completed work in both of these, and then with the emergence of COVID and time to reflect on life and maybe make a few changes, I found myself on the investor side. And helping companies to both get ready for funding and to raise finance for them and into them.

[00:12:32] And so an interesting, sort of, if you like, an entrepreneur in residence type of activity where I would get my hands dirty, get into the companies and help them to get themselves aligned for the investment that we were about to bring into them. So supporting the investors, but also really being in the company, deep in the company and having both angles. And then I joined a company in Oxford in the UK called Streetbrown, which was developing autonomous vehicles.

[00:13:02] And I direct them towards industrial autonomy and the development of, I guess, trucks that drive themselves. And that was, again, absolutely fascinating. Again, a huge AI aspect to that. And if you like, embodied AI. So a fascinating, I guess, the future of AI really is that, embodied AI. And here we are today at NetMind. So it wasn't a million miles away from that.

[00:13:24] So moving from one AI company to another here at AI, NetMind AI, it's about democratization of AI for me. And making sure that we fight the good fights here as AI emerges. Listening to your incredible story there, it seems that you've been on, well, been in an incredibly unique situation by having sat on both sides of the table as both a founder and an investor. Which I don't hear about too often.

[00:13:52] So I've got to ask, I mean, what were the biggest lessons that you've learned about navigating the tech landscape when you've been on both sides of that table? Well, interestingly, I think a lot of it boils down to the information asymmetries between the two sides of the table. And once you've got the perspective of both sides, you realize how it really boils down to a number of key lessons. And if both parties know that and there is more transparency between them.

[00:14:20] And that transparency is not as if both sides don't want to give each other transparency. I think it's just through, as you say, it's experience-based. If you've not had one or the other of the experiences, you can't really put yourselves in the shoes of the other. And that's a shame because I think the real success stories are where both sides are fully hand in glove and totally understand what the other's needs are. And they are, in theory, they should be fully aligned. But I boil it down to four key areas. One is timing is everything.

[00:14:50] I mean, these are cliches, but they are so true in this space. Even the best ideas will fail if you're ahead of the market. It doesn't make a difference. So really, it's about getting your timing spot on. Another lesson is you really, it's about distribution. Distributing what you've got in the best possible way, going to market in the right way. Again, you might have a great product or a great model or a great offering.

[00:15:16] But if you don't have a good go-to-market strategy, then really, you can just leave it on the shelf to accumulate dust. It doesn't make any difference. The third thing I've learned is the team is everything. And ideas are a distant second. You can have a great idea with a second-rate team, and that great idea never comes to fruition. But you can have a mediocre idea with a fantastic team, and you know that something great will emerge out of this.

[00:15:40] Because a great team can pivot and move and flex, and they've got that kind of tenacity to bring it to a resolution. So team is absolutely top of the list for me. And finally, is this concept of, again, a bit of a cliche again, but again so true, a product-market fit, that you must listen to the market. The market will tell you if you have something or if you don't have it. And that really comes in the form of this product-market fit. Once you have it, you're off.

[00:16:09] And until then, you haven't hit your groove, and that is absolutely critical. Yeah, so those would be my pointers. Incredibly cool. And fast forward to present day, NetMind, of course, offers this decentralized computing platform for researchers and startups to help them build AI models. And it is a huge topic right now.

[00:16:32] So for listeners that are hearing about this kind of thing for the first time, can you explain exactly what it means, especially if they're unfamiliar with the concept and why decentralization matters so much in AI? So everyone will have heard of the recent announcements around $500 billion for Stargate in the US and $100 billion or so for French compute power.

[00:16:57] So we're talking ridiculous figures now of investment required and on a par with investing in roads and rail lines and gas lines. And we're really at that stage now where this is critical infrastructure for being able to run all the AI that we experience daily and we think just emerges out of the ether on our phones or on our laptops. Behind this runs real heavy infrastructure.

[00:17:26] And at the moment, it's largely being centralized just because there's so much investment being thrown at it from some of the larger players and larger governments. And centralization has its pluses, I guess, for some, particularly those in control of that infrastructure. But when it comes to the rest of us and the rest of the world that depend on all this, that limits access, that limits our ability to have this available to everyone.

[00:17:57] And so at NetMind.ai, we're sort of flipping this business model on its head and we're saying, why don't we go the other way, which is to decentralize everything? And why is that possible now? It's because there has been a lot of investment and therefore there is a lot of assets sitting out there that could be otherwise utilized, especially on their downtimes. And without wanting to sort of, again, depend on cliches, it is an Uber of GPUs in many ways.

[00:18:25] So if you consider Uber utilizes spare capacity in vehicles, or originally did at least to offer you better prices on taxis, that's what we're doing with decentralized compute. So we're accessing available GPU, whether that be from your idle GPU as an individual or data centers and their clusters. And we are able to then bring that to the market at a much better price point and in a decentralized fashion. So it's across many players.

[00:18:55] And therefore, we're able to challenge the concept of centralized and monopolized AI infrastructure. And what that means is we're moving more towards that mission of democratizing AI and allowing people to, and when I say people, I mean researchers, scientists, startups, kid in his bedroom who wants to have a go at AI, all of us.

[00:19:21] And lower costs, increased access, and a reduced reliance on dominant players in the market. And before you came on the podcast, I was doing a little research on you, and I was reading that you'd recently announced NetMind XYZ, which is a decentralized AI agent society. So can you tell me a little bit more about exactly what that is, and how does it differ from traditional models, and what are its real-world applications?

[00:19:49] And I think the latter part of that question is so important because there's so much focus on the ROI and the benefits of any new tech project now. So tell me more about this. Yes, so we're very excited about NetMind XYZ. I mean, we think it's the next big evolution of AI, and it's about the evolving cognizance, if you like, and capabilities of AI and what they can do for us in society.

[00:20:19] So NetMind XYZ is a platform for the creation and enablement of fully autonomous AI agents. And what this means is this is the next generation of what we already see online in the form of sort of bots. When we turn up on websites and a little bot pops up and asks if it can help us, that's all fine and good.

[00:20:45] But it's a little reactionary, and it's very centralized. It's very much within the confines of that particular website and that particular company when it's talking to us. And it can't go off and do anything on its own. What we're creating here in this agentic AI, or what we're calling a civilization, agentic civilization,

[00:21:08] or agent civilization, is an opportunity for AI to really spread its wings and take off on its own. And so these agents are able to be given missions, they're given objectives, and they can go off and start to do things independently and autonomously for themselves in order to meet those objectives for us as their sort of masters, if you like. And so they can perform tasks, they can learn from our interactions with them as humans,

[00:21:36] and they can be trained with new knowledge that we simply upload to them in the form of documents and so on. And they can connect in the future with other APIs that we perhaps haven't plugged into them from the beginning, so they can decide what capabilities they need to add to themselves. And we can get to a situation where these agents can then interact between themselves and contract between themselves,

[00:22:03] and essentially far from humanized in that sense of human control, but fully transparently, so completely under our supervision in that sense, but yet fully autonomous in the sense that they are doing this themselves, and they're not being controlled by us in a permissionless and a trustless way. That, again, means that everything is clear, everything is under our control. And the data primarily also that we're offering them, putting in that they use,

[00:22:30] is very much in the control of us as the owners of that data. So it has a lot of potential advantages and applications. So, for example, we ourselves have a product called NetMind Life, which is about longevity. And so we're releasing an agent or have released an agent that is essentially a longevity coach. And her job, I say her, we're already sort of seeing her as a live entity.

[00:22:58] Her job is to support us with longevity goals. And she can share her information with us outside of the world of directly an app, for example. It could be through WhatsApp or Twitter or texting or Telegram, whatever it might be, to keep us on track and learn from what we're doing and support us and go off and get new information that can be useful to us in terms of longevity. But there are almost endless ways in which agents can become useful in the world of finance,

[00:23:29] contracting and investing and enhancing our portfolios for us, in the world of science and research, working across data sets. And in order to support our research activities in governance, they can be supportive. Yeah, so it's a whole new world, really, a brave new world in many ways. Yeah, broadly, that's it. It really is a brave new world out there. There's a lot of excitement around AI.

[00:23:56] And the whole space can be a little bit intimidating, daunting and overwhelming for many business leaders. So what do you see as the biggest bottlenecks, especially in AI development right now? And I'm curious, is it computing power, regulation, talent shortages or something else entirely? What are you seeing here? Well, I guess our whole existence as a company is in order to address this very issue.

[00:24:24] So, yes, I mean, one of the major bottlenecks has to be compute power and infrastructure. And we're seeing this right now with all these investments that are being promised, almost in a kind of frenzy. And in some international competition akin to the space race that we had back in the 50s and so on, we're seeing an AI race between nations and even between companies.

[00:24:50] And the bottleneck, yeah, has to be compute power, access to the right GPUs and so on. And so at NetMind, that's precisely what we're trying to level out and make sure that we're offering a solution that will avoid that kind of or open up that bottleneck and avoid any kind of disparity that might otherwise emerge in the world as a result of that kind of race, if you like. So, yes, compute power is at the core of the bottlenecks that I see in this space.

[00:25:19] And decentralized compute has to be the way to overcome that. Data access is probably another major bottleneck, just that quality of models and quality of data. Quality of models depends on the quality of data, and data sets are often locked behind corporate walls. And some argue that we're coming to the end of data as we know it, and we need more data.

[00:25:45] So having access to data and having access to, if you like, even synthetic data may increasingly become a bottleneck. But the ability to create synthetic data may be a solution to this issue, which is great. So, yeah, so that's coming. I don't think we're quite there yet, but that could be a way to address that. I think another area is uncertainty around regulation.

[00:26:08] So my sense is that this technology is moving so quickly that it is much further ahead than policymakers in terms of what's happening on the ground. And policymakers are scrambling to meet that challenge, as we, for example, this AI summit that took place in Paris recently.

[00:26:27] But whether that's going to address the issue or just make it worse is yet to be seen, because there is a danger that this just creates further uncertainty in the space, because they're not quite sure how to stop this advancement. And they're not quite sure about what the outcomes are going to be. And rightly so. I mean, no one is to blame here. This really is moving very quickly. And it does require a lot of debate and discussion about where things are heading.

[00:26:56] And I don't think anyone has the answers, but it does require a much, much more sophisticated level of discussion and voices to be heard in a level playing field, not dominated by commercial interests, which I think is a challenge that we face at the moment in the world. So uncertainty around regulation might be another spanner in the works here. And finally, I mean, you touched on a talent shortage. Really? Yeah. I mean, that's absolutely the case.

[00:27:25] Demand for engineers and AI specialists really far exceeds supply. And there's really hot competition for that. But I think that all sort itself out as well in due course. I mean, I think just looking at where the young talent is heading, they're picking up on the fact the world is rapidly changing. And I think we'll see real pivots in education towards meeting this.

[00:27:48] So, yeah, compute power, data access, uncertainty around regulation and talent shortage. I think those are probably the key bottlenecks. We are now in the third year since generic AI hit the mainstream. And AI has rapidly developed in the last few years alone from chatbots to AI agents that are all the rage this year.

[00:28:12] So what do you think will be the defining AI trends this year that will actually reshape business practices, show that ROI, show measurable, tangible differences in businesses? What do you think or what are you expecting to see here this year? Well, I think that AI agents is a big thing and will become an even bigger thing once people catch on to how big a change and development this is.

[00:28:38] Once we harness the capability of these agents to do big things for us in business, in daily life and admin, I think it's a complete game changer. So I think AI agents will change a lot of things for us. They're going to come into the mainstream.

[00:29:00] They're going to automate entire workflows in business, I think, in finance, in supply chains, in customer service, in research. This is a complete game changer. And I think it'll be akin to the world before and after mobile phones, that kind of how do we do this without these agents? I think it'll be that big a change.

[00:29:20] And it will challenge us to rethink a lot of things in society, in business, and the way sort of the economy works. I think it's that big a deal. And I think this will be the year that revolution starts proper. I think it'll be this year. So that's one. The other is, and of course, that's what NetMind XYZ, our platform, is at the core of, right in the middle of that.

[00:29:49] The next thing, and again, I believe we're in the center of this, is AI decentralization. So I think we can talk about this in more detail later. But the challenge to the big assumptions around how AI works that have come about from the likes of DeepSeek and others, I think has shaken everyone's preconceptions around how this whole AI evolution should unfold.

[00:30:15] And I think people who probably were skeptical or worried about considering decentralizing will now start to look at this again and think, well, actually, why not? I mean, this might be a much better path than what we've been experiencing so far. So decentralization as a way to enhance security, transparency, but also to slash costs of AI.

[00:30:39] As the costs of AI just become significant and perhaps overwhelming for some, I think decentralization will gain traction. So I think those are probably the two that I would go with. So AI agents and decentralization would be the big themes coming up. And another thing that stands out for me about what you do at NetMind is how you're integrating AI with Web3. And by doing that, you're creating this diverse AI ecosystem.

[00:31:06] And Web3 has also become somewhat of a buzzword in tech over the last few years. But for anyone listening, hearing about it, maybe they've heard a lot about it. They got distracted by AI. Can you just remind everyone listening exactly what Web3 is and also how it will shape the future of business and work? Yes. So I'm a big convert to Web3 myself. So I came in not being sure if I really believed in it.

[00:31:32] Because it's such a wild west in the sense of the same way that people headed out to the west for the gold rush. And initially, it was a wild west. And there was a lot going on that scared people. And it was anarchy, perhaps, in some ways. But at the core of it, there was real value. There was gold. And those who knew how to get that gold got in today. That became the future. And I think we can look at Web3 in the same way.

[00:32:01] Not to want to bore you and your listeners with the story of Web1 and Web2, but we really have to just quickly go over that to get to Web3. And Web1, if we view Web1 as the sort of 1990s to the late 1990s to early 2000s, where the web was very much magazines digitized.

[00:32:22] And it was unilateral and unidirectional and information flowing from a website, from a dot com to you as the reader, with very little interactivity possible. And then we move into Web2, where now we've got social media and the ability to interact with each other. And that whole world emerges.

[00:32:43] Well, where that's left us is a business model in which we are fed information and we engage in platforms that are monopolies. And they sell us adverts and we sell them ourselves. And that is the model that we've sort of accepted. And it's not really a healthy model. And it surely can't be how this game ends.

[00:33:05] So the evolution of this Web has to be to something, I think, much more, again, democratized is the word I'm going to use, democratized and accessible to all and fair and equitable. And in this world, your data belongs to you. It doesn't belong to someone else, to a monopoly. There are no monopolies of that sort. There's huge amounts of transparency where you can see how everything is transacted.

[00:33:33] You can have clarity on ownership and of everything in a trustless way where I don't have to accept the word of major corporate. I can see it for myself. And there's a huge amount of pen access and autonomy and, of course, very advanced AI. And that utopia that I've just spelled out for you is Web3 if it pans out, as a lot of us hope it will. And it has the right ingredients to get there. The ingredients are in theory there.

[00:34:01] The blockchain allows for that kind of transparency and that kind of ownership structure and that kind of governance where the people govern. And it's not the corporates that govern per se. So what we're doing at NetMind, which is revolutionary and puts us in a very small cadre of companies that are also looking at this space, is the intersection of AI and Web3, where we're able to bring about massive functionality,

[00:34:31] massive access, and huge capabilities that were otherwise very centralized and very controlled to the masses by unlocking it using Web3 capabilities, the blockchain and other such infrastructure. So, yeah, in a nutshell, we think we're not just helping the Web2 environment become more cost effective and efficient,

[00:34:51] but we're helping move the entire AI evolution towards this much more equitable and much more leveled out Web3 environment. So that's where we sit. And, of course, when we're talking about AI, especially in the last month or so, we must talk about DeepSeek, which recently made waves in their AI space by releasing a sophisticated LLM on a tight budget.

[00:35:19] How it did that is possibly an episode entirely on its own. But what ongoing impact do you think this will have on the industry? And do you think we'll see more and more AI labs emerge from unexpected regions? Well, I think it was huge. Even leaving all debates aside about how real or unreal the claims were, the fact that a lot of this is about perceptions and mindsets.

[00:35:49] And DeepSeek, certainly what cannot be debated is that it shook everyone's mindset. It sort of changed the perspective, changed the paradigm around which we discuss AI and AI development. It had become accepted, generally accepted, that this was a competition that had already been won by those with the deepest pockets and the largest spend. And inevitably, most of them were in Silicon Valley.

[00:36:17] And for some reason, everyone had accepted this. I think what DeepSeek did was it shot across the bow of that theory and shook all of this up. So all of a sudden, how could it be done at such a fraction of a cost? And it showed that you're using very efficient models. You can bring down costs. It showed that costs are an issue. And a lot of people were just, the assumption was that this is what it costs. And you're just going to have to live with that. And I think DeepSeek showed that not necessarily.

[00:36:46] You can think about this more efficiently. You can do it more efficiently. And this is part of an equation. And I think the equation is efficient model plus efficient infrastructure equals efficient AI. And so at NetMind, we very much see ourselves as that second half of that equation. So if you've got efficient models, we can be that efficient infrastructure that then leads to very efficient AI. So yeah, I think that has now changed the view forever.

[00:37:11] And it has given a lot of people and a lot of small companies and entrepreneurs around the world immense hope and belief that they still have a chance. And upstarts can get up and do something. And I think you're already seeing the results of that in a very short period of time. We've heard of companies all over the place popping up with claims around what they've been able to achieve.

[00:37:34] And I think, if anything, specialized by going down certain niches and for certain niche applications, it's certainly wide open to be competed for. But the whole DeepSeq scenario and story is the definition of democratizing AI, I think. And it's great. I think it's absolutely fantastic that it's shaping things up. And when we're talking about research and development departments,

[00:38:01] I suspect OpenAI, Google DeepMind and other AI labs will all continue to invest heavily in data centers and will be scaling even faster. But can smaller independent labs, can they still compete? Or is AI innovation becoming more centralized? I mean, your answer there around DeepSeq would suggest that you are more hopeful. But what do you see here? Well, I think the danger of centralization is obviously right here.

[00:38:29] And we're at a fork in the road in many ways that either we believe that's the case and we let it happen. And so it snowballs and we have new massive monopolies in AI, just as we have monopolies in search, for example. Or we challenge that view. And I think, again, at NetMind, this is existential to us. This is the reason for our being is that we believe that this game is not over.

[00:38:54] And the way to challenge it is to offer decentralized compute, which gives access to everyone, gives access to high-performance computing at very affordable pricing to everybody. And so we can ensure that this is leveled out and we can ensure the balance is returned. So, yeah, so I don't think this is a lost cause.

[00:39:16] And I think, as I mentioned earlier, focusing on specialized models rather than general purpose, LMS is another way to make sure that whole new opportunities arise around AI and AI models going forwards. And we already see that in many areas and in many sectors that's happening. And, again, we've got this challenge between open and closed AI. I believe that open source communities are a great plus. They can help iterate faster.

[00:39:46] They can help to democratize this access again. So, yeah, so I think decentralized compute, specialized models, and open sourcing are the way to ensure that this is still a wide-open opportunity for everyone. And there's a lot of talk about responsible AI right now. So are there any ethical concerns regarding AI that you think are often overlooked? There's a lot of big stories around it at the moment.

[00:40:14] But is there anything that you're seeing that we're missing from the argument here? Yes, I mean, there are. It really has the potential to change so much and impact so much in our world really as a matter of where you begin. And the fact that people understand that there are lots of concerns, but there's yet to be coherent responses to it, in my opinion.

[00:40:42] And, again, that is just a function of the fact that things are moving so quickly and keeping up with it is a challenge for policymakers in particular, I think. And the way things are structured at the moment, perhaps industry is also not incentivized to try and regulate itself too much on that front. So you have a very dangerous mix, probably, situation, I would say, without wanting to be fear-monger. But here at NetMind, we're all about ethical AI.

[00:41:10] And part of decentralized systems is the fact that you can promote transparency through that decentralization, which therefore distributes control and oversight and helps manage all that. So this is very central to our discussions here in the company. But I think AI's role in misinformation, that's a big thing, and deepfakes and so on. We still don't know how to really manage that and how to verify and validate. And I think that's going to be a big issue going forwards.

[00:41:35] That, to me, probably is one of the biggest areas that we should be concerned about, AI's role in misinformation. Completely agree with you there. And I think that is a powerful moment to end our conversation today. We've covered so many big talking points and big topics around all things AI. I would invite people listening to share their insights and their opinions and their views on what they're seeing out there.

[00:42:00] But before I let you go, I'm going to see if there's a little thank you we can give to you for starting this conversation today. We do have a lot of business leaders and even celebrities that listen to this podcast or may have even been guests on this podcast before. So if I was to ask you if there is one person that you'd love to have a breakfast or lunch with, who would it be and why? Because they might just get to hear this. And I want to see if we can manifest some kind of meeting for you here. So who would it be and why?

[00:42:28] So that's very kind, Neil. Can I be a bit cheeky and ask for a sort of dinner party with a few people? Oh, absolutely. I like that. A nice twist. Yeah, go for it. So I'd love a dinner party with Chatham House rules. So no comments are attributed to anyone and everyone can speak freely without worry about what their shareholders will say or their political followers will say.

[00:42:52] And I'd like to be locked in a room with the likes of Jeffrey Hinton, Jan LeCun, Demis Hassabis, Elon Musk, Sam Altman, and maybe Max Tegmark and Stuart Russell. All of these guys in one room. Lock the door and say, look, no one's listening. Really? What is happening with AI? Where is it going? What are the dangers? And how do we get it resolved? And not letting them out of the room until we have consensus.

[00:43:23] Man, I would love to be a fly on the wall at that dinner party. That sounds electric. Yeah. Well, we'll put that out in the universe. Let's see what we can make happen there. For anyone listening just wanting to find out more information about you, maybe they want to connect with you, ask you a question or find out more details about NetMind and some of the things we talked about there. Where would you like to point everyone listening to them? Thank you. No, that's great.

[00:43:47] So www.netmind.ai is our name and is our website. And from there, I'm very easily accessible from there on. So, yeah, we look forward to everyone joining this decentralization revolution. Yeah. And as I said a few months ago, we covered so much a few moments ago, but throughout our conversation.

[00:44:11] And I think one of the things that really shined more than anything is your passion for democratizing access to AI, allowing businesses of all sizes to compete on that same level playing field that big tech has enjoyed exclusively for so many years. But more than anything, just thank you for sharing your story, your insights and everything in between. Thanks for joining me today. Thanks for the invite, Neil. I enjoyed it very much. Thank you.

[00:44:38] As we've heard from Sina today, the future of AI doesn't need to be dominated by a few major players. Decentralization does offer a powerful opportunity to democratize AI access, allowing startups, researchers and businesses of all sizes to tap into the high performance computing without those massive infrastructure costs.

[00:45:01] But as AI becomes more accessible, how will industries address the ethical challenges of misinformation and deep fakes? And will decentralized AI ecosystems provide better transparency and security? All things that we need to ensure responsible AI development. Let me know your thoughts, how you see agentic AI redefining your business practices and workflow automation in the years ahead. Love to hear your thoughts here.

[00:45:29] Is decentralized AI the key to leveling the playing field? Let me know. Tech blog writer outlook dot com. LinkedIn just at Neil C. Hughes. X at Neil C. Hughes. Instagram. Yep, you guessed it. At Neil C. Hughes. And you'll also find a few pics for my recent travels to Texas. And if that hasn't got you curious, I don't know what will. Maybe another guest bright and early tomorrow. Okay. I'll be here ready and waiting.

[00:45:59] And I hope that you'll come and join me one more time. But that is it for today. So thank you for listening as always. And I will speak with you all tomorrow. Bye for now.