Life on the Edge: AI Transforming Smart Infrastructure

Experience "Life on the Edge," where AI and edge computing intersect to transform smart infrastructure worldwide. In this episode, host Joseph Celestio of Blaize is joined by Francisco Sotto of VSaaS.ai and Daniel Pfeiffer of OrionVM to showcase cutting-edge advancements in AI-driven solutions and hybrid infrastructure. Discover how Blaize’s secure, power-efficient AI platform is driving real-time analytics and solving critical problems in sectors like retail, defense, and smart city solutions. Learn how VSaaS.ai enhances video analytics and OrionVM’s custom cloud stack powers future-ready AI workloads with remarkable efficiency.

Explore the game-changing potential of these technologies and partnerships that are reshaping industries globally. Join us to see how AI innovation is creating real-world impact and transforming the edge. Don’t miss this exclusive insight into the future of smart infrastructure and AI solutions. #edgecomputing #machinelearning #smartcitysolutions #aiengineer #datavisualization

#aitechnology #smartcities #hybridai #aipartnerships #videoanalytics #practicalai 

#cloudcomputing #orionvm #edgecomputing #blazeai #aitechnology

CHAPTERS:
00:00 – Welcome & Introductions
Overview of Life on the Edge podcast, Blaize, and today’s partners VSaaS.ai and OrionVM.

01:13 – What Blaize Enables: Practical, Hybrid AI
Joseph sets the stage on Blaize’s mission and value in real-world AI deployments.

02:08 – Partner Stories: VSaaS.ai & OrionVM
Founders share their origins—from wildfire detection to supercomputing-inspired cloud.

05:09 – The Infrastructure Challenge in AI
Why cloud, accelerators, and IT complexity make AI deployments difficult.

08:07 – Scaling AI for Thousands of Cameras
VSaaS.ai explains cost barriers and how the partnership eliminates them.

11:46 – Making AI Turnkey: Cloud + Edge + Software
How OrionVM, Blaize, and VSaaS.ai simplify deployment for real customers.

18:36 – Real Customer Impact & Cost Savings
Examples of scaling from 40 to 600+ cameras and reducing infrastructure cost.

23:49 – The Clear Market Signal for Practical AI
Final insights and the future of practical, physical AI powered by partnerships.

Transcript

0:05

Thank you everyone. Welcome to life on the Edge podcast, where we explore how AI and edge computing, together with cloud and data center, are transforming the life on the edge for communities, for market verticals, and for the physical environment that we have worldwide. I'm your host, Joseph Sulistyo, senior vice president of marketing at Blaize. And today I'm joined by these two innovators that I have and that I'll be proud to call partners. That's driving the next generation in smart infrastructure and AI. I have Francisco Soto, CEO of SAS, that I and I have Daniel Phifer, CEO and VP of partnership at Ryan VM. Francisco. Daniel, thank you for joining me. Great to be here. Thank you. So before we start, let me provide a quick, introduction about Blaize. Blaize is a trusted AI platform for a practical and solution driven intelligence. We enable hybrid AI infrastructure, which is heterogeneous in nature, to deliver secure, power, efficient compute and software to solve real customer problems and drive measurable business outcomes through critical sectors in the markets of verticals such as smart infrastructure, retail and defense. I would like to also to give the opportunity for Francisco and Daniel to introduce themselves and also their company. Francisco, why don't we start with you and Daniel. You can continue. Hi, everybody. Thank you. You're so great to be here. My name is Francisco Soto, founder and CEO of VSasS. We are video analytics platform. We started like seven years ago. So we tried to solve the problem of CCP control rooms by enable AI for structure as a service. I'm Daniel Phifer, COO and VP of partnerships here at Orion VM. Orion VM was founded in Sydney, Australia. The founders of the company were inspired by supercomputing at CERN, and we have built our own custom cloud stack that's ideal for hosting both AI and general purpose workloads with exceptional performance and efficiency. We were one of the first pioneers to build a Hyperconverged stack with InfiniBand. And InfiniBand is the interconnect used in 450 out of the world's top 500 supercomputers. And we now have, prominent reseller channels across North America and Australia. But we are expanding globally very quickly. That's excellent. Daniel, actually, can we touch a little bit more about that? And also, can I expand a little bit more on the origin story? Why don't you tell us a little bit more basically, how did that journey, how did that kind of transition happen? You start this in the supercomputing architecture, right? And hype in in HPC. Yes. How did that foundation lead to the cloud model for AI infrastructure that you're building today? Yeah, no, it's a good question. So, at the time, the founders of Orion were, looking at the technologies used in supercomputing, and were very ambitious to take that tech and make that available for general purpose cloud. And so we, we built, as I mentioned, one of the first hyperconverged stacks, where we can now, merge, compute, storage, networking to into a single layer and very efficiently scale. And then that was used to now power a white label cloud service for various partners, telcos, MSPs, system integrators. We simplified we basically take a complex, set of infrastructure. We make it very easy and we effectively sassafiy IaaS. We make it easy for our partners now to brand and build their own cloud computing solution and bring it to market under their brand, and do that very quickly, and have that brand identity and ownership. And also a differentiation in terms of the performance and cost savings. Like I like the word sassafiy It sounds so sassy with, sassafiy your iaas. And iaas in this case, because I buy an infrastructure service right from that perspective. So I guess the idea is the is the first foundational model in the in any cloud offering, right? You first start with the infrastructure, you provide compute storage. And you know any of those type of a those commodities typically. Right. You get from the from the server and system. Yes. But in this a infrastructure is a little bit different because it is a little bit more complex. It is is is not that well defined as we have a typical server, right, in this case, right, definitely. And we have, we have a product called a micro pop where we can deliver a managed cloud appliance on prem or into the data center of choice. And to give that control and give a private cloud, into a facility where, the partner wants ownership control and, you actually be able to optimize that stack. That's kind of a differentiation for Orion. There is obviously the hyperscalers are the big guys. You know, you can get a multi-tenant solution, but usually there's a sacrifice in terms of security. Also there's a lot of those systems are are cheap to start out, but then very expensive as you scale. So we kind of fill a gap in the market. We, we see between buying and building your old cloud or using a hyperscaler. That's kind of the sweet spot for Orion. I love that sweet spot. What's what is actually what makes us to be, you know, kind of a dynamic, robust and kind of a keep the differentiating right. Very competitive in the market. So absolutely love that. Well, Francisco, you, you know, I knowing you, you know, talking to you basically, within the last year here basically you I know you as a partner, you actually you have a compelling story as well, right? And, Ray, you started with, your company started with solving wildfire detection challenges. Share us. Basically, how is that early problem in in kind of wildfire detection? The shape, VSaaS right now, your platform, your your roadmap and then where you're going to go. Yeah. No, we started like with this problem with wildfire. It was, electrical company. It has so many have so many electrical substations around Chile. So there was a big wildfire. But then so before we start, we used to come from the IoT space. So we tried to connect every everything. And these customers told us, hey, you can put in everything. Why don't you connect my cameras to detect wildfire? So we were like, not again with my co-founder. So. But, we connect those cameras. And the problem that we did understand with we, we understood the problem of wildfire. But, then we understood the product of emergency that 24 seven, because most of this company has this infrastructure like camera, CCD control room, and they have behind that all that infrastructure. They have people watching everything. So all the emergency was, the trigger alarm was a human. And the problem with that is a human is not ready for it. So they buy these expensive infrastructure like a camera, everything. But, the most important component that was who triggers that alarm was that was a person or a set of person watching this. We we try to detect those fire or smoke and you see an algorithm. But we understood that the problem was the algorithm itself was solving this problem, that it was an emergency. And somebody has to be 24 seven. And then we saw that, oh, that problem. Many companies face that this problem of alarms the wildfire. What for us the critical are most obvious. Thing is, then we understood that not just electrical company have this problem municipality retail. All they have this problem. They have the infrastructure. They have people behind that. So, we understood that, first of all, the opportunity was enable this asset infrastructure and put AI on top of it. And then the second problem, how how do you make it not expensive? Because I always make a joke. You can drive over with your car by buying a Ferrari. It looks cool, but as success, what you get, it will never scale. So there we we solved that. And that's why at some point, we met place. Our decision was how we scale it and this partner and we obviously with OrionVM and we, we, we believe that, the problem of scale in this connecting thousands of billions of cameras in the space of emergency, we can make it happen using AI. That's excellent. I, you know, obviously, I heard some of those, those those triggers. Right. And I think that, the connection may, you know, a lot of people basically may not see that, like, the obvious things that start from connected IoT. People just want to be connected in, you know, to in this physical environment in which what happen. And then from that detection, as you mentioned, there's an urgency, there's an emergency aspect of it. Right? Because if something happens, I have to know right away on real time. But the problem is but how is the price if I want to connect not one camera? Because all this infrastructure that we connect is, 1000, 10,000 kind of project. So then becomes intelligent that the customer say, hey, I understand the, the value of AI but the problem that I'm not willing to pay for that because they have two options. I move to AI Yeah, but then it's too expensive to stop. So the challenge is that the speed of implementation is not moving faster because there are so many infrastructure available today in the AI space that, that for this area of surveillance, people are not willing to pay for that infrastructure right now. So what we have created is a way that that problem is solved because we take care of that and nobody's seen it, that everybody's like in this shadow of of reality that AI is going to solve everything, which it's it's going to do that. We already know that. But the challenge that is coming and now we trying to solve that is how do we make it accessible for a company to have ten camera or have the company that has a million camera? Yeah, that's that's a great group with them. That's a great point, Francisco, how you get there. Right. Because everybody knows AI is transforming everything. But how do I get there economically? How do I leverage my existing infrastructure? And how do I do it in a way where I can integrate with my existing I.T environment? That is a challenge. And I feel like that's kind of the outcome of our partnership between Orion, Blaize and VSaaS. We we bring expertise at different levels that solve that problem in a turnkey, simplified way. You mentioned about sassifying the iaas tell me a little bit more about those things that in AI, how would you sassafiy those infrastructure AI can have a great complexity in terms of the hosting environment. So you have various accelerators, you've got your compute. And a lot of these companies aren't, built to actually have a sophisticated, IT department themselves and build that out. But they want a certain amount of control and they want a certain amount of, scale where they can start maybe small and then then scale to, very high load and support lots of cameras, lots of, lots of clients. So, that's something. So, Orion, we're in that cloud space where we simplify that process. We have a platform where a partner can very quickly come and, set up their environment and deploy their applications. And, you know, we we've integrated with VSaaS, so their, their system and their platform for managing these cameras and then leveraging Blaize for the inferencing, it can all can be done in a configurable, configurable turnkey manner. And so that complexity really becomes reduced. But now you have a sophistication of functionality, that really now can support, a very, complex video surveillance environment. You can leverage existing CCTV, and VR systems and you don't as Francisco was saying, you don't have to rip replace. You can you can take what you already have and now add this layer of sophisticated threat detection on top of it very quickly, Francisco, to you on this one, based based upon that infrastructure, now based upon Daniels, let's say, you know, assessment and also what he has done, what what Orion VMs he provides is ready. Do you agree with that? I mean, we started with that philosophy the first day seven years ago with this electrical company. We understood that customers already have a structure. So. So you don't need to replace that. Yeah. It's not going to do that. And in another industry, the same philosophy in this transportation industry, Uber made the same thing. You know, they believe that any item transferring to a taxi can change the industry or what happened with, Netflix. They they thought that you don't need to go to a market to get a video. You can do that in the same in a, in the cloud. So absolutely they were trying not replace the industry is just telling the industry, hey the industry already has camera. So that's the first thing that changed the space. The second thing is that I have to somehow in terms of emergency, understand the emergency. It's not just the detection because wildfire, the good thing about wildfire that everybody start that. But the problem is not detecting fire is detecting is how this fire is going to spread. We have worked that over the past seven years. We incorporate that models. And when when we find partners like OrionVM or Blaize that we can make it happen faster. It can really grow today. The industry somehow that other companies that are trading together, nobody can do this thing by themselves is, is, is extremely expensive. And time to market is today or yesterday. I believe its yesterday. I think we are solving this problem end to end. Daniel was saying that sometimes end user that doesn't have these IT department, so they don't know how to do it. I mean, they don't understand the AI, but they look back to the company, say who's going to manage this? So I agree that's one, one part of the problem. Yeah. Yeah. Even the AI who I'm where I'm going to buy this all this AI cheap infrastructure. The good news is that we already understood that. So we are solving it. And people who are like customers, all the companies that we are talking to and we are telling them, hey, where the right partner to make it you become, reseller or you end user can to transform your entire company, as a SaaS model, because you already have that expertise, that complexity, and what we solve and remove that headache for the client, they can now focus on delivering their core services. So that's another thing like this. Complexity takes away from their their main, the main value that they're bringing in terms of delivering a service or have a product offering. So by the fact that we simplify that, they can actually be more effective. Now on what they're focused on, whatever it is we're providing, either from the software stack or from the hardware stack or from the actual infrastructure itself. Right. The manage it as a service. It has to solve business problems. Daniel, break this down to me. Basically when we talk about solving, you know, real business issues, right? Yeah. You know, and also tell me about, tell us about the some of the deployment that you've had because as a cloud services veteran here, seeing everything from that perspective essentially as it goes through. Okay. Now it's comes to deployment. It's not science experiment anymore. How do I support this. Yeah definitely. And cloud cloud you know is is the hosting environment for all these applications. Right. So we understand that very well. And we also understand that you need innovation and a tech stack that can be cost effective and also easy to use. So yeah. No, we're seeing that across all industries. We're seeing we're seeing the ability to now bring the right partners together because, as you I think mentioned at the start of this, it's it's about the data center, it's about the cloud infrastructure, and then it's about the application layer and having the right configuration and the right, integration and orchestration between those components. And that's not a simple thing. I mean, there's, there's companies and companies that are focused on one layer of that one silo. But to bring that all together in a way that's, smooth and optimized is a challenge. And that's, that's where the partnership between Blaize VSaaS and Orion VM is we are experts at what we do. But then when you aggregate that together, we're we're now delivering a solution that as you're saying, Joseph, is solving business outcomes. It's allowing video surveillance. Distributors to now leverage this solution with their existing, cameras in the existing ecosystems they have and now add a very sophisticated layer of threat detection, smart and smart inferencing on top of that, in a way where they can now deliver those core services, which is this is a chance, not an easy thing to do. Francisco, why don't you mention a little bit about some of the cool things that, you know, like the, moment from the customer side. One is that it's very easy for us is, before some customer have one operators that they manage like 40 cameras in real time. Now the same operators have managed like 600 cameras. Our goal is 1000. And customer like before they were like, oh, you're like this, this when they see it, I mean, they opened their eyes because it's not just the, the manpower that it's reduced. They look at, oh my God, the productivity of that person there, because it becomes very powerful when they saw what we could do for, say, $25 per camera, they it was like too good to be true because there's, a lot of systems that, force you to rip, replace and, CapEx outlay to, to deploy a sophisticated service. And the fact is, now with our partnership, we can take any IP camera and now add this, this layer of, threat detection and a whole, you know, video data surveillance platform on it very easily and very affordably. So, there was that sort of light bulb moment, where we had, a person after person go, hey, you know, there's people that solve one part of this problem, but nobody that solved it in the way that you guys have done it. We convinced one municipality. We started with 50 cameras, and we connected the cameras from the neighbor for the cameras that were looking at the the street. So before that, most of the municipality had to buy the street pole put the camera, put the fiber and the cost of that infrastructure was like $20,000. for a pole just the infrastructure. Yeah, we reduced that to $200. That's insane. So the customer was because no, nobody believes it. And yeah, I didn't believe it. that was funny. But when you tell the customer, hey, let's do a demo, let's deploy this, the customer first, like, hey, for instance, one guy told me, you are just like changing my mindset because over the past 30 years I just put poles in the street. We we here, on and, you know, kind of, on and on again about leveraging. Right. And I think that the word itself is that complement instead of rip and replace. And I think we have to kind of, to to take that as a, as our mindset as we move forward. Right. Is that the, you know, AI does not mean that you need to replace. You need to actually to change anything. You have to change anything that you do. In fact, this guy can help you. And and this is where we are, this this kind of a de value. And I can see directly, basically if anyone ask me why, you know, what is it about VSaaS that. That's, that's that's cool. What is it about Orrion VM that's cool. I will give you this one... word. Very simple. You guys work. It works. Let's see our closing thoughts. Daniel, start with you, I guess. Call out first. You know, it's still early in the cloud days. It's the first inning of the baseball game. You know, there's there's been a lot of a lot of hype and noise in. And we're still in early days and things that are being developed and promoted now our foundational and, you know, you need to find the right path for adoption. And I think what I'm excited about is we can, with some of these solutions, you can show of value immediately upfront without a huge financial commitment, without a huge, technical investment, you know, work with partners that can show you the value by, giving you demos, POCs, with your existing environments, and that if they can't, you know, maybe there's something, not fully mature there. So we're it's early days. And look for the innovation. Look, look for something, you know, maybe under the radar. That's not just the completely, you know, the traditional solution. And there's a lot of emerging tech, a lot of partnerships happening that are bringing these solutions that solve a bunch of problems. So, it's exciting because, you know, we're seeing, that aggregation and that collaboration happen. But to get that out to the consumers, there's, you know, more effort and more marketing is needed. Excellent. And to you, Francisco, the word now is, is, is confusing because every day there's all this new data center. So there's a niche of those type of clients that they will buy this huge infrastructure. But in another hand, this huge demand that they don't they're not willing to pay for that. They're willing to consume that. And we are trying to focus on that niche. I'm going to call that to be practical AI, a practical physical AI. Right. And then so, Francisco Daniel, thank you again for sharing your insights. It's been incredible. So what we heard today is is clear, folks. There is a clear signal. There's a clear demand in the market, where people want to deploy. Yeah, people want to see AI in their life. And the partnership that we have right now allows us to do that. And with the right technology and the right mindset, we're able to bring you this solution to our listeners. Again, thank you for joining us. Thank you for listening to us. Be sure to subscribe. You know, click that subscribe button on life on the edge on on our channel, for more conversations with the leaders that will shape and has been shaping, AI, physical AI, practical AI today and for the future. Until next time. Thank you.