The Future of Edge AI & Portable Data Centers with Rahul Narsimhan (Hive Radar)

Scott Mann

Welcome to PodMagic, real conversations about solving real IT challenges. I’m your host, Scott Mann, SVP Global Sales at StorMagic. We’re always exploring how simple, reliable technology can benefit you and the people you serve, whether you’re running branch offices, retail stores, or supporting customers on the front line. Our goal is always bring interesting guests, deliver some value, and have some fun along the way.

Now, today’s guest is someone I’ve had the privilege of knowing and working with, for what he corrected me, for 14 years. And he has constantly been one of the most forward-thinking minds in the business from his early days in security to creating insanely clever, ultra-portable data center. He’s always been solving those tough problems. Joining us from Canada is the founder of Securicore and the CEO of Hive Radar, Rahul Narsimhan.

Rahul Narsimhan

Hey Scott, you got my name right. Good to see you and good to be here.

Scott Mann

Fun backstory. I had to pronounce Rahul’s last name on stage in front of thousands of people and I thought I did a great job and he came up to me afterwards and he was like, “you know what, Scott, it was a good effort, but you did terribly”. I was a little more nervous doing it again this time.

Rahul Narsimhan

You got it right this time.

Scott Mann

I appreciate it though. This is fun because I’m very familiar with everything that you do and your business. We’ve known each other for so long in different capacities and this isn’t an introduction to me of what you guys are doing, so I think this will make for a very fun experience within the podcast. Let’s get right into it, we can share some of those stories along the way.

In our world, we’re seeing this massive push towards edge. When we first met, the edge wasn’t a buzzword yet, but I want to start looking at that big picture. Are we witnessing a true inflection point in how enterprises think about IT architecture, or is this just a natural evolution of what’s already happening?

Rahul Narsimhan

There’s a lot to unpack there, it’s a natural evolution. But you remember when we did our very first joint call back in 2013 or 2014, we went to a small municipal customer and he was used to running bare metal server with just one Windows instance running and we talked to him about virtualization and how you could run multiple VMs. And he was completely, what Hocus Pocus is this? Why would I want to run three servers on one physical hardware box? We are at the exact same point here right now that edge has now become a little more mainstream. The curveball that we have got right now is the integration of AI. I think people are coming grasp, okay, what do I do with AI at the edge? And that’s where the challenge is going to be for the next little while.

Scott Mann

When you went through that I just had flashbacks, I can remember sitting in that office, in that boardroom talking.

Rahul Narsimhan

He comes from the old school. He is used to installing a Windows server, from a CD or whatever else and he has got his Windows server set up, but then we talked about virtualization. He had heard about virtualization, which was where VMware was the only game in town at that time or running parallels. And then here we are talking about the latest, greatest in HCI. “Hey, you can run multiple workloads, you can have it completely resilient”. And he just could not fathom. It’s also the learning curve that’s there, right? Because you’ve never heard of the technology. He’s just used to the old school. Now he has got to learn something new, so I think there was a little bit of resistance that’s there. And it’s the same now, I think what’s happened is, Edge is no longer an emerging technology. Edge has been there for 14 years. It’s somewhat new for StorMagic because StorMagic’s biggest thing is SvSAN. That’s your bread and butter, which has been there for years. You’ve moved into HCI in the last I think two years or three years. There’s a little bit of a new curve, a learning curve that’s there, but it’s essentially a mainstream product right now.

Scott Mann

I think the edge always existed. It’s just redefining the edge and edge AI is like you mentioned, that’s to your point, that’s the evolution of it. Even for us, the same customers that are coming to us with these edge AI use cases are the same customers that, some of them have been using our solution over the last ten years. It just wasn’t called that, and the needs have evolved. The requirements because of AI and everything has evolved.

Rahul Narsimhan

Correct. But even the customers are learning, right? Edge AI is like the buzzword. Everybody wants to do AI, but half of them don’t know what exactly to do with AI and in their environment. They’re learning and we are trying to figure out how to guide them on the right path, as to what you can do with it. So that’s the other challenge that’s there.

Scott Mann

Yeah. Absolutely. From your perspective, what’s changed in the last few years that makes deploying infrastructure anywhere not just possible but necessary?

Rahul Narsimhan

Whether you look at Proxmox or you look at Scale Computing or look at VMware, Edge has always been there. You can either run it in the cloud or you can run it in your own premises, whether it’s a co-location facility. I don’t think anything’s changed from that. What has changed is the complexity. There have been shifts. During COVID, everybody went to the cloud because everybody was working from home. Now the whole mantra is changing. It’s coming back. There is also a couple of other things that’s happening, the fatigue for everything sitting in the cloud and paying the online subscription fees where the costs are unpredictable. There is a push to bring everything back in-house where you’ve got a little more predictability in pricing. So pricing is a little more predictable. If you want to run one workload, say for example, in a cloud instance is cheap, that’s fine. But when you run multiple workloads, it’s not cheap. Now you add redundancy to it. “Hey, I want to run it in multiple zones within say East Coast”. That becomes more expensive. Now you want to run it in multiple, you know, East, West, Central, whatever, the price goes right through the roof. How do you bring all that stuff and make it easier?

And the sheer volume of data that’s getting generated. Sure, SpaceX has got Starlink and bandwidth is becoming easy, but there’s still latency. How do you give that under 10 millisecond latency? Your applications have got quick access to it, but only if it’s local, you’ve got fast access to that stuff. It is going to continue evolving.

If you have seen in the last two, three months we had a few outages, both Amazon and Azure had outages where there’s somebody who had his heated bed just stuck on heating and he couldn’t turn it off because the cloud control was gone. There’s different stuff, for example, the point of sale. A restaurant’s point of sale stopped working because they had no access. I’m not saying that it’s got to be cloud or the edge, there could be a balance in terms of hybrid. You’ve got stuff in the cloud, but you’ve also got a backup plan which is the edge, which, if cloud goes down, I have got my backup plan there.

Scott Mann

It’s less thinking about edge versus data center as a destination and more thinking of it as a complete architecture. Physics doesn’t change, bandwidth is bandwidth. It takes so much time for data to travel from there and back. Eventually maybe it’ll feel like there is next to zero, but then at that point AI continues to evolve and you need even less micro milliseconds, whatever that is.

Rahul Narsimhan

For example, if you look at Starlink, the good part with space-based communication, you’ve got connectivity in areas where you could not dream of connectivity. But what that has done is that whatever you’re sending to space has to come back to a ground station. That ground station is connected to the rest of the internet feed that’s going everywhere. So there is going to be some latency. If you look at the current geopolitical nonsense that’s going on, there could be interruption, there could be outages, there could be rogue nations that are targeting your space network that’s there and taking down communication. You have to have a fallback. I don’t have any internet;  can I run my entire operation without connectivity to the internet? That’s the biggest challenge that you’ve got. You can run stuff in the cloud, but if cloud goes down, connectivity goes down, what do you do? Does your business come to a standstill?

Scott Mann

The next big buzzword is data sovereignty and having control over yourself. Even while you need the public cloud in some instances, still being able to maintain your own sovereignty within your data center, it’s only going to grow in importance, and it’ll be that next big buzzword as AI continues to grow.

Rahul Narsimhan

Actually, it’s already there. When you look at stuff running on OpenAI or Google Gemini or any of this stuff, companies will warn you, don’t put your proprietary code on any of these platforms to do the modeling. They say keep it local. You have to build your own GPU server farm that’s there, which is doing all the inferencing locally for you. That is going to happen, it is not going to change. It’s going to be a little more amplified because they know the dangers of this stuff sitting in the cloud. I think it’s just going to amplify.

Scott Mann

Yeah, agreed. When people hear edge infrastructure, pretty much everyone thinks that remote mines, defense or disaster recovery. We know that’s not the whole story, but what does that whole story really look like?

Rahul Narsimhan

That is a low hanging fruit the way I see it. Because when somebody talks about a portable data center or tactical data center, the first thing that comes to mind, military, defense forces or emergency response or disaster response, and it’s not something unique that I’ve built. We’ve got a different story behind why we created the product and the life that it’s taken right now. There are many companies, and I think you guys had one of these companies on the podcast several months back, which manufactures a portable data center too, but they’re only geared after one market or focused on one market, which is defense. Because some of these products are literally like built like a tank. A short drop, blah, blah, blah, whatever you do. It’s not that they’re bad products, they’re great products, very proprietary. Which means that the only guys who have got a budget is your defense, where you’ve got governments bankrolling their entire thing. We didn’t want to create anything which is just defense or emergency, that’s low-hanging stuff. Can you make it affordable for the remaining 95% of the consumers that are in the market.

For example, you take retail chains. You’ve got a thousand locations that are there, even with a giant enterprise operation they don’t have a thousand technical guys who are field reps that are going to go out. They probably have twenty guys sitting in the office and they’re going to roll out all the IT infrastructure for say a retail chain, right? If you look at it from their perspective, they’ve got say 5,000 locations to roll out. When you go the traditional old way of doing this, you order your switching rack, you order your firewall, you order your servers, and the guy’s going to get his airfare, hotel stay, and by the time the rollout happens, it’s going to take four years before they’re done rolling out all the sites. What we have done is we have spun the whole thing around on its head. He says, “Hey, you can get this entire thing, pre-configure it, ship it on site. It’s already pre-configured. One cable to plug in and you’re off”. You’ll be completely deployed in six months versus four years.

There’s a slightly different shift, and how do you keep the cost down? The other guys, they’re expensive. We are at one-fifth the cost. There is a little bit of cost dynamics there. How do you make it affordable for all these entities? The other emergency disaster response kits are great and it’s not that you can’t use ours, you can use ours too. But we’re trying to cover everything, the entire spectrum.

Scott Mann

It’s making affordable data centers more relatable to everybody and less niche as this edge AI market continues to grow.

Rahul Narsimhan

I won’t say it’s less niche. I mean obviously AI is something new that’s come up, but when you look at these portable data centers that are there, traditionally they’ve been very proprietary, the hardware is very proprietary. When you’re building proprietary hardware, it is not going to be cheap because you as a company are building your own custom servers, and once you’re on that ecosystem, you’re stuck on that ecosystem. We had to think differently. How do you take commercially available off-the-shelf hardware and integrate it into a custom design chassis? But we had other challenges because the initial plan was, keep it very simple. But then as we went along, the more things got out. That’s why I said, it’s taken a life of its own, but we are trying to cover every market from education, from not just emergency disaster response or mines, but for education systems, for retail chains, for grocery stores, for lab research teams. I mean the use cases are many. And even within defense, for example, just because you have a tank doesn’t mean a tank has to be for every environment. Some environments require a tank, some require a Humvee, some require an ATV type thing or whatever that’s there. You can mix a match. That’s the beauty about the product. The area that requires a tank use that super hardened product. The area that doesn’t require a tank, you’ve got a more affordable product that’s there. So different thinking.

Scott Mann

Let’s expand on that. You’ve been an inventor all your life. You’ve got entrepreneurial blood flowing through you. To invent something, you see a problem, you see a need that needs to be addressed. When you were creating Hive Radar, like what were some of those needs that you were trying to address or those problems that you were trying to solve?

Rahul Narsimhan

The inventor streak has been there for a very long time. When I was a consultant, I was trying to build my own anti-spam appliance back in 2003, till I stumbled upon Barracuda networks. And I said, “Hey, these guys have done a better job”, so I became a value added reseller and we started promoting and selling Barracuda anti-spam appliances and that’s actually when we met.

The origin story of Hive Radar is a little different. The Hive Radar was originally supposed to be a real-time triangulation or location services app. Basically you were doing a lot of wireless deployments, we were coming up with an app where you can, on a map locate all the wireless devices, which could be individuals walking around with the phones in a physical facility. Think about Harry Potter, that Marauders Map that you could see Dumbledore walking around. It’s the same way. He could see every student walking around, but we had a different idea behind it.

But anyway, that project did not happen. The next evolution of Hive Radar became the wireless survey kit. Because we were doing wireless site surveys for businesses and the old gear weighed about 150 pounds. We condensed it down to an airline carry-on friendly. That was our very first design pattern. It started as a fun project and continued, and we’re happy doing what we’re doing. This was selling online all over the world. We’ve got for our wireless survey products, we’ve got engineers in Fortune 1000s that are using our product. There is some awareness within the wireless community.

Fast forward to now, the Portable Data Center was not even on our radar till we actually met you guys at Platform in Vegas in I think 2023. If somebody doesn’t know what Platform is, that’s Scott’s previous place where he was with Scale Computing. They used to have an annual event in Vegas. We ended up seeing some of your engineers walking around with a guitar case where they had a kludgy three node server instance. That’s when we looked at, “hey, we can do a better job”. We did an initial design concept. Some of your colleagues at that time were involved in the project. That’s where the original design started off from.

Since then, it’s completely evolved. How do you take all this product and make it simple to deploy? But now you’ve got multi-vendor technology. How do you give a unified monitoring interface for multi-vendor products? Because StorMagic has got its own interface. Mako firewall, which we use, has got its own interface. The switching vendor has got its own interface. Now I’m not trying to recreate the interface, but I want to give our users a unified dashboard which allows them to monitor, which is what they’re going to be looking at on a daily basis because changes are not something that happen daily. You set it and try and look at it maybe once in two months or once in a month when there are new updates, patches, or new VMs you have to create. You don’t have to keep doing it on a daily basis. That was the other challenge. As things have evolved it’s added on but whatever we are doing, we’re trying to make it fun for us. If it’s not fun, I’m not going to go down that path. And it has to also look cool, aesthetics matter. Sure, you can drive a Russian Lada from here to Whistler, but if you’re going in a BMW, obviously it’s a nicer car. It’s got a little more bells and whistles and everything else. That’s where we were going with Hive Radar. You can buy a rack case and put your servers in it, but it’s just not the same thing.

Scott Mann

I speak with a lot of enterprises and the big thing in today’s world; they’re looking for that turnkey. You’re basically bringing everything from the virtualization down to the switching layer into a turnkey solution that you can present an interface with and all of those that work together as well if they want to have their own tie into whatever that interface is that they’re using with.

Rahul Narsimhan

Absolutely. The other challenge that you find, you can buy rack cases a dime a dozen. They’re from four hundred bucks to two thousand dollars depending on what you want to do with it, but they’re just rack cases. You can mount your rack servers on it, but there’s nothing cohesive. But what challenges that you find, because now we have got internal wiring and lighting and all this stuff has to get industry certification. We had to go through our FCC certification, C E certification, UL reports, all that stuff, because you cannot sell a complete package. It’s easy to build it as a lab one-off, but if you’re making a commercial product which has to be sold, you have to go through all the compliance. No choice.

Scott Mann

That gives you that global scale. We’re talking from Canada right now, but this is a globally available solution, and I know we’ve done events with you literally around the globe. Just a couple of many to go.

We’ve talked about resiliency. How does your how does your definition of resiliency change when you move outside the traditional data center?

Rahul Narsimhan

Resiliency can mean many things. You can have data resiliency, you can have data center resiliency, you can have, should my hardware failure happen, I’ve got backup hardware failure. But in the grand scheme of things, the way I look at resiliency is, how soon can my business be up and running should something happen. I think that is the big picture for everybody. Having your integration to a backup solution and having backup data in the cloud is one part. But say if your building burns down, you’ve got nothing left, how do you get your data up and running? You can run a cloud instance on Azure and try to pull all your data which is there, and your bills will be right through the roof, in addition to the insurance chat challenges that you’ve been dealing with.

Or you have a complete turnkey solution. Where you can spin up your critical applications and bring it online quickly. That is the way I look at resiliency in terms of how soon your business can come up. How soon can I get everybody back online, as opposed to looking at just the physical piece of hardware which gives you redundancy. My business has to be in operation.

I’ll give you a case in point, I’m not going to give the name, this particular chain in Canada, they got hit by a crypto locker. They shut down 70 locations all over Canada. They were losing a hundred thousand dollars in sales on a daily basis while they were doing their triaging, trying to figure out where it got hit and what got hit because they didn’t want to obviously pay the ransomware guys money. But what makes our solution a little different, listen, you continue doing your triaging out there, we can bring a complete unit up and running so you can continue functioning your entire storefront while you slice and dice your current infrastructure. Sales should not go down.

I think you guys had a similar thing during one of those unrests that happened in the US. This is a previous job you were at, where the grocery store got burnt down and they had set up a makeshift tent outside to get the entire point of sale up and running. Same exact thing. How soon can I get my business up and running? So that’s where we are trying to think out of the box.

Scott Mann

These conversations typically seem like it’s very complex to do all of this. There’s a lot of different moving pieces to doing it all. You come back to that same message of ‘turnkey simple’. How do you make the edge simple?

Rahul Narsimhan

There’s a webinar that I attended; it was a partner conference at Barracuda Networks back in 2009. One of the senior VPs out there did a presentation and his slide was a picture of the Apple remote, which had four buttons, and a Windows remote which had 25 buttons. That image has literally made permanent residence in my brain. I can do everything with that four-button remote with my eyes closed. Especially with my age, nearsighted, you have to take your glasses off, look at it, okay, am I pressing the right button? With those four buttons, I can do it eyes closed. Whereas with a fifty-button remote. You’re looking at it, “hey, did I get the Netflix button, or did I mute the TV or what exactly did I do?” That simplicity has stuck in my head. You can make the greatest product that’s there, but it’s complicated, your chances of adoption are zero, or even if it gets adopted because of your connections, it’s not going to stay there for there for too long.

The other challenge that you have is in terms of employee training as well. People move. Now you’ve got a new guy who has to be trained. It’s a complex product. He has to go for a whole certification cycle. Then he has to be playing around with the product for the next four or five months before he becomes completely familiar. That simplicity has to be the key. Any data center is complex, you’ve got switching, you’ve got firewall, you’ve got switching rules, VLANs, et cetera, that you have to set up. You have to set up your StorMagic cluster for example, whether it’s a tool or cluster, you need to make sure that data is being synchronized. It is a complicated product. How do you make it simple? When we put this data center together, we actually ended up pre-wiring everything. Everything on the back is pre-wired. It’s already pre-labeled, and we are only making it available through value-added resellers. It’s not available through online retailers. There is a certain amount of work that a value-added reseller has to do in terms of simplicity. “Okay, do you know what? I’ve got to set up the VMs, I’ve got to set up the VLAN, I’ve do the port assignment, configure the firewall”. Once you’ve done all that, you send it to the customer, he has that single cable to plug-in experience.

There is complexity, but we have simplified the complexity, making sure that every unit comes configured the exact same way. When you’ve got hundreds of these deployed, you know where exactly every port, every cable is going to be the same across every box. Every hardware is going to be the same across every unit deployment. That’s the challenge that we had to put together to make it simple.

Scott Mann

There’s a quote that always resonated with me and I think it’s Da Vinci. If I’m wrong nobody quote me on it, but it’s “simplicity is the ultimate sophistication”. In order to present something, there’s a lot of complexity under the hood that you have. Basically, that’s the value of the turn of a turnkey solution.

Rahul Narsimhan

Correct and you hide the complexity. It is complex, but you hide the complexity. Just give them a simple interface to manage. What is happening behind the scenes is completely different. That is what makes the adoption easier. Absolutely.

Scott Mann

Moving on. At the CEO level, how do you decide when to lean into an emerging model like edge, versus waiting for the market to mature? We’ve hit on that a little bit, and how emerging really is edge, but give me your take on that.

Rahul Narsimhan

I don’t think edge is an emerging model. I think edge has now come long way from emerging. If you look at VMware, technically it was emerging model, and it’s a stable model now. It’s been deployed in millions of sites. Same thing with all the other ATR solutions, edge is there. What is an emerging model is the edge AI part. I think edge AI is the wild card out here, where people are still coming to grips with it. What you’re going to do with it is the other challenge. We’ve got our own set of challenges there because hardware is an inherently risky business. You guys are purely software, you don’t really care about what hardware platform, as long as you can certify on as many hardware platforms. The flip side story is, I want to make sure that my hardware is certified with as many software platforms, but I also have to worry because the hardware keeps changing. If the hardware is changing nonstop, I have to have some sort of commitment from the partners, because it’s a commercially available product that we are making low cost. Vendors like Asus, et cetera, are committing that we are going to keep this form factor alive for the next eight years. That’s important because if you’re building something that’s there, the hardware form factor has to be there for a while. Otherwise, I’ll be left with a hunk of metal that’s essentially garbage.

So that part is important. AI itself is evolving. I think the GV10’s mini supercomputers only came out last year. There’s already talk about, AMD has got its own version that’s coming out, how all that stuff integrates into a model which can run your ATI workloads and AI workloads all in the same thing. Two years back when we designed the product, we were only looking at HCI workload. AI workload was not even on our radar. It was not there at all. It’s only in the last six months. We went GA with a product in March. We did our first roadshow with you guys in April. And between March and now, we’ve got AI hardware that’s being integrated into it. We are talking about something which is a small form factor, it also has to be power efficient green tech. You can’t build a 20-foot mobile trailer, which is technically a portable data center, and call it mobile. Within reason, what can you do with it? There is a genuine worry that technology is going to change, new hardware is going to come out. But that’s a risk I have to take because I can’t wait for another three years and say, “hey, when the market matures”, I don’t know what’s going to happen between now and three years.

Scott Mann

It’s about future proofing it. Edge AI, what everyone’s talking about, what they’re going to do. This hits pretty well with everything that’s happening around the world with the hardware crisis and pricing. There’s this balance and the question for any executive level is, how do you balance that ambition with financial discipline, in this space that’s still evolving? They don’t have all their edge AI requirements down completely yet and are being impacted by hardware prices, tariffs, everything. What is that balance? How do you find that balance in this world?

Rahul Narsimhan

Depending on who in my organization you speak to, you’ll get a different answer, because I’m terrible. There are some advantages. The money challenges come in when, for example, you’ve taken a fair amount of loan from the bank, or you’ve got investors that have invested money into the company. The financial challenge has to be there because you need to stay on message and you need to be able to build a product which is going to deliver immediate returns out of there. We are completely self-funded. We have got no investors, we have got no debts, we have no loans from anybody else. We have got a profitable side of the business, and we are diverting some of our extra resources into doing what we are doing right now. In a way I don’t have my back against the wall, so there is some flexibility.

But at the same time, we want to build a cool product. We also want to have fun doing the product. I’ll give you an example, actually, it’s right behind me. I’ve got the 3D dummies of my portable data center on the wall.

Scott Mann

I was wondering if it’s 3D printed or in the wall. That’s good.

Rahul Narsimhan

No, it’s not 3D printed. You’ll be actually surprised. These are actual cutoffs from the actual data center in the front. When we decided to put this together, the challenge that we were having is that we could not ship our portable data center, which weighs about a hundred pounds to, say, South America to Chile for a trade show and then bring it back. That alone was costing us $2,000 and you ship it and they bring it back. There was some logistical challenge. We said, “hey, you know what. Why don’t we make a cutout? It’s easier for us to carry this and show them the product which you can pack in a suitcase”. I said, I’m going to make 40 sets of these, which is 80 pieces front and back, and I’m going to leave behind with some of my partners. Would you believe it? How much does this end up costing me? $36,000 US dollars. When you look at this stupidity, you say, “hey, is Rahul fiscally responsible?” That is one example of no, not fiscally responsible.

Scott Mann

It just takes one good trade show pays it off.

Rahul Narsimhan

We know we are in for a long haul for the next eighteen months. We’re setting the roadblocks in place. When we went to Germany, we left our portable data center with your team there. We are going to be back there again. We want to be able to do more shows, so we are not hauling it back and forth. We did the same thing when we went to South America, to Chile. We did the same thing when we went down to the UK. We left some with a couple of partners out there, plus StorMagic has actually got our portable data center. We’ve got three or four in the US with different partners. We have got a trip coming up of South Africa. If I take the portable, I’m not bringing it back with me, it’s going to stay with a partner out there because we plan to be in each of these regions over the coming months and years. We’re putting all the building blocks in place to make sure that we are successful after 18 months.

I’ll give you another example of stupid stuff that I do, just to show how fiscally irresponsible I am. Much to the dismay of my programmers in the office. Our name is Hive Radar, and we had two stalls in Germany with the defense and the IoT section. Would you believe how many people came up to our booth and looked at the data center, even though it says, ‘portable data center’, looked up announcing, “how does this radar work?” They want to know if I was actually manufacturing a radar. Even though it says, ‘portable data center’. It’s just a matter of the show because it’s a defense show, the mindset technology, “hey, it’s a radar, what are we doing?” So, guess what? We’re building a small little device, which is actually a radar. It’s an open-source project, it’s just going to be included, no charge. There is a Wi-Fi radar which shows all the wireless devices hovering around near your area. The other one is another radar which shows all the planes flying over your head using ADSB data. You say, “hey, you know what? We’ll actually have some fun and put it together and just give it away.”

Scott Mann

I like it. Just like a little play on the name and just feed into the misconceptions. I mean, that actually segues perfectly into my next question. What’s one misconception about edge infrastructure today?

Rahul Narsimhan

People think edge is complicated. Or only in places which are deployed by big guys, whether it’s a mining agency or whoever else. I think edge is a little more prevalent. We’ve got an example of a project through our reseller in Africa where they’re looking at schools, five hundred schools across Africa where they’ve got not enough power, they depend on sunlight, they want a full education system that if it requires to be run off grid, you can run that off grid. We are trying to make it more affordable. It’s going to be there, required by everybody, whether it’s convenience stores or retail chains or whoever. It’s not just going to be mainstream stuff for the big guys or defense. I think it’s available for everybody. That’s the biggest change I see.

Scott Mann

I agree. Edge is seen as the far remote edge on a vessel in the middle of the sea or something like that. But edge is just anything outside of the public cloud. There’s different nuances to every need within the edge.

Rahul Narsimhan

Exactly. And how do you run it? It’s not one way or the other way. I think a hybrid is probably a more resilient way where you’ve got stuff in the cloud, which is required, you’ve got stuff at the edge, that same resiliency comes into play. My business has to stay uptime, whether I’ve got part in the cloud, within means, within reasons. That’s the biggest thing.

Scott Mann

I got one more question for you. This is my typical crystal ball question. This is how we wrap up every one of these podcasts and it’s always good to catch you off guard. You have a different perspective on the market, even edge. I talk about the edge quite a bit and you’ve got a very different perspective on it for me and your solution fits into areas that I might not have thought of. “Hey, this is a great fit for us as well”.

If we have this conversation again in three years, what will be fundamentally different about how infrastructure is deployed and managed? In three years versus now?

Rahul Narsimhan

I think edge is going to be more prevalent. But at the same time, I think your original starting question comes in comes into play, the inflection point. I think the inflection point is the AI part. The AI is the wild card out here. We do not know how it’s evolving, how it is going to impact everybody’s lives over the next three years. What changes are going to come into place? I have a feeling that everybody’s going to be deploying edge. They’ll have some component of AI which is going to be deployed on top of that, whether it’s for local inferencing, for programming, data sovereignty. All that is going to come into play, but I think it is going to be taken for granted that there has to be an instance of edge that’s running. Whether you’ve got a cloud instance or not. But I think edge is going to be more mainstream. Absolutely.

Scott Mann

Yeah, it’ll be more ubiquitous. You have to manage data any place, anywhere. Maybe that’s not three years, maybe it is three years.

Rahul Narsimhan

I think it’s already there. You look at, depending on who you read, the industry reports which came out one year back or two years back, which says “hey, the edge market is going to go from 30 billion to 300-400 billion dollars in by 2030”. But I have a feeling those numbers could be way off, because at that time they were not looking at AI as a component or AI as a driving factor. That includes your hardware, software, AI orchestration, everything that’s there. You guys are doing the HCI layer. You’re looking at how can I run AI workloads on the HCI platform. I’m looking at, I don’t want to be singled out as a hardware vendor. We’ve got the hardware part covered. We’ve got the HCI part covered, whether it’s StorMagic as an instance which is there or anybody else, but we are also working with other companies out of there. How can I do AI orchestration on top of this? How can I do inferencing on top of it? I’ve got research teams, biohazard activity that happens all over the world. How can I make sure that I’ve got something that is set up in a completely remote, isolated place where I can do air inferencing or RNA analysis completely off grid? Also if I have connectivity, I’ve got connectivity back to send all my data back to the central HQ or lab.

It is going to become mainstream, and you look at stuff happening in the warfront which is there Ukraine or what happened in the Middle East, your central cloud, Amazon got hit by the one of the missiles. When you’ve got a whole cloud, your operation goes down, a regional site goes down, what do you do? I think it is going to be more of a mainstream product going forward.

Scott Mann

I see what you’re saying, you guys are more than hardware. You’re bringing together all these solutions from the software layer down to the hardware.

Rahul Narsimhan

We are adding some initial software stack. People talk about digital twins, where you’ve got sensors, et cetera, and engines, and you’re getting real-time metrics, which gives you predictive AI or generative AI saying, “hey, based on the past historical outcome, et cetera, I expect the failure to happen in the next one week”. What we are doing as in AI, we are actually miniaturizing some of these models and actually putting sensors on top of it and generating it.

What we have done also, with our portable data center, we’ve got a physical box that’s there. We’ve got a digital twin of that inside our monitoring solution. Inside we’ve got a single board computer inside a portable data center, which has got somewhat of a digital twin version of a full portable data center, which shows you my complete edge data center that’s running, the fan speeds, how much CPU, how much temperature based on this, and predictive information, if a VM is over provisioned. We’re generating all that data to be able to let customers and let you guys make informed decisions. “Hey, this is what’s happening. Okay, we need to optimize it.”

Scott Mann

I love where you guys are going with it. I think you’re going to be very successful. I’ve already started to see some of the success. There was a conversation yesterday with my team and I was like, conveniently I’ve got a podcast recording tomorrow with Rahul, so I’ll let them know after the podcast. It’s very neat to see what you guys are doing and I love to see how you’re bringing all these different technology vendors together and the tagline ‘Hive Radar more than just a portable data center’, it’s forward thinking.

Rahul Narsimhan

No, we make the real radar. We have a radar.

Scott Mann

Rahul, thank you so much. Appreciate it. It’s always good to connect, and next time I’m in Vancouver, I’ll hit you up.

Rahul Narsimhan

Absolutely. Thanks, Scott. Nice talking to you.

Scott Mann

For everybody, thank you all for tuning into PodMagic. where simple, reliable. IT meets real world impact. If you enjoyed the conversation, be sure to subscribe and share. And we’ll see you all next time.