The EDGECELSIOR Show: Stories and Strategies for Scaling Edge Compute

From Rock Band to Energy Rockstar: Decentralizing Data Centers and Innovating Heat Transfer with Mark Bjornsgaard of Deep Green

December 11, 2023 Pete Bernard Season 1 Episode 11
The EDGECELSIOR Show: Stories and Strategies for Scaling Edge Compute
From Rock Band to Energy Rockstar: Decentralizing Data Centers and Innovating Heat Transfer with Mark Bjornsgaard of Deep Green
Show Notes Transcript Chapter Markers

Can you imagine heating a public swimming pool with a data center? Well, that's exactly what Mark Bjornsgaard and his team at Deep Green are doing. Join me as I chat with Mark about his incredible journey from rocking in a band to becoming an energy rockstar, and how his exploration in venture capital led him to pioneer a future in energy and climate impact.

Data centers, notorious for being power-hungry, are facing a paradigm shift. Mark and his team are at the forefront of this change, decentralizing data centers to optimize grid usage. We talk about new technologies that leverage Kubernetes orchestration of workloads across data centers that not only maximize utilization but are also designed to harvest the heat generated for local housing, swimming pools, and more.

Through this distributed data center and energy harvesting strategy, Deep Green can heat thousands of flats in the UK. We discuss the crucial role of carbon regulations for the power industry. Mark shares his fascinating story of scaling from a small heater on the side of a hot water tank to a data "furnace" for cloud providers.

Join us to learn how Mark and his team built a playbook for rapid, efficient project implementation, and how these trailblazers are leading the way for a better, more sustainable energy future.

Listen to Mark's band at - https://open.spotify.com/artist/49sAZLRBJgtum6oW6pjrrT?si=VshqwJZdT82F0BvcM8a3Yg !!

Want to scale your edge compute business and learn more? Subscribe here and visit us at https://edgecelsior.com.

Pete Bernard:

When you ask people what Edge Compute is, you get a range of answers Cloud Compute in DevOps, with devices and sensors, the semiconductors outside the data center, including connectivity, ai and a security strategy. It's a stew of technologies that's powering our vehicles, our buildings, our factories and more. It's also filled with fascinating people that are passionate about their tech, their story and their world. I'm your host, pete Bernard, and the Edge Celsius show makes sense of what Edge Compute is, who's doing it and how it can transform your business and you. So let's get started. I think these days people pick their people kind of get into their ecosystem and all their stuff kind of works, and then it's like don't touch it, exactly exactly.

Mark Bjornsgaard:

And then when you have to, having upgraded computers such a hassle, isn't it and then you, yeah, having gone through the hassle and the bloody thing doesn't work, is really more annoying.

Pete Bernard:

I know, yeah, you said this a lot of times. Well, I'm glad to connect. Yeah, I know it's afternoon there and it's morning here, so that's good, and it's about. I'm here in Bellevue, washington. Where are you where?

Mark Bjornsgaard:

I'm dialing in from about 25 miles south of London, so in a town called Guilford, so just south of London, oh yeah.

Pete Bernard:

Guilford. Yeah, I remember that. I worked for a company, Phoenix Technologies, that was a BIOS company, many years ago. I think they had an office in Guilford.

Mark Bjornsgaard:

They might do. There's a lot of pharma, there's a lot of pharma around Guilford and there's a lot of sort of AI and machine learning around kind of cruelly Gatwick area. So yeah, it's-.

Pete Bernard:

One of my bad habits is I forget to introduce people and just start talking. So let me introduce you first and give you a chance to give us a little context. So, mark Bjornsgard from Deep Green, did I get that right? Not bad, not bad.

Mark Bjornsgaard:

I've had a lot worse.

Pete Bernard:

Okay, I'm trying, I'm trying, and you're calling from the UK, of course. Why don't you give us kind of a few minutes of context of who you are and all the good stuff?

Mark Bjornsgaard:

We'll go from there.

Pete Bernard:

I mean.

Mark Bjornsgaard:

I'll keep the sort of history brief, but up until about a year ago I was running a venture capital firm that I built and had a number of exits. We sold a healthcare business in the US called GSK and we had a bunch of other stuff. But we're really energy folk at heart and about seven years ago we were working for a large energy company in the UK called British Gas and we built a very, very crude prototype of what everyone now understands as a data furnace, as in the ability to capture computers. Fast forward. Three years ago that business grew and became a business called Heater. But I felt there was a commercial opportunity so I actually bought us out of that.

Mark Bjornsgaard:

And then about a year ago we invested quite heavily in our first EDGE data center with Heat Recapture. So in March this year we announced that we were heating a public swimming pool in the UK about 60% of it and we were giving the heat away for free. So we're giving that public swimming pool about £22,000 a year and we've literally built a supercomputer in the plant room of the public swimming pool. And this was a, for instance, a proof of concept to say, to prove out the economics, prove out the operational and industrial capability of what we were doing. And then suddenly we were famous across the world. Suddenly we were trending number two on Twitter. So for four hours I was on national TV news. So my face for radio. Suddenly I go to data center conferences and now we kind of get high-fived as the pool guy.

Speaker 3:

And seven years ago you couldn't get arrested doing this right.

Mark Bjornsgaard:

Seven years ago you had Heat Recapture or Energy Official no chance. But of course the world's changed. Edge computing is a reality, the robots are taking over. Energy is expensive and pricing is crazy in Europe. So here we go.

Pete Bernard:

Cool, I do want to dig into that. Let me ask have you always been involved in sort of the energy space? I mean even, what was your journey to get to be this energy rock star that you were?

Mark Bjornsgaard:

talking about. Yes, ironically, rock star is the right word because I did have a misspent youth playing in. So I started off my career as in the music business. I'm a history and politics student. I'm a troublemaker by nature but we are quite anchored in energy. So I've invested in large industrial heat stores, so these massive thermal stores where you basically heat aluminium and rock to like 400 degrees in a perfect vacuum. That business is scaling. We sold an EV subscription business to BCA, british Car auctions, a couple of years ago. So we're energy folk and we sort of learnt our center craft.

Pete Bernard:

Wow, I see, I see, but did you play actually in a band?

Mark Bjornsgaard:

Oh yeah, there's proof a bit out there. There's us supporting a band called Mochieber in Germany in front of a I know there's a lot of folks in tech that are in music. I noticed a guitar. I couldn't help noticing.

Pete Bernard:

Yeah, and almost everybody I have on the podcast has some background in music.

Mark Bjornsgaard:

I know it's strange. I think we're naturally sort of curious types, maybe.

Pete Bernard:

Maybe here's the math or something I don't know what it is, or the structure or the lack of structure, maybe, but no, that's cool. I think it's great, actually, just as a sidebar, that to exercise different parts of your brain. So for me, like tech, tech, tech, and then on the weekends I'll do something totally analog or whatever, and just you got to kind of balance it 100%.

Mark Bjornsgaard:

Yeah, you definitely. You live a dull and less creative life if you, as you say, if you don't give yourself space deal. We recorded an album back about 20 years ago and it spent 20 years in my loft and then eventually, about two years ago, my wife said I've had it on the to-do to mix the album, and we did it with a really big time producer I called Ian Grimble. So about two years ago we finished the album, so we finally got something on Spotify. Yeah, I agree. I agree, it's everyone. You might think of it as a distraction or you might think of it as a time sink, but actually it's not true at all. Of course it may.

Pete Bernard:

No, yeah, no, it's important. It's just like physical exercise. Creative exercise, you know, is good balance. So anyway, that's the PSA for today. But, anyway. So let's get back to the nuts and bolts of this thing. I mentioned this to someone. I had lunch with someone from Nvidia on Friday. I said, oh, I'm going to talk to this person about, you know, taking the heat from data centers and heating pools and stuff, and she was like what? How does that?

Pete Bernard:

work. So now, of course, I hadn't really dug into it with you yet, so, but I think the actually take a step back before we get into the how it works. I think people will be fascinated about that. I did, I published a and I think this is probably how we got connected. I published a paper, an edge note, as I call it, and some TikTok videos about what I called AI's environmental armageddon, sort of the.

Pete Bernard:

You know, sort of this insatiable desire for AI workloads and high performance computing, coupled with kind of a lack, basically, of environmental resources to power those workloads and all those cat poems and other things that we're trying to do, and so that's kind of a. That's just an inevitable collision that we're having. And you know, I actually published an episode today with Qualcomm talking about low power silicon. Yeah, and so when you step back from it, I think in the video I talked about, there's a number of you know probably five or six or seven different things that need to happen to solve the problem low power silicon better. You know transmission infrastructure, you know the ability to measure, actually measure, you know the power that's used for workloads, but you know the other part, too is just fundamentally in terms of constructing and, and you know, deploying high performance computing or edge computing. That's doing AI workloads, just doing that in a different way.

Pete Bernard:

And I think that's where what you guys are doing is pretty interesting, because it's really take almost flipping the whole model on its head. You can play the other way, you know, as opposed to like. So for folks that maybe don't know, and you've been to Azure data centers, they're like these giant. You know 10 football fields of concrete out in mill and nowhere, with, just you know, 1000 miles of cable and blah, blah, blah every superlative you can think of. And it's cool, you know, but it takes, you know you have to run a ton of power to it, many megawatts and 100 megawatts or whatever, and then you have to, you know, cool the thing, because you're running these fantastic Nvidia chips, as you were talking about, at very high temperatures, and so it's a you know it's a thing, and then that was all well and good until we realized that there's just not enough power and water to actually do this, and so maybe you can kind of dig into where you're seeing this from, and because I know you're also from the UK perspective, it'd be good to hear.

Pete Bernard:

I mean I'm, you know, in the US. Here we can talk about data centers and big things in Texas and Arizona and stuff, but you know there's no Texas in Arizona, in the UK.

Mark Bjornsgaard:

So no, no, no, no. You basically hit all the high notes of of the challenge and what's and, as you say, we, we've been planning this for seven years and we are looking at completely the other end of the telescope, and when you do that, all of these perceived challenges kind of become opportunities. Now, of course, you'd expect me to say that, but but if you take, just take simple things like like low power chips, well, if we can catch 95% of the heat, which we're doing in Xmouth, do we really need low power chips? Actually, if, if, actually, if we use, actually, what we're doing is using the electrons twice. So we're, you know, we, we use the electron for the compute and then we're using the electron to heat the pool, actually, that the hotter chips would be better. Well, exactly, we actually, we actually like inefficient chips.

Mark Bjornsgaard:

And then you take power and the grid our grid is, I think, probably as constrained as yours. When you decentralize, I mean, if you run a hundred megawatt data center, you need 220 acres of solar, right, it's never going to happen. So when you do it like this, and you and you, you, you actually break up the data center and you actually then realize that you've got plenty of solar on site car parks, solar roofs, whatever to then power the data center. So then again you take with batteries and solar, you then take, you pretty much take the data center off grid. So and then and then, and then.

Mark Bjornsgaard:

The really interesting thing is that since our moment of fame in the spotlight, we've had some like 7,000 megawatts of committed capacity sign up. Right, but that's all with wiring around, we don't have to do anything to the grid. There's no, we'd have to upgrade a substation, you have to lay any more cable, it's all there. So so yeah, if you look at it from the other end of the telescope, you saw a suddenly thing. Actually, this is, this is the solve. Essentially because you start to say well, actually you know what In the Europe, heat pumps aren't really, don't really make much sense at the moment. You know electricity is still more expensive than it has. But if you use the electrons twice, it turns out we can make heat pumps super efficient, and so. So that's, that's where we've got anchored. You kind of take all of the perceived challenges, of what people are fretting about, as challenges, and then when you, when you say, actually the opportunity is everywhere that you can reuse heat. You actually actually there's. There's hundreds of gigawatts of available capacity. Yeah, just in the right place.

Pete Bernard:

I guess the challenge though the inverse of that, just to put the other hat on is you're deconstructing the data center into smaller chunks, and you know there's an argument that if you have all of these disparate pieces out there and they're not all fully utilized, then you know you've got a room full of each 100s that maybe aren't fully used in one area that should be used in another. I mean. So how do you think about that?

Mark Bjornsgaard:

Yeah, you're absolutely right, shuffling the low workload is very much part of the plan. But actually when you, we know that we've had the virtualization, we've now got Kubernetes and now we're going to WebAssembly, so actually we know that the software layers are actually going to do quite a lot of the heavy lift in the next five to 10 years that are going to make more assets, much more. You know all of these, all of its. You know Kubernetes and WebAssembly. It's like it's compostable, right.

Mark Bjornsgaard:

So actually what we assume will happen and I think we're really seeing it to some degree is that actually your asset utilization becomes much better because you can stop duffel workloads around, and so that's definitely part of the plan, which is and then, what the great thing is, you don't have 10,000 H100s stranded in some data center where you can't actually use the heat you can actually start to and we know Nvidia really well and I think they're particularly excited about what we're doing in terms of inference. You know smaller instances of AR. So, you're absolutely right, it is the challenge now, but I think if we're in front of the puck, we kind of can see actually the better way. Actually, decentralizing is actually again a better thing to do, because you can actually just sort of shuffle loads around.

Pete Bernard:

Yeah, no, it does get a lot more agility and put the capacity kind of where it's needed as well and modularize it definitely. I mean I think you know so that I think that's one of the discussions is, where is it heading? I think now that you have a much better orchestration layer, like you said, you can sort of load balance these workloads and plop them down across this kind of more of a virtualized data center than a centralized one.

Pete Bernard:

The question was around inferencing versus training. So you know, right now there's a tremendous amount of demand for training, training, training, training and training's hard right and it requires lots of big chips and lots of compute. And that's kind of the essence of what has unlocked AI in the past few years is the ability to do this kind of massive training. And then there's the inferencing, and so you know there's some folks are like well, the big data centers will do the training and then the kind of the edge data centers can do more inferencing. But are you doing any like actually doing AI training in any of these?

Mark Bjornsgaard:

Yeah, xmuth is 56A100 cars, so there's a couple of there's about a million worth of Nvidia Kitty and XMuth Leisure Center and that is doing exactly that. Yeah, it's being used by Civo, one of our brilliant partners called Civo, another brilliant partner called Alsys Flight, who are both kind of high performance compute specialists. So, yeah, absolutely. And now our next wave of sites will be doing a 14 megawatt district heating system. So that's kind of training, right, you can land a pretty meaty kit in a 14 megawatt tent all the way through to the next wave of pools, which are kind of the smallest, of about 200 kilowatts. So, yeah, that's where our kind of estate we imagine will be used.

Mark Bjornsgaard:

And the reality is, most normal day don't want this workload, right, it sucks up all your power and cooling it's a nightmare for most of them. So actually we're seeing, I mean crazy. So the first, the day we became famous, a very, very large infrastructure fund that in Europe rang us up and said we want to, we effectively want to kind of back the business until exit, and that's the deal we've been negotiating. So come the beginning of December we're going to be announcing backing for this so we can roll out at scale and then we can provide. And the reality is, once you land solar and batteries behind the grid, behind the meter, the costs are just so much cheaper than a conventional data center. Yeah, yeah, definitely it. So, yeah, it's.

Mark Bjornsgaard:

But the interesting thing about what we're doing as well is that it won't necessarily all be high performance compute. So, because it's edge computing, right If, because you're plugged into the edge, actually we're starting now to see true edge use cases emerge. So content networks, gaming companies, all the people who, yeah, lower latency type Proper use cases, and I've always been fascinated by this kind of idea that you actually can build an edge network without actually having any edge, any edge net use cases. You sort of build it and then at least edge has got the kind of it's, got the infrastructure to start to. You know, do what we think it's going to do.

Pete Bernard:

Yeah, no, that's interesting. Yeah, I mean, there's a kind of there's a whole discussion around latency these days where it's like latency is one of those attributes where, if you, if the latency is low enough than any incremental lowering of the latency doesn't really have any value. So you know, I've talked to a lot of companies where it's like, well, you know, instead of 50 milliseconds for five milliseconds, like well, but you know, 50 milliseconds is plenty, you know, like then you know, it does.

Mark Bjornsgaard:

It depends on the workload. It certainly does. Workload I mean.

Pete Bernard:

Yeah, CDNs is a good example where you definitely want stuff close to the action.

Mark Bjornsgaard:

Well, yeah, and I think you know 90% of the people who've gotten you know who we're partnering with now are anchored on the ESG bit. You know the carbon bit Because in Europe certainly we've got there's a ton of legislation now coming for over the next kind of year or months where you know you, just you know, if you're not reusing the heat, you're not doing, you're not actually taking care of the carbon in this equation. You know, yesterday it was announced in the UK that the prioritization of whether you get a power connection is now going to be based on a number of factors, including whether you are a sustainable planability based into your proposition. So you know a stick. There was a really nice phrase somebody used on a panel I was doing earlier which was like the carrots and the sticks are just going to get bigger and bigger.

Pete Bernard:

Basically, yeah, no, that makes sense. And you know, I mean, I think in tech sometimes we lean toward, you know, the market forces to solve everything, which you know usually doesn't work in the long run. But I think in this case you definitely need some guardrails for the better public good to sort of make sure people are incentivized to do the right thing, even though I'm sure they would do it anyway in their heart. But it never hurts to have a little bit of, like you said, carrots and sticks that are large carrots and sticks.

Mark Bjornsgaard:

Well, Germany, Germany, to do that, yeah, Germany, by 2025, you won't be allowed to build a data center without starting to reuse the heat. So I mean, this is becoming very real. Csrd, which is legislation, again you know, yeah, it will. As of May this year, if you run over 100 kilowatt data center, you've got to report your carbon. So, yeah, over the next three to five years, certainly in Europe and I don't know as much about the US, I'm sure you have a better view, but certainly in Europe, this, yeah, it's going to become Well, certainly the reporting of the usage.

Pete Bernard:

I mean that's something that's missing and it's one of the points is we don't instrument very well to understand power consumption, water consumption. You know it's all very obfuscated and I think that would be huge, like if people went to chat GPT and saw for each query. You know it's like in the US we had this thing called energy star, which is like when you buy a fridge it's like oh this is like $4 a month, whatever.

Pete Bernard:

If you had some sort of energy star for AI where it's like, well, this cat poem is going to cost you like three trees or whatever, then I think you know companies and other folks would be able to measure, like, how much you know environmental resources, am I really consuming, based on the AI workloads that I need to run. And that's where you can start to say, well, maybe I'll run some of those locally or run them on my PC, or you know what's the eventually the? You know what's the eventually the. We'll get you talking about orchestration. I think we'll get to the point where the energy consumption will be a factor, just like latency will be a factor in the workload, and where it's run and people will like let me run this workload as green as possible. Yeah, maybe that's running it at a certain time of the day or on a certain platform. Yeah, and it'll be optimized for power consumption eventually. But we have to measure for Google.

Mark Bjornsgaard:

I think I'm right saying Google Cloud already does that, so it's already got and it's already taking half hour data Of, because we'll know that when, even when somebody says I'm buying renewable energy, it's not always green right, it's only green when right. So so I think, yeah, that is very much on the cards in the green software movement in a number of different areas with business. That is that thinking. Again, what I know in Europe is is pretty well advanced and yeah, he's.

Mark Bjornsgaard:

And look, reality, if you set a net zero target and you know this right, you know you're what would you do? You know your cart, you the cart, the embodied carbon, the scope, three carbon in your, in your data center is you know, yeah, eventually the stick becomes so big you have to. I mean, we won one, one client recently that we sort of put a proposal together for they were spending something 450,000 pounds a year over two data centers and and we've, with very mind, we're we're offsetting about 75 dollars a tonne, all of the scope, all of the carbon in the computer Cell. So we've taken their bill down by half. But also, that's decarbonized right, that their net zero target is now Right, that much easier because it isn't creating the problem.

Pete Bernard:

Yeah, right, right, right. What's the big deficit? So how big are these kind of deconstructed edge data centers? I mean, give us an idea if I were to see one on the street. Yeah, is this like a shipping container or is this a?

Mark Bjornsgaard:

Smaller than that. Yeah, I mean, if it's, if it's a 200 kilowatt Site for a swimming pool, a normal kind of what you think of as an average kind of public swimming pool, you know, 25 minutes people, but it's a container, basically a container on site, but but it's very dense. I mean, obviously you, you know, you know we're getting, you know we're getting like a hundred or more kilowatts of rack. You know, in terms of, in terms of the density of the and I'm at the edge of my technical abilities here but certainly in terms of the density you're getting from from immersion cooling, you're, you're, you're way way better. So you're squeezing a lot of compute into a very much smaller space.

Pete Bernard:

Right, right, yeah, so you. So it's like a shipping container ish for a swimming pool and then for these the larger ones, I mean that's 200 kilowatts is probably not enough to do no, then yeah so if you want to do like multi-megawatt, one is like it's a lot of shipping, say so yeah, bark.

Mark Bjornsgaard:

And the next, the 14 megawatt site. That starts as a six megawatt site.

Pete Bernard:

That's like 10, 15 shipping containers, you know, wrapped and stacked with a, with a frame around them but, but and yeah, and those would be built like near Housing, where the heat would be used more for housing. Yeah, that's, that's a district heat.

Mark Bjornsgaard:

Yeah so exactly that they have that. So this particular development is 10,000 flats. They've got a, they've got a temporary energy center. At the moment they're building a permanent energy and that is, as you say, massive heat loop. It's got Europe's largest water source heat pump into the Thames, so it's dragging heat and Thames out the water and we're priming that heat pump. Yeah, but but again, because it's free heat. We, you know people welcome rain, they're welcome with open arms, right, it's just free and the transfer and the transfer mechanism is water.

Pete Bernard:

So you're basically doing water, heating the water and then the hot water is unused to no oil.

Mark Bjornsgaard:

That's how you're oil, yeah, yeah. So the computers, for in this instance we can do direct to chip and and we can actually do a, a cruise. But in that particular instance it's going to be high-performance compute fully immersed in inert oil. The oil you don't even have to pump it round. A convection current starts and then you and that convection current just feeds it through a heat exchanger. So it's, yeah, it's very, very simple. I mean, one of the reasons we did Xmouth in the way we did it is because we wanted to show how simple it was. It's, it's right now, without doing a disservice to the amazing work that the technical teams have done. It's just a tub of oil with computers in it and heat exchange.

Pete Bernard:

Wonderful.

Mark Bjornsgaard:

Yeah.

Pete Bernard:

Yeah, no, it's so. But I guess the other thing to To sort of kind of buy into this whole new process, like you said, it's looking at the other end of the telescope, it's really taking a different strategy toward you know kind of data center buildouts. So for the hyperscalers they would have, they have to really sort of buy into this deconstruction of the data center. You know, and in order for this to really work because you can't transfer heat from a data center in the middle of Arizona, it's too far and it doesn't make any sense it has to be sort of like you're saying, just like the workloads need to be where the action is, the heat needs to be where the action is for it to be transferable.

Mark Bjornsgaard:

One high-scaled in particular is very excited about what we're doing. So we're having really good conversations there, Because I think most of the industry is anchored around the fact that at some point, you know, up to 70% of the workloads are gonna be at the edge of the network, and so if that's an inevitability at some point in the near future, then actually this whole kind of it's much less controversial, right? It's just okay, we'll just do it there now, yeah yeah, interesting.

Pete Bernard:

Is anyone doing this in the US yet, or is this still a European?

Mark Bjornsgaard:

No well, they may well be and I might be doing somebody a terrible disservice, not that we know of but of course your energy market's different. The pricing of energy is different. You know, you've got a lot more space. Menlo Park yeah, there are a few sites that sort of put their hands up when we first sort of spoke or we first announced what we were doing. So we hope to be all-sided sooner rather than later. But I think Europe is probably leading the charge just because of the nature of our energy market. So there's a-.

Pete Bernard:

Yeah, no, and I'm sure regionally, I mean around the world, there's lots of energy markets like that and they, you know, not necessarily the US, but so there are other places outside of the UK, which in Germany is obviously some stringent-.

Mark Bjornsgaard:

Yeah, there's a brilliant company called Cloud and Heat in Germany doing it really well Slightly different model, I think. They're charging for the heat, quite technical orientated. And then there's another great company called Carnot in France doing it as well. So it's, and they've just started to raise larger sums of money. So they're kind of you can see the momentum. Well, it was Microsoft's white paper In 2010,. Microsoft wrote the white paper on this. They wrote, they coined the phrase data. So Microsoft actually set the hair running, if you like. Yeah yeah.

Pete Bernard:

Right, right, right. Oh good, that's a good UK phrase. Set the hair running.

Mark Bjornsgaard:

Yeah, there you go. There. It gets the dogs in the east end of London. Got the hair running, yes.

Pete Bernard:

There you go. Sometimes I use this phrase here, and so I used to work with a gentleman from the UK and he always used to say horses for courses, yeah right right. You know like you have to have the right things right, and no one would ever know what I was talking about when I said horses for courses?

Mark Bjornsgaard:

Maybe it's not.

Pete Bernard:

You know what I'm talking about. But exactly Now, wow, that's fascinating. So it sounds like you're kind of at the center of center of some new, pretty exciting stuff, I mean. So how big is the organization you're working with? It's just a few of you in a garage here, or is this like a big one?

Mark Bjornsgaard:

Well, there's 10 of us now. Once we raise. Once we raise, it's going to be a lot more of us very soon. So one of the challenges here is scaling it fast, yeah, but that's why we. That's why we a number of people wanted to back us and we've plumped with a particular type of fund with kind of enough recognition in the market that people will understand that that's signalling a kind of a proper sea change really.

Pete Bernard:

Yeah, yeah, let's see what else I feel like we've. I feel like I now much better understand what's going on.

Mark Bjornsgaard:

Yeah, well, I mean, I'm sorry I was late as well. Maybe one. Maybe we can come back on and we can do a kind of walkthrough of next wave of sites and we can kind of do a bit more of a deep dive, that'd be fun.

Pete Bernard:

Yeah, no, I think it'd be great to kind of check in on progress on this stuff, and especially in different geographies, as it's coming up to speed, I'm sure, like with you. So the site you mentioned before, the one with the housing, so is that up and running right now?

Mark Bjornsgaard:

That will be that, and one pool in Manchester and one pool, another pool in London will be the next wave of our site. So that will take us to about six or seven megawatts, and then the idea is that we'll scale very, very fast from that. So we plan to be doing three or four sites a month, and so we'll, as I say, we've got thousands of sites to go at and yeah, it becomes a scale.

Pete Bernard:

Well, these projects so yeah, I can imagine these construction projects mean they take a long time. So I mean there's a lot of you know. Oh, go ahead.

Mark Bjornsgaard:

No, I was just gonna say, interestingly, we installed Xmouth in three days, so it doesn't take. Actually, the interesting thing about the whole thing is that because you don't have any dependency on power there's there's more fiber in the ground in the UK than there is in the whole of North America and People welcome you on site we can actually spin up a site in less than three months. The only the biggest constraint is a network actually. So, um, yeah, we can build really fast. Yeah, very yeah.

Pete Bernard:

That's really interesting distinction between the normal center industry and it's because also the they, the solution is modular. I mean, are they literally shipping container boxes?

Mark Bjornsgaard:

Yeah, Go to all standing I was good.

Pete Bernard:

Yeah, I was gonna say, like in the software world we're kind of spoiled because we can build things quote-unquote, you know, recompile them, ship them and update them and build them every day. But in the real world, where people have to put shovels and grounds and stuff like that, yeah Can't really recompile that stuff. So it takes a lot of planning to To sort of do things in the real world with with real equipment. But I think modularizing it, like you said, makes a lot of sense because then once you identify the site, you've got the power, you've got the connectivity. Then it's just a matter of yeah, yeah, snapping them together.

Mark Bjornsgaard:

Well, we pretty much what we spent seven years doing. That was that's the challenge. It's it's making sure, right, just crank the handle on them really. So all contracting or paperwork or legal, all that, all the spec is all fun, you're right. Oh, it's okay, and you just can then apply a playbook into science and be super fast.

Pete Bernard:

Cool. Well, I look forward to seeing more news about more of these sites coming live and Lots of swimming pools I guess there's a lot of swimming pools in the UK and indoor swimming.

Mark Bjornsgaard:

Yeah, well, I'm one half thousand and we've lost 300 already, so we've half of them are at risk. So yeah, we have. So so then, couple of thousand pools that are that are. Yeah, I mean, you know energy crisis is, yeah, it's really impacted, so really, really important.

Pete Bernard:

You know it's interesting. I mean there's a total side note. So I'm in a little Homeowners Association here. We have like 40 families, we have a, we have a community swimming pool, right, Just bought new heaters last year for like 15,000 bucks or whatever. You know, someday, I mean, maybe these things get so modular and small that it's like instead of buying heaters we buy a little data center pod you will run some workloads on behalf of someone and heater.

Mark Bjornsgaard:

Well, you will already I have one fitted to my house and the business started. The business started with this heater, which was we can hang it on the side of a hot water tank. So it's a tiny little day hangs on the side of a hot water tank and heats that whole to hot water tank for free. That's where the business actually started. So, um, yeah, that that reality is already already here.

Pete Bernard:

Yeah, yeah, huh, interesting, but you're not commercializing like home data center heater.

Mark Bjornsgaard:

He's off. Yeah yeah, go heater HEATA dot-co. If you go to that website, you'll, you'll find, you'll find that already in play. Yeah, yeah, that's where, that's that's where.

Pete Bernard:

Wow, this is good. This is a lot. There's a lot of called action here for all the listeners to Dig in more, but that's cool, including myself. I think I might go check it out.

Mark Bjornsgaard:

Yeah, well, I'm happy to come back on and speak more about it. Yeah.

Pete Bernard:

Yeah, definitely, definitely. Would like to have you back on and check in on progress, but this has been great, super educational and you know we're talking to lots of different folks in this space, like when we started. There's lots of different ways to help help with solve this problem. This is obviously a really fascinating one that seems to be really picking up steam, and We'll definitely be on the lookout for it. Any kind of closing words of wisdom?

Mark Bjornsgaard:

No wisdom, but no, thank you very much. I had me, it was pleasure to chat. So yeah, we'll and yeah we look forward to, because he say kind of announcing oh, the announcement is now quite a thick and fast.

Pete Bernard:

Yeah, fantastic, all right, so deep green, and we'll keep their eye on that and begin to some of the stuff that we talked about. Cool. Thanks so much. All right, thank you so much, mark Cheers, mark cheers. Bye, bye, bye, bye, bye, bye, bye, bye, bye, bye, bye, bye.

Speaker 3:

Bye, bye, bye, bye, bye, bye, bye, bye, bye, bye, bye, bye, bye, bye.

Pete Bernard:

Bye, bye. Thanks for joining us today on the Edge Sells your Show. Please subscribe and stay tuned for more and check us out online about how you can scale your Edge Compute business. Thanks for joining us today. Thanks for joining us today.

Exploring Edge Compute and Heat Recapture
Being an Energy Rock Star Journey
Decentralized Data Centers and Reusing Heat
Inferencing vs Training in Data Centers
Heat Transfer Advancements for Data Centers
Scaling a Modular Data Center Organization