We're going to get started with our next presentation of the afternoon. I would like to introduce Doug Recker, CEO of Duos Technologies. Welcome.
Hello, everyone. Thank you for coming in and listening to our little pitch today. We appreciate it. What we're going to start out, we want to tell you a little bit about who Duos is. These slides will obviously be on an 8-K today. We have some very important information we're going to share with you. We just got the word, legal approved it, so we're going to announce it. You're the first to hear it. It'll go out to the market tonight. We're extremely proud of it. It's a huge announcement for us. But let me go through this boring stuff first, and then I'll tell you about the exciting stuff. What we do is basically, we build and operate data centers. In the very beginning, Duos Technologies, I'll give you the history.
Duos Technologies Group was a rail inspection portal business that was basically going out of business. Over my 30 years experience of building and owning and operating data centers, I sold my last company called EdgePresence, which basically deployed in tier 3 and tier 4 markets, smaller data centers. So, I sold that, and in my backyard in Jacksonville, Florida, there was a company called Duos Technologies Group. They did rail inspection portals. So basically, that company was suffering. So I went to that business and I said, "Look to the shareholders and the board, let me turn this into a data center centric business. You have the platform for it.
I'll get out there and build the stock back up and build your investors get their money back." So what we did is recently, in the news you'll see in July, I sold that business off, the rail business. It was basically burning cash for years. So, we got rid of that business. Now we are 100% focused on what we know how to do, which is over 30 years of building and owning and operating data centers. How we're different is in a bunch of ways. First of all, where did my little clicker go? Okay, here we go. Just to go through kind of what we do. When we build data centers, we actually aren't the hyperscale, 100 MW. We found a niche, which I've seen for years, which is basically the 1 to 20 MW range.
Now you ask yourself, "Well, who needs 1-20 MW if everybody is looking at 100 MW sites?" Who needs those are the folks that are sitting on GPU right now and need to burn them. The reason why we go to 1-20 MW, basically, is we go to stranded power. Where there's power already there, we go and take it down, whether it was a Bitcoin miner or in a tier 3 or tier 4 market, which is our strengths. You'll see power that's sitting at the substation that they've built out for years of growth that never happened. That is what we call stranded power, and its power that's available. When we go to these markets, we can actually deploy in under six months. We can build a 1 MW facility or a 20 MW facility under six months.
That's the key there. The key is obviously power. The second key is fiber. If you can find both of those together, you're in a winning combination. The other piece why we're so popular and why we can get this done in six months is we build modular. If you ever looked at our website, you'll see these modular buildings that we build, but they're not really where the GPU is going in. We build modular on the actual infrastructure side. Basically, you see the long lead times, you see the issues that they're having in these 100 MW, 50 MW, GW facilities, is they can't get that equipment in time. When you're in the 1-20 meg range, that's not really off the shelf, but it's pretty close to off the shelf.
I can buy 1 meg generators and parallel them to 10 meg in under four months. Really, that market you can deploy quicker and you can deploy at around-- Our latest numbers show we're deploying around $6 million-$6.5 million a meg, unlike the $12 million-$14 million that the big ones are using. The reason for that is because we build everything in a factory. What we call our E-House is our electrical components. That's our UPS, our switchgear, our main transformers, all that are built remotely. Then Trane builds our chiller plant, which is a closed loop system. They all come to the facility at one time. Basically, we have just land that we'll take, and we'll build a tilt wall building really quickly, and that's where the actual cabinets and GPU go inside.
Or we'll take over an existing building and bypass everything in the building and bring the infrastructure outside. That's really our niche on how we deploy quickly and why we're deploying in the $6 million-$6.5 million range. The next question, and like I said, we're filing an 8-K, so this whole presentation will be on there. Basically, we're the landlord, right? We build the facilities. We don't own the GPU. We really don't really care what they bring into our facility as long as they pay our rate. Our rate right now is basically, it's very strong. We charge about $180 a KW, and then we do what's called a power passthrough. It's almost like we build a powered shell for our customers. We don't mark the power up. It's a straight passthrough.
When you see our numbers, Chris will go over those in a little bit. Basically, our profit is on that 180 per KW, which is really unheard of right now in this market. The reason customers are paying that is because we can deploy so quickly, and they can get their GPUs to work. They are not sitting on them. They are getting the pressure to burn them. It used to be the cluster had to be a MW. Now they are looking at clusters that are 4.5 MW. With that said, we just deployed our first cluster, with Axe Compute, who has a tier 1 customer, tier 1 hyperscaler, that we just deployed 4.7 MW in under 75 days. That customer got to revenue very quickly, got to turn their GPUs. Supermicro came in, delivered their product. We had it up in 75 days.
That gets them to revenue quickly, and it gets us to revenue quickly. It is very important that you understand that we do not own the GPU, we do not take the risk. I do not care if that GPU goes like this in price or comes down. Ours is a fixed price, and now they are 5 to 10-year terms is what we do. Basically, our model is we basically buy the facility first, or we find the powered land first, then we go to contract. We sign with the customer. The big issue you have seen in the market and you have read all about is people are not getting their GPUs up. They are signing contracts. We have a customer, a tier 1 customer, who signed 80 plus colo agreements last year. Three of them came to fruition. 3. 3 came up.
What they are doing is they are signing with a bunch of people hoping the power is going to come up, and it is not coming up. We actually have the power. The customer goes and sees the site. We are not waiting on the utility. Usually, the transformers are already on the property, so that is a big win. Then we can actually put them in service quickly. Then they deploy their GPU, and it does not matter if it is a hyperscaler CDN, it does not matter who they are. We personally are providing a colo environment where we guarantee the power is up 24/7, the five nines, and our SLAs show that. We are bringing it back to the traditional colo player. We are just like a DataBank. We are just like a QTS, but on a smaller scale.
I used to own Colo5 Data Centers were just like that, the big brick and mortar data centers. But I saw that there was going to be a market for the 1 to 20. Let me give you an example. Let us just use one of my old data centers, like in Jacksonville, Florida. That was a 2-meg facility. Those customers, I sold that in 2014 to Cologix. Those customers, a majority of them are still there. But now you have customers calling and saying, "Doug, I have 120 cabinets. My average usage is 7 to 12 KW. Now I am switching to a new platform, which is AI. Now I need 60 KW." They cannot do that. They cannot upgrade the facility for several reasons. One, that facility is landlocked, so there are buildings all around. You cannot add 16 more generators out there. Plus the cooling.
There's a whole new cooling plant. That issue is huge for enterprise. Everybody's looking at hyperscalers, looking at neo clouds. What happens to the Blue Crosses of the world? Where do they go? They don't go in the Microsoft's data center. They don't go into CoreWeave's data centers. They have their own structure. They have their own networks. They have no place to go. You're going to start seeing this 1-20 MW is a very popular spot to be and a need. The great thing about that is when you are selling 1-20 MW, you go into a community, it's a lot easier to sell a 1-20 MW and show them the tax benefits and show them that you're not using their water, you're not destroying their environment. They won't even know you're there.
It's a lot easier to get in and to grow our business. That's why we're strong where we are and we're deploying. The big issue in the market right now is, hey, how are you allowed to build when nobody else is? We're in that small niche of that 1-20 range. We actually just signed a contract and deploying in Roberta, Georgia, a 4 MW customer. That 4 MW customer is a neocloud, but it's a very well backed neocloud. You're asking yourself, 4 MW? You'll start to see that Anthropic came out with their latest news where they're going to deploy. I would predict they're going to deploy hundreds of 1-2 to 3 MW sites for inference. That's just one. You're going to start seeing these pop up everywhere.
Just a little background of what we've delivered or what we're delivering underway. We signed a 55 MW deal with Axe Compute. That's two different sites, a 35 and a 20 MW site, and those sites are going in Ohio, Carolinas, Oregon. They're going across the country. But that's a hyperscaler tier 1 that's behind Axe Compute. We'll take those, but we'll also take the enterprise. Don't forget about the enterprise, because you have the FedExes of the world, Blue Cross, you have UnitedHealthcare, you have all those other customers that can't sit in a traditional old school data center anymore. You're going to see that grow. Don't lose sight and keep focused on the hyperscalers. The real market is going to emerge as well as the enterprise customer. In service today, like I said, we turned up Columbus.
That's billing 7 MW right now. When we first started the business, we wanted to prove to the market that you could actually deploy GPU quickly. We took the risk, and we bought $144 million worth of GPU about seven months ago. We deployed that. We showed that we could do it. We did it in under 75 days. Those GPUs we just sold, if you saw the press, we just sold those to Axe Compute to their tier 1 hyperscaler. Now we are strictly just a colo provider. But we wanted to prove that you can do it in a modular environment very quickly. The hyperscaler, number one, we're advancing into Columbus. We're upgrading that to 10 MW. Right now it's 7 MW at that site. The 55 MW we signed. Here's the other key. The key to this business right now is power.
Everybody knows that. Power and land, right? We've identified over 100 plus MW of 5-20 meg tranches that are deliverable within 60-90 days. That's really what our market is. We don't go into a market hoping the utility comes to us and says, "Yes, we can deliver you 50 MW or 20 MW." But it's going to take us some time. We can't operate our business that way, and we can't commit to our customers that way. It has to be already there. It was either there with a heavy industrial, a company moved out, or it's a Bitcoin miner that had 5-10-15-20 meg site there that went out of business or is not using all the power. That's really how we identify and grow quickly into these markets. Power first site selection, integrated module build.
Like I said, we build everything offsite. That's why we can meet that under six months timeframe or quicker, and in-house sourcing, which we'll talk here a little bit about, which is extremely important to our business. The biggest part about this business is can you deploy on time, and can you deploy at cost where it makes sense, where you can make your money back in a quick term? I've built data centers, like I said, for almost 30 years now, over 30 years, and the key is how you build your data center. I can build a data center at $15 million a MW, it'll never make money. I can build a data center at $6.5 million a MW and have a three-year or under payback. And that's where you want to be in this infrastructure business. You want to be in that market.
How you do that? How you do that is you start a business, which we've done, that actually buys direct through the manufacturers. We buy direct from Schneider, Vertiv, Eaton, Corning. How do you do that? Nobody can just go and buy directly from them because everybody would do that. We hired some key folks that worked in the industry for over 16 years, for the Accu-Techs of the world, the Wescos of the world, and they committed to bringing their customers to us. What we've done is we started that division, which is basically our infrastructure side of the house. We buy all our infrastructure straight from the manufacturer. That means when I buy a cabinet or a UPS from Schneider, I don't have that 30% markup. I buy it straight.
Now, that division that we started, they sell to the DataBanks of the world, the QTSs of the world. They sell to the hyperscalers, and we have a very low margin on that business. 12%-15% is our average, but I have four people in that business, and I believe year to date, we've done almost $30 million in that business. It was really smart to bring that on so we have that buying power and keep our costs down. We're real proud of that department. These are some of the areas that we've deployed. You're going to hear today in our press release, in our 8-K, that we've secured over 100 MW worth of land and power throughout the country. And before I turn it over to Chris, I'd like to talk a little bit about our announcement.
The attorneys just cleared it right before we came on stage. We just signed a 20-MW deal in the Carolinas. That is with a very mature, well, got to be very careful, a very mature cloud provider, a public company, that is bankable, and we are excited about that. You will see that deployment hit by the end of April of 2027. We are excited about that. With that, I am going to introduce Chris, our new CFO, and he can shoot through the numbers.
All right.
Thank you.
All right. To hit on the number piece, one of the things we like to look at is everything from a unit perspective, unit economic perspective. In particular, looking at this, as Doug mentioned, we deploy at about just under $6 million per MW, and with land, it comes of about $6.5 million per MW. That is our build cost. Looking at how that lays out, then we are generating about $2.2 million of revenue per MW on an annual basis and about $1 million of EBITDA per MW on an annual basis. All in, that puts it just under a three-year payback timeframe per MW. We think that is pretty good, pretty good in the industry overall, and nobody really can compete with that. We are pleased with how we can deploy that and deploy that rapidly as we go forward.
Looking at a little bit of the history of the numbers, this is not the greatest comparison in the world because up until this year, this business was completely different. Doug came in early this year, really transformed the business, and in all reality, we have only been a pure play colocation business since August of this year. A lot of room from that perspective. We do expect to be just over $50 million of revenue this year and a positive EBITDA for the year as well. Generally speaking, again, that is where we are headed. Feel comfortable with where we are at for 2026. Looking at 2027, everything will be up and running.
As Doug mentioned, we have got a lot of contracts in play and that are signed to date, so we expect to have a little over $160 million of revenue in 2027, and sizable EBITDA growth as we go through the next year. Just summing it up overall, again, the way that makes us a little bit different from other players in the market, we are pure colocation. We always go with contracted, signed agreements upfront. We are looking at power land in that kind of call it 10-20 MW range where we can deploy quickly. We are deploying at a fraction of the cost of anybody else in the market. We have got so much demand, quite honestly. We have got demand. We have got line of sight to properties and deployment. We are working on the capital side and getting that up and running.
As Doug just mentioned, we are pleased with a pretty massive contract that will be just under half a billion dollars on its full run rate basis, so we are pleased with where that is. We are really starting to see a lot of growth at more of that tier 1 investible grade hyperscalers in the back end. With that, I think that leaves us a little bit of time for questions.
Can you go through the model real quick with them on this last deal so they understand it is a 10-year
Yep. Yeah, the overall perspective of it is a 10-year committed contract. Total contract value is about $10 million, and that is a 20-MW payment side of things. Great contract overall, long-term, great customer behind it. We're very pleased with it and hope to add a few more of those on behind it once we deploy.
And we're roughly what per MW? $2.2 million?
$2.2 is per-
MW
on revenue.
This is about a $450, $460 million
460, yeah.
deal for over 10 years. We are real proud of that one. There is more to come behind that. Once we keep delivering, do what we say we are going to do, there is a lot more behind that in that size range, anywhere from 5-20 MW with this particular customer.
You put up a slide earlier where you showed $6 million roughly to get into the 1 MW, then when you showed $1 million to get that, then you talked about a three-year payback. What was I missing there? That looked like a six-year payback to me. What am I missing?
It ends up being, I guess it's a little under five years total per MW.
On the bigger one, yeah.
On the bigger ones, yeah.
Yeah, the smaller We build 4 MW ones that are under three, and the larger ones obviously take longer, but still under a five year at that rate, we're still good. Yes?
What are your capital needs requirements over the next 12 months, particularly with this new contract? [audio distortion]
So, a couple different things. Clearly with the contracts we have executing, there is free cash flow coming in. We do have, at the end of last quarter, we had $112 million of cash on the balance sheet. Clearly, we have deployed some of that into capital projects in Q3 as well. There is definitely a need for additional capital as we look at the market. The company has raised a lot of capital in 2026, mostly from an equity perspective. We have no debt per se on the balance sheet today. We are going to look to raise additional capital. We are always looking, to be honest with you. We are always out in the market looking. We will focus on debt going forward, and that could be more at a project level, and then also complementary on more of a corporate level as well.
But there is going to be a need for additional capital, and we are actively looking at that.
[audio distortion]
Primary debt.
[audio distortion]
Primary debt.
[audio distortion]
We do not have an ATM on our shelf currently. Anything else?
Questions? Yes.
When you go out to get the debt, about what percentage of the loan is to the value of the properties that you purchase? Do you think it will be?
If you are doing a pure asset backed, it would be about 50%-60%. A traditional bank would give you up to 80%. We are not really seeing that in this market, in this environment right now. It is more based on the creditworthiness of the ultimate customer, less on the land itself. If I look at our business long-term, if we look three to five years, I would love to have a great term facility that truly is asset backed with a revolver sitting on top, more of a true ABL type structure with a syndicate bank. I do not think we are quite there yet, even as an industry. It is more either institutional debt, you could look at mezzanine structures. But it is more backed on how the customer looks from the overall transaction.
But still, if you look at it from a pure asset perspective, probably in that 70%-ish range is comfortable. Most debt instruments, or providers, they are looking past that to ultimately who is the end customer, what does the contract look like, what is the time commitment, are there escalators involved with it, and ultimately, who is signing the back end of the paper. Anything else?
Will that debt amortize over 10 years. [audio distortion]
I am still working on terms. I do not know. Potentially, yes. Yeah.
[audio distortion]
There is a shelf open on our stock today. I think we've got about 40 million-ish shares available.
Yeah. About 160 million left on that shelf, I think.
Yeah.
Have you been taking down shares off the shelf?
We have not. The company did raise equity this year, and yes, it came off the shelf. But, since then, no, we have not been touching that.
Yeah. I think we're in a strong position now that we show the revenue producing off of Columbus, like we said we were going to do. Our numbers are tracking, like we said, for the year, so we're on target there. Plus now we're very picky on who we choose as customers. Now we have a backlog of probably over 20 plus customers. Now we're picking and choosing which one is the best for us for creditworthiness so we can take that. That's why we're focused strictly on debt. Pretty good.
Great. All right.
Well, thank you so much, everybody. Thank you.
Appreciate it.
Thank you for having us.