Duos Technologies Group, Inc. ($DUOT)

Good morning. Welcome to Duos Technologies First Quarter 2026 Earnings Conference Call. Joining us for today's call are Duos' CEO, Doug Recker; and CFO, Leah Brown. Following the remarks, we'll open the call for your questions. Then before we conclude today's call, I'll provide the necessary cautions regarding the forward-looking statements made by management during this call. Now I'll turn the call over to Mr. Doug Recker.

Welcome, everyone, and thank you for joining us today. Earlier today, we issued our earnings press release and at the end of last week, we filed our 10-Q for Q1 2026. The Copies are available in the Investor Relations section of our website. I encourage all listeners to view press releases and our 10-Q filing to better understand some of the details we'll be discussing during this morning's call. . At a high level, our first quarter results reflect the continued execution of our strategic transformation towards a data center-focused platform with our Duos Edge AI Technology Solutions division emerging as our primary growth drivers. As expected, results from the quarter reflected our in-progress transition away from the legacy rail operation and the planned wind down of the new APR asset management agreement, which was the primary driver for the revenue in the period. At the end of the same time, we remain on track to exceed our $50 million revenue target for this year, supported by our strategic partnership with Hydra Host and our growing pipeline of AI infrastructure deployments. Before I get into the exciting updates on our Duos Edge AI and Technology Solutions divisions, I'd like to first update you on our Rail Technology and Duos Energy subsidiaries. Since our last call, we've continued to make progress on the Rail division divestiture. The company is currently going through a fairness opinion on the value of the Rail division and this process is expected to extend into the second quarter. As previously discussed, this was a thoughtful decision that will enable us to redeploy capital, reduce SG&A and focus on higher growth opportunities. We will provide additional details as the progress moves forward. Turning to the Duos Energy Corporation. We saw a ramp down with reduced reliance on Duos services this quarter. As a reminder, in December of 2024, Duos entered into an asset management agreement with New APR Energy to help find new contracts to engineer, procure construct and operate fast power plants. This was pivotal for us to make our data center business transition that is currently underway. As we previously discussed, the AMA will conclude later this year but we will retain 5% equity stake in the parent of APR Energy. In Q1, the company reported $1.55 million in revenue with a cost of goods sold of approximately 544,000. This was a step down from the previous period, and we expect it to continue to wind down in the coming quarters. Now I would like to discuss our data center strategy and our newer line of business, Duos Technology Solutions. As we build and deploy data centers at scale, controlling costs and optimizing procurement is critical given the capital-intense nature of this market. As a smaller buyer relative to hyperscalers and large colocation companies, we needed a more efficient way to procure equipment, which led to the creation of the Duos Technology Solutions. This division enables us to reduce procurement costs for our own deployments while creating a new asset-light revenue stream, serving enterprise, hyperscalers and contractor customers. I am pleased to report that Duos Technology Solutions experienced traction throughout the first quarter. We successfully signed eight new large data center operators and increased our backlog to approximately $14 million, all of which is expected to ship and be invoiced in 2026. Our pipeline for the technology solutions is several orders of magnitude greater than the backlog as of today, giving us an additional confidence, and our outlook, specifically in the revenue ramp for the second half of the year. This new line of business held low overhead is high scalable while also being supported strong customer commitments. We expect the revenue generated by the technology solutions to not only replace the revenue from the new APR [ M&A ], but also provide better margins. Now I want to shift our discussion to the core of our new data center focused organization, Duos Edge AI. The demand for edge computing and AI infrastructure continues to grow rapidly, and we believe Duos is well positioned to address this demand through our modular data center platform, following our recent capital raise, including the $65 million financing completed in March, we have significantly strengthened our balance sheet and are well capitalized to support near-term deployments and future growth. Our focus for the first half of 2026 is to continue executing our sales strategy to acquire new customers in our markets to fully utilize capacity of each EDC . During the quarter, we made significant progress across two key revenue streams. The GPU as a Service and high-power colocation. Under our GPU as a Service agreement with Hydra Host, we expect to deploy 2,304 NVIDIA GPUs across our edge data center platform. This contract represents approximately $176 million in total revenue over a 36-month term with total anticipated revenue of roughly $50 million projected margins exceeding 80% and approximately $40 million in expected EBITDA. Importantly, in addition to the GPU as a Service revenue, this partnership is expected to generate ex rental colocation revenue of approximately $25 million over the term, further enhancing the overall economics of the relationship. We have already received $15 million down payment with an additional $3 million deposit pending currently. We are actively executing on initial deployments. We continue to expect revenue from this agreement to begin ramping in the second half of the year. Separately, we were awarded a high-power colocation contract to deliver 4.8 megawatts of critical compute capacity to support a leading hyperscaler high-density GPU cluster. Together, these agreements represent a significant commercial inflection point, establishing two complementary high-margin revenue stream and validating our edge data center platform and scale. At the same time, we are also seeing increasing demand for high-density data center capacity, driven by AI and advanced compute workloads with demand now measured in megawatts rather than kilowatts. These higher power capacity EDCs should provide much higher monthly recurring revenue for Duos. Duos currently has 10 megawatt contracted and an additional 15 megawatts planned for deployment in 2026, and we continue to expand our pipeline of edge data center opportunities to support growing demand for our AI training, inference and high-performance computing workloads. Geographically, we are expanding into multiple regions across the country. including Maryland, Iowa, Georgia and Texas, as we position the platform to serve both enterprise and hyperscale customers. Within our existing EDCs, we've also begun hosting open houses for the surrounding communities as well as prospective customers to provide an opportunity to explore how edge data centers enable faster connectivity, localized computing power and AI readiness. We've recently announced a few of these communities initiatives and expect to host several more over the coming months. Since announcing our recent contracts, we have seen strong inbound interest from hyperscalers, [ neo ] cloud providers and other large-scale compute customers. Supporting our growing backlog and pipeline, we are currently evaluating new power partnerships that will enable green solutions and faster deployments for our megawatt sites and expect to provide exciting updates in this area in the near future. In closing, we believe Duos is at a pivotal inflection point. We are transitioning to a higher-growth, higher-margin business model building strong visibility through contracted opportunities and pipeline and positioning the company to deliver meaningful revenue and EBITDA growth as we move through 2026. Now I would like to turn it over to our CFO, Leah Brown, who will go over our financials for the first quarter of 2026. Leah?

Thank you, Doug. This has been an encouraging and productive start to 2026 for Duos. The first quarter included several landmark announcements, strategic financing strong backlog growth, strategic investment and meaningful progress towards building a stronger, more scalable company. I will now walk through our first quarter 2026 financial performance and highlight key operational drivers that shape our results. For Q1 2026, total consolidated revenue was approximately $2.7 million compared to $4.9 million in the first quarter of 2025. Total revenue for Q1 2026 represents an aggregate of approximately $44,000 of Technology Systems revenue, $562,000 of Technology Solutions revenue approximately $532,000 in Services and Consulting revenue, $1.5 million from related party services and consulting agreement, and approximately $30,000 of Hosting revenue. The decrease in total revenues was primarily driven by the planned down draw from the Duos Energy and New APR asset management agreement, the AMA that Doug mentioned previously. The company delivered materially stronger gross margin in Q1 2026, generating $1.6 million in gross profit achieving approximately 59% margin, a significant year-over-year improvement. This was driven by a reduction of cost of goods sold largely reflecting the impact of the transition of the AMA, the associated decline in related costs. The company also recognized approximately $900,000 of revenue during the first quarter of 2026 and 2025, related to its 5% nonvoting equity interest in the ultimate parent of New APR. As this revenue has no associated cost of revenue, it contributed at a 100% gross margin. The company reported net loss of approximately $3.5 million for Q1 2026. compared to a net loss of $2.1 million for Q1 2025. The year-over-year increase was primarily driven by lower revenues resulting from reduced scope of services, Duos Energy provided under the AMA with New APR as well as higher operating expenses. As we discussed on previous earnings calls, achieving positive adjusted EBITDA and in Q3 and Q4 last year were important milestones for the company, reflecting the early benefits of revenue scale and margin improvement. In Q1 2026, adjusted EBITDA was negative $1.5 million. We did not report adjusted EBITDA in the prior year period, but on a comparable basis, this reflects the impact of the items discussed earlier. While we did not achieve positive adjusted EBITDA in the quarter, we expect improved profitability as revenue ramps in the coming quarters. Let's shift to the balance sheet. The company ended Q1 2026 with $33 million in cash and cash equivalents. Our cash increased significantly compared to December 31, 2025 as a result of our $65 million capital raise in March, which strengthened liquidity and enhanced our ability to support operations and fully fund our planned investments as part of our agreement with Hydra Host. As of March 31, 2026, Hydra Host has secured a customer for the company and this customer provided a deposit of $15 million to the company in May 2026 and with an additional $3 million currently pending. Now I'd like to turn to our 2026 outlook. At the end of the first quarter, the company's bookings represented approximately $43.5 million in revenue, of which all is expected to be recognized during the year, included contracted backlog and near-term anticipated awards. In addition, approximately $1.1 million of the contracted Technology Solutions deferred revenue recorded in 2025 and will be recorded as revenue in 2026, further supporting the company's performance. Based on these committed contracts, and near-term pending orders that are already performing or scheduled to be executed throughout the course of 2026, the company is reconfirming its expectation for the total revenue in 2026 to exceed $50 million. Let me briefly walk through how we bridge from approximately $2.7 million of Q1 revenue to our $50 million full year target, which we know is a key focus for our investors. The primary driver is our GPU as a Service business, which we expect to contribute approximately $26 million largely recognized in the second half of the year as the project comes online and utilization ramp stock. In addition, we expect to generate approximately $26 million from our Technology Solutions backlog, which provides a solid base of committed revenue. This includes $2.9 million currently recorded as deferred revenue that will be recognized in the second half of the year. We also remain on track to recognize $15 million of bookings as revenue in 2026, supported by an additional $25 million in backlog. We also anticipate the balance of guidance to be recognized due to incremental contributions from colocation and infrastructure services, driven by customer expansions, new hosting deployments and continued capacity build-out, along with new customer wins we are actively pursuing. Together, these visible drivers give us confidence in reaching our full year target. To reiterate, due to the timing of revenue recognition, a significant portion of revenue is expected to be recognized in the second half of the year. During this time, we also expect to return to positive adjusted EBITDA. Doug, I'll now turn it back to you for final comments.

Thank you, Leah. Our first quarter of 2026 reflects continued momentum as we execute on our AI infrastructure strategy and expand our edge data center footprint. The industry recognition we received this year underscores the strength of our positioning and validates the path we're on. We believe our strategy is aligned with several powerful industry trends, including the rapid growth of AI-driven workloads, increasing demand for high-density and energy-efficient infrastructure. The shift towards secondary markets with available power and a broader move towards modular, faster deployed data center solutions. At the same time, evolving power, cooling, sustainability requirements are all reshaping the competitive landscape, further reinforcing the importance of the scalable, cost-efficient and speed-to-market solution. We are entering the remainder of 2026 with a focused, disciplined and a growing pipeline of opportunities, and we believe we are well positioned to deliver capture the market opportunity. And with that, I will open up to questions, everyone.

[Operator Instructions] Our first question today is coming from [ Rafde Khalit ] from Ascendiant Capital Markets.

This is [ Rafde ] for Edward Woo. With your progress in the U.S. data center market, do you have any plans to expand internationally?

Right now, good question, because we are getting a lot of inquiries internationally, especially South America. I was actually in London last week that a lot of interest. But right now, our primary focus is to keep proving the model out here in the U.S. and probably stick with doing our 25 megawatts this year and our 50 next year in the U.S. .

Great. And one more question. With such strong demand in the U.S., have you seen new competitors enter the market or any change in the competitive landscape?

Actually, the -- there has been some movement. Obviously, [ Armada ] is in the business with their -- it's a different approach. They're more privatized with Microsoft, but you're starting to see the need for inference. You're starting to see the need to compute more locally and the power, obviously, is an issue. So you're starting to see a lot of movement going the modular way and going after that 5 to 10-megawatt range, so you can deploy it quicker.

Your next question is coming from [ Scott Buck from Titan Partners ].

Doug, on the Hydra Host GPU as a Service agreement, can you provide the status of what hardware deployment and site readiness look like? Trying to understand whether we start to see some revenue in the third quarter versus even later in the year?

Yes, absolutely. So as of Thursday of last week, [ Super Micro ] and NVIDIA have received everything they're doing the rack and stack at [ Super Micro ], so the cabinets will be fully utilized and shipped on site. So actually, that brings us about a week -- it takes about 3 weeks off of our lead time. So fingers crossed, we're looking in that for it to start building instead of August, July 1. So everything is pointing in that direction. So we should be a month out of schedule. That would be a $4.4 million in revenue starting Sorry, go ahead.

Great. Can we potentially see this partnership expand to other locations?

The Hydra Host partnership?

Yes.

What we see with the hydro's partnership going forward is, obviously, on this first model, deploy GPU as a Service, right? We actually bought the GPU. That's not our model going forward, but they do have tons of customers -- actually over 12 customers that are interested in 5 or 10 meg that other folks have bought the GPU that they need to deploy. So our partnership with Hydra Hosts will keep growing, and it will grow on the colocation side.

Great. I appreciate that. And then last one for me. You scaled up some costs during the quarter. Do we expect that to continue through the remainder of '26? Or does the current kind of underlying cost infrastructure to support the anticipated growth through the end of the year?

The cost of our infrastructure are you referring to like our $6.5 million per meg?

Sorry, Doug, you took up some sales and marketing costs, I think, in the quarter, and I think it may be a little bit of [ D&A ]. So I'm just curious whether you need to continue to add to OpEx to support the top line.

Yes. So let me talk about that real quick. So obviously, all the investors on the call today realize that we are moving the Rail business out. So the challenge has been separating the two. A lot of folks look at us as a Rail business and then they dig in and they see what we're doing and then they're extremely excited and happy. So what we're doing is we put a lot of capital in the very beginning of the year, and we will do that going into the second quarter to really distance and separate the two businesses. So there's been a lot of marketing expense for that. And obviously, [ Gateway ] we've hired, who is doing an excellent job for us, and we can already see the calls coming in correctly and the investors having the right pitch and having the right expectation of what we're doing. So it will fall off around July, August time frame because we're making great progress. So I think we'll slim that down, but that definitely was a need that we had to do.

[Operator Instructions] Our next question is coming from [ Alan Klee from Maxim Group ].

Yes. How do you think about the CapEx, your CapEx spend over the next 12 months?

Yes. So the CapEx spend over the next 12 months, we're looking at deploying our first -- another 5 meg site and roughly, we're at [ $6 million to $6.5 million ]. So roughly $30 million is what we're anticipating in the next 2 quarters to deploy to meet our goal. And then we'll probably deploy another $30 million towards the end of the year to stay on track. We will obviously procure more product for next year to make sure we hit our number for next year. But we are on track. We're well funded to hit our number of the 25 megawatts this year, and we can do that with the funding that we have.

Yes. That's clearly a competitive advantage. And then strategically, it looks like your contract that comes on later this year, it's a 3-year contract. How do you think about what you do with the GPUs after that contract is over? Do you think there's an option that they could get renewed?

Yes. There's two options there that we're actually looking at. So one, the market is saying, right, and it can change. But the market is saying that those GPUs are going to be worth $50 million to $58 million in that range after the contract is finished. So we have two options. One, we can turn around and sell those GPUs and go back to a straight colo play, and then we're out of the GPU business. or we can actually go back to that customer, which is common from what we understand. We go back to that customer. And we're not getting 100% of the revenue that we did on the first term, but probably anywhere from 40% to 60% of that normal revenue. So we'll look at both applications. It just depends. If we want the capital to expand, we'll probably sell those GPUs. So we can use that capital to put back into infrastructure.

That's helpful. And you did mention you also have colo opportunities with the Hydra partnership. It started out with you buying the GPUs going forward, you could be getting customers that already have used to deploy in those type of situations, what would your responsibilities be?

Sure. We actually -- so when we build a 5-megawatt site, a modular 5-megawatt sites, say, in Iowa, our responsibility is to bring power, cooling and connectivity. So we are basically a co-lo just like a [ QTS or an Equinox ], anybody, the big brick-and-mortars were just very small, and we provide all the services. They bring their own gear, they bring rack and stack. They bring the infrastructure as far as the compute we provide the infrastructure as far as the power, cooling and the reliability of the [ Five 9 ], the backup power, the generators, that's our core business.

Okay. And then for those opportunities, do you view them? And then I guess you would sign on to longer-term leases with potential customers? Is that the way to think of it?

Correct. Those are typically 5- to 10-year terms. So obviously, we like those terms a lot better.

Okay. Great. This is impressive what you're doing -- maybe one other question. Just since I'm a little newer to the story. But for , could you go through a little bit of what Hydra Host is bringing to the table? What expertise?

Yes, absolutely. So Hydra Host is basically a GPU as a Service company. They do not own the GPU. Their specialty is selling and supporting the GPU. So basically, they have the, let's say, the hyperscalers as a customer, they basically go to companies like myself or investors or data center operators that want that GPU revenue. So they'll go and buy the GPU. So the customer owns the GPU. Hydra host just manages the GPU, the install, they manage the sales. So basically, they bring you revenue and they support the GPU. You own the -- you take the hit on buying all the GPU, but in return, you get the revenue and it's a revenue share. They get a small portion of the revenue. So it's for people that aren't in the GPU business that want to be in the GPU revenue business basically.

Makes sense. And then in terms of what you said, your responsibilities are with colo with power and interconnect and all that? Explain also like who you partnered with to do those things and what their background is.

Sure. So basically, our equipment is back. Obviously, at [ Schneider ] Electric. We use a lot of stir electrical equipment. We use [ Vertiv ]. And then in-house, our team in-house, we have roughly 22 [ folks ] that what we do is we monitor with our NOC, we have two NOCs. We have one in Jacksonville and one in Amarillo, Texas. Those NOC monitor the pods 24 hours a day. So that's break/fix. AC unit goes down. We just back within 2 hours. Everything that's built with our POS is just like a Tier 3 data center, it's what's called [ N+1 ]. So everything has a redundancy factor to it. so you have time to fix it. So if something does go down, it's not hurting the business. It's -- you're still delivering the [ Five 9s ]. You have time to fix it. That's why we have dual generators. Everything you see is what we call an A and a B feed, and we maintain all that. Now we do some it out, obviously, the contractors that are in those markets, but we control the dispatch, we control the contracts, we control all the servicing.

Right. And as you go forward and are looking at little opportunities or building out, how are you this my list? How are you strategically thinking about like finding power opportunities?

So when we look for power, we're a different breed, right? So we're not going into a community, looking at 100 megawatt. We're going to where power is what we call stranded. So when they build a substation and say a city in Iowa, right, that they build a substation, they build it to 20 megawatts because they know that community is going to grow, there is actually extra power there. Our team goes out and finds where there's 5 to 10 meg stranded power, and then we go contract it and move quickly. A lot of times on our two sites that we have basically, they were old bit mining sites. So the power is actually there. They thought they would use a lot of power, they never did. So there's 10 meg available at the site. What we do is we do a lease with the actual land owner, and we'll take that over for 20 years. And well the land owner makes money on the power as well. So it's to their benefit to bring somebody like us who actually is going to use the 5-megawatt to 10 megawatt, then a bit miner who goes up and down 1 month, it's a May 2 months, it's not consistent load. They want to make money off of power, so they like our model.

Next question today is coming from [ Justin Tapper from Shay Capital ].

So question the 10 megawatts on the colocation business. So the 10 you've signed, you guided to 25 this year, so you'll do another 15. And then next year, I think you said 40. Maybe can you just talk about the demand out there? What type of customers want this to the same customers that would want like you mentioned before, the 100-megawatt sites? Why would they want something like a 5 or 10 megawatts from you? Maybe you could just help us to sort of bridge the gap of demand there? .

Absolutely. Great question. Thanks, Justin. So yes, so what we're looking at in markets right now. Let's back up. So when we deployed the first one with Hydra Host, we were starting to get tons of calls for 5 to 10 meg. I've been in this business 30 years. I'm thinking why all of a sudden some want 5 or 10 meg when everybody else is looking at gigawatt and 100 meg just it's crazy the power they need. So what's going on is that the 5 to 10 to even 20 meg sector is going to be the hot new sector. That is for training or for inference models. So what's happening is they need to deploy their GPU and they need to deploy it quickly. Everybody knows that people are sitting on GPU. They've bought them, they've invested. Now they need to burn them. So to actually get 5 to 10 meg up quickly, we can do that under 6 months. You can't do that in the other models that people are deploying. So the Microsofts, the Googles of the world, all of them, they're all getting an inference and they know they need to capture these pockets to do their inferencing. So the 5 to 10 meg range, I can tell you right now, I look on my wall right here, I have 21 [ neo ] clouds. If I had 5 or 10 meg, they would take it. So the need is there. They're trying to build their networks out now for inferencing. It's finally there. People have talked about it for a while. Now they're doing it. You can see press releases from Google, how they're looking at doing 20 sites right now, the same thing. So it's time and they can deploy quicker. And obviously, speed is of the essence right now.

Got it. Great. And then just a follow-up for me, just on the balance sheet. So the $33 million in cash, I think Leah mentioned the $15 million received prepayment with another $3 million on the way. So as we -- that was in May, so then that's additional cash to the $33 million you filed as March 31?

Correct.

Right. And then just one also two. So a question. So the -- on the FTC website over the weekend, there's a filing that Elon Musk purchased ATR Energy. Doug, just can you confirm, is that the APR Energy you have the 5% stake in if there's any details you can provide us there would be great.

Yes. I'm not at liberty to say that today, but it is the same, obviously, the same company, but I can't discuss that today. Hopefully, we'll have some news from them shortly.

Next question is coming from Nico Sacchetti from RBC.

Yes, that last question was my first question. I read about, there's no details released, but it looks like that will capitalize for you. Whatever the details are, if this goes through your 5% of whatever the number is, is going to come into Duos, correct?

Yes, sir.

If the sale takes place. Is that going to be taxed? Is the number that we just do 5% of whatever the number is, the number that's going to show up on your balance sheet? Do you have any idea? Do you have any carry forward? Like what will that look like?

So just looking at the funds that we would receive. The agreement has a waterfall effect. So it's not a straight calculation just doing on the transaction.

Yes. Sure. I just meant is it -- from a tax standpoint, not what is the number, but let's just say he buys it for $100 million. I understand it, we don't know exactly what the 5% is, but the 5%, is that going to be taxed as like a long-term capital gain where you're going to net out in amount of it? That's my question is just to speculate on the number, obviously, we don't know. I'm saying, if it goes through, what would the tax look like? Would you -- would it be a gross or a net number? That's what I'm asking.

So I would say at a high level, we do understand that, that is a capital gain. But we don't want to devote any definite calculation around that transaction at this time. .

But I think I can answer your question a little bit better, Nico. With this movement of the rail business that we're doing, I think we'll have a substantial amount of NOLs. So I think we'll be in good shape. But we'll report to you as soon as we know.

Sorry, a substantial amount of what? I just didn't catch that.

The movement of the rail we just had a substantial amount of what Well, NOLs. I mean we we've lost a lot of money in that division. So we'll.

No, I know I mean, behind the scenes, I'm sure you're excited, right, because this is getting the company into the actual company that you want moving forward, correct? Like focused on what you want to be doing on the data center.

Yes. That's exactly correct. And if you look at our business, obviously, I came into this role and I brought this product to this business for our shareholders. It's the best thing going in the market right now. We just need to separate and focus like a primary example, we it has -- everybody knows this on the call. It's been a challenge, right? So we've burned through $900,000 on that division. We need to exercise that. And we're doing that here. And I would like to commit to you, I'll have that done as soon as possible. We're almost at the finish line with that. So we're excited about that. We're excited about a bunch of stuff this week. So we're extremely excited about where the company is going, and this is just things like this just another arrow in our quiver. This is good stuff.

That rolls into my next question. usually, you're really fired up on these calls, and it seems like it's a little more dampened this call. And that's with the $2.7 million of revenue for the quarter. Obviously, that's not what I think anybody is looking to own the company for is a number like that. So I'm curious, do you think that this is like understanding this pivot is happening, the work that you're doing is maybe you're booking it now, but the revenue isn't recognized yet. Is this like the pivot quarter or quarters? Like is this something that we should expect or should have expected? Or was something that you were expecting where you booked all of this. You've got this -- the Tech Solutions backlog we just aren't recognizing it in this quarter. So we shouldn't look at $2 million as like, wow, what a flop of a quarter, you are doing work that's going to get paid in the next couple of quarters and moving forward, where, if anything, this number, is it immaterial? Or is it worth like questioning this quarter? That's my question.

Yes, prime example. If I could shift all that stuff because that business, you have to ship it, right, to recognize revenue. I could have booked $14 million this quarter, right? So -- but I made it clear on the last earnings call that we're going to see this revenue start kicking into second, third and fourth. So -- and that's always been our model. Our model right now, obviously, is keep going, keep doing what we're doing, get these other sites up because once I like that, I was say, that's another 5 megawatts at $2 million a megawatt, right? So and you really start seeing this kick what we're doing is exactly what I wanted the team to do is build these, deploy them keep focus, keep your head down, the infrastructure division keep running. I apologize if I'm not excited, but are working 24/7, and it is good stuff. I'm sorry, I did that, but I you are extremely small.

Yes, I just -- that was the $2 million. I just wanted to ask the question. You mentioned Iowa where in Iowa, is that work happening?

It's the name of is [indiscernible], I always say it wrong. [indiscernible] [ Scokin ] Iowa, I'll put it out there. We have a press release coming here next day or 2. It's right outside of Illinois. So I still get that low latency down to [ Surac ], which is important. And our hyperscale customer wants that location, and they want the one in Texas as well, the one in Texas is right outside of Amarillo. So we're partnering now with somebody that is going to be a great partner of ours moving forward. They're already a partner.

It's not a major metro area in Iowa taken.

No, no. But yes, but you're only 20 miles out of a major market area. Yes. And that's where the power was stranded, that's where the power is there. I literally, if I had 10-megawatt worth of infrastructure, like coming off the line, I could like 10-megawatt there today. It's there because it's transmission down. It's beautiful.

Okay. Okay. I think I have two more questions. Just around this whole backlog like revenue, the fact that you do the work now, but it's not showing up. And then I saw some backlog things even earlier in the Q&A, you were talking about deploying 30 megs. There's a lot of like language barrier for like are we deploying something? What does that translate into revenue? So just to try to -- for the sake of getting things like understandable. And even like in the release that after the quarter, you received a $15 million prepayment from a customer. So like just trying to make sense of the different wording and what numbers or what? I am hopeful like the backlog numbers looked like they were broken down into booked backlog data center was $43.5 million. Tech Solutions looked like it was $14 million. Are those numbers right, that sounds right?

Yes, absolutely. .

So that just means that the $43.5 million in the $14 million are booked to business for what type of time frame? Or is there a time frame?

This year.

Okay. Is that something that you will start to report moving forward as like a broken down like back those two -- it would be very helpful if you did. So if you would, this really helps clear up the story. So then the $43.5 million in backlog. Is that inclusive of the $15 million prepayment?

So the $15 million prepayment, which there is going to be an additional $3 million that is pending right now. We will recognize that over the life of the contract. So the 3-year customer contract, you won't see $18 million being booked immediately. That will be over the life of the customer contract.

So my question is the $15 million prepayment that you highlighted in the release is that -- is that like cash flow coming in? And is that included in? So is that different than your reported booked backlog?

That is included in our reported backlog. But that will not be recognized until -- for the life of the contract, which is 3 years.

So yes, you get what I'm saying, right, is just to try to make sense and get the clearest picture and you highlighted the $15 million prepayment in the call. And then I'm just not sure where that fits in with some of the other numbers. And then the guide is like Doug is on the megawatts and so there's a megawatt guide, and then there's also a revenue guide, and so I just want to make sure that we're always talking about the same things, and it's not -- we're not talking about megawatts booked and built versus revenue because that I feel like that pendulum kind of swings back and forth with the conversation. So obviously, for the sake of getting everyone on the same page to get your stock to be valued with all of these good things that you're doing to try to just get the picture as clear as possible. And it sounds like we're on the cusp of maybe this railcar and this gas-powered turbine business being removed, and I think that will only help clean up the situation.

Absolutely. Absolutely.

Okay. Last one is just around like the actual units. So you mentioned competition, more competition, you mentioned [ Armada ] on this call. You mentioned there's a lot of like interest in this 5 to 10 megawatt range. And that, to me, is semi new information just from the standpoint of you guys are the only ones really doing it. The clean room is this huge competitive advantage. So I just want to clear up like what I watch a video of them with a semi mobile data center and someone is lowering one onto a Navy ship, like is that a bad thing? Is it real?

Yes. Nico it's real. It's a totally different application that we're doing. Yes. So they're really -- when I say there's people going in the market, I would just talk to more modular, right? So they're not going to deploy 3 meg, 4 meg, 5 meg. They're not doing that. And they're not for multi-customer, right? They're just for one privatized customer, do a lot for the government. So I would just use that as an example. But you see a lot now if you go out in the industry and you're in it like us, you'll see a lot of 3D renderings. You'll see a lot of people saying, "Look, we're doing inference. We're doing this. No one's actually done it". And that's why the two main hyperscalers that came to our facility, they toured last week, Corpus Christi. And they went down there to physically see it. Now that pod is not the high-density pod. They wanted to see physical work done and they wanted to see how we do -- how we build the quality of work. And they both signed off on it. So that's how we're winning the market is we've done this. I've done this 9 years, 10 years now. I've put over 30 of them on the ground in my career. You can go look at the first one, you can look at the one we just put down 3 weeks ago. So we do know what we're doing. And it's not rocket science, but we've got it down, right? We've got it down.

So what your comment was more of a positive that there's a lot more interest in this modular type of idea rather than there's a lot more like competition coming.

Yes. I'm positive. The reason why I'm positive goes because of this. One, it brings hype to the industry, right? Everybody is starting to look at it, and people are starting to make moves. The second is, I'll go up against any of them every day. And if you were a part of my sales organization, I have one trick, here's my trick. When we go to sell somebody like this is how we want our first hyper, I said, "Look, I'll pay for you to go see their pod. I'll fly you there. We'll take a tour together, and then we'll go see mine. And if there is no pod to see, you have to sign with me". And I haven't lost yet.

This is what I meant by the fired up. This is what I'm talking I'm sorry. you know what I mean, like I'll go up against anybody tells me that you're confident in your unit. And so what I meant by who's real is like what you just said, like there's a lot of prototypes, but who can actually make -- who can manufacture this at scale? Like is it you and [ Armada ]. Is that really the only ones for this place where you would compete? And then what -- is there any difference between you and [ Armada ]? Is it the clean room? Is that enough to protect you if this -- if there's not really barriers to make in a rectangular box like and there's all this interest, is that clean room enough to protect you?

Yes. Clean room is enough to protect, but they're using more of a ship in container. And I started my career with those 10 years ago. You can't do high density in there. So these are what they're doing is one customer, one to two cabinets. My customers, nobody wants that. It's really for -- let's just say Johnson & Johnson or a hospital or manufacturing plant. They just need a small one for their own privatized compute in AI. That's what they do. So they're in a different market but they are deploying modular. That's it. So we're not even apples-to-apples. I just wanted to mention that there's somebody doing module. A lot of people see somebody else is doing it. It's not the same thing at all, and it's more privatized. But it does give hype out there, right?

And I like No, you're right. If you Google the stuff everywhere. So you mentioned that you've done 30 of these, you also mentioned being able to sell these GPUs at the end of the contract, if needed. And I'm pretty sure somewhere over -- in the last 1.5 years at Taco, you mentioned that the first data center you ever built one of these is still packed with that older technology. It's still fully leased out. Do you really think that you would ever sell the GPUs? I mean, unless something goofy happened with AI. I mean someone would want that based on the comment about your first one, build 20 years or whatever place.

Absolutely something going to want. So in full disclosure, there's people want -- they're trying to buy our Hydra's contract right now. Oh, yes. Oh, you're telling me. We get calls I'm telling you right now, gentlemen, if I had enough capital to deploy 10 meg sites, they would be full. I would put them up for auction on a Tuesday, and they would all be gone by Thursday.

Well, that's got to be -- there's got to be someone that's hearing that and understanding that unit produces revenue that would be worth putting up, what, 3 years' worth of that capital to borrow it to or however that structure. Anyways, you don't need to answer that. That's more of a fact.

Yes. Remember our strategy. Our strategy was to do this and to show that we can do this, number one, and we can house GPU, and we can deploy quickly. And whose eyes did we want to get on this. We want to get the NVIDIA, [ Super Micro ], those folks. Because to be honest with you, I'm not going out for -- I don't want to go out for any more equity. I don't. If I can't have somebody backstop me like an NVIDIA or a Dell or somebody like that, which they see the need, number one, they're seeing it now, and you're going to start seeing them do this. I want to be first in line because we're the first guys out there holding a flag saying, "Look, come look at our stuff. It's real. We're doing it. You can talk to our customer". We have $144 million worth of GPU. So we're in.

Well, if anything, would -- if I look at you and think why would you be talking to NVIDIA, is it -- would it be correct to say you would help NVIDIA because they have this backlog of GPUs that their customers have bought but can't take delivery of because they have nowhere to put them. And you can do these things in 120 days where these things are being pushed back on being built in these rural areas, the big data centers. If anything, they're incentivized to help you get these things built because it actually releases their own backlog to them to recognize?

You got it. You've been sneaking in my office, reading my playbook. But yes, that's it.

Exactly right. it just logically makes sense. These big data states are getting pushed back and so you guys can do them quicker anyways, there's advantage to an NVIDIA or someone like that is the reality Okay. Last question is on power. At some point, with all the stuff being built, we're going to hit the capacity of our electrical grid. I'm sure you know more about the into. I'm hoping that you do. Like what would have? Like something has to happen if this really is going to continue at the pace that they're saying it will for a power. And so I've heard a lot of ideas out there. One of them that's like this holy grail but has a lot of like skepticism is nuclear with like this nuclear SMR or whatever they call it, that's like a power source. Number one, is that even a reality? Number two, what do you think will happen when we hit the electrical grid capacity. And then number three is like my understanding of these things is like, yes, it's great energy and efficient and low power, but it's also like a bomb that would be sitting next to your hardware like the most important stuff you want to save. So even if it works, in practice would people even use it.

Yes. So here's our secret, right? The SMRs are going to take a long time, right? I don't see a lot of communities allowing a nuclear mini plant coming in. I think that's years out, it would be a good idea if it happens years out. We're not looking that far out. We're looking for today. So if you noticed, and you heard on my earnings call, I talked about an alternative green company, right? Well, you've seen the guys like at [ Bloom ] that have done this for Google just basically bought out their production for the next 2 years. Those are fired on natural gas. They can deploy on our side, probably 6 months, 6 to 7 months. They can bring 10-megawatt up where you're not touching the grid at all. And it's somewhat green. It's good on the environment. It's not bad for the environment. It doesn't use water. It's all natural gas. Those are the type of partnerships and things that we're pushing to go down those path. And hopefully, you'll hear something soon. But it's -- that's the way to go. I'm telling you right now because it doesn't touch the local community and you can deploy those and it takes natural gas.

Well, that's your next idea after this clean room as you should patent is building with an electrical windmill on top of the solar panels. And then I mean even the pushback on the nuclear, even if it worked, it's like I think it's just new -- that's what I was asking you. It's just like public reception where like your boiler and they think, well, that's safe or that boilers blow up all the time. So anyway...

I can take that offline. Yes.

Our next question is coming from [ Richard Jackson from Strategic Assets ].

That last conversation was extremely helpful. Most of my questions were covered, but I got two more here, one short, one long. You said that when you when the client owns the NVIDIA chips, your responsibility is the maintenance, the power and the connection, I'm assuming that means that your financial responsibility to connect these new centers with fiber?

Yes, yes. So the fiber carriers come. So when we find a site, we make sure that it's rich and fiber around the traditionally long-haul fiber or buy a highway where that's where all the fiber runs down. And then the carrier actually brings it in because once the customer says he's there, the carriers come because they use so much bandwidth. It's worth them to build into the infrastructure and they need multiple paths. So we actually don't own the fiber. We don't own the connectivity, the carrier does and they sell directly to the customer, but that's what brings them in. Once you build power and the customer signs, that's when they come in.

Okay. So the fiber is your responsibility, but you obviously strategically placed these places where the cost of the carrier is minimal to the contract, correct?

Exactly. Exactly correct. Okay.

That's helpful. By the way, I was at Corpus Christi when it was announced so that really helped me crystallize going on here. [ Bill Radfords ] is a treasured employee, you got there. He's also...

So you actually went to the site?

I did.

Excellent. Thank you.

Hopefully we kind of know what the hell you're doing here.

So what were your thoughts? So a lot of our investors don't get the opportunity to go to see the actual pods. What was your impression?

The two things that I found most enlightening was, number one, you don't use water. It's a pure air cooling system, but the way build structures, the stacks is explaining that air can do what it typically can in other centers because of the way he has the airflow.

That's right cold [indiscernible] containment. .

It help me understand how you get away with that and still provide 100% availability. And I understood why the marginal costs are so low. The center pretty much runs itself. You just got to react to problems. . Anyway, that was very helpful for me. But I'm having at that time modeling all this, and maybe a phone call offline would be better, but...

Yes. Call me anytime, that way we can take our time and go through it, but that would be great. We'd love to have that call.

Okay. Great. Let me just throw out three assumptions I got. You tell me from way off on these, all right? . So it costs somewhere between $1 million to $1.4 million to build these 10-megawatt centers. correct?

No, no. Per megawatt, it's about $6.5 million per megawatt. The $1.4 billion are for the 300 KW ones. Just so you're clear.

$5 million per megawatt. And you're trying to lease these out and doesn't seem like it's too difficult to do it within a reasonable amount of time, a lease rate that's in the ballpark of 30% to 50% of that construction cost. Is that about right? .

Yes, it's a little under $2 million per megawatt in revenue. .

Okay. Okay. Okay. So it's costing you $6.5 million you're generating $2 million. Okay. And that equipment look pretty steady. I mean how long do you think those things are good for ballpark?

They got our generators, everything else is good for 20 years. The only thing that you're going to swap out of your batteries, right? Your UPS, those batteries are got an 8- to 10-year life, when you swap those out, you're looking at another infusion of probably, let's say, high side $80,000.

Okay. Okay. Okay. Now I'm getting excited. Okay. Now on the revenues that you're posting, are these revenues people paid you to build your centers? Or are these monthly lease revenues? Or is it all mixed stuff and I should plug over one by line?

No. It's what they pay us to lease. Traditionally, there's a onetime install charge that we charge a customer to move them in. But our model, our revenue comes from reoccurring. So that's what the customer pays to be in that facility a month. And they pay based on power. So how much power they use is how much you can base their revenue on.

So the revenue you're targeting at the end of the year, how much that's going to be repeated every year? -- half, 80%? But it's least on, right?

Yes. The recurring revenue is going to be through the life of the contract. So for our GPU as a service, that's a 3-year contract. And then most of our colocation contracts are anywhere from 5 to 7 years. So that's recurring.

Yes. the only the ones that aren't recurring are obviously the infrastructure division side. That's the equipment side, where we sell equipment to data center operators like ourselves. And you'll see that...

I know that you said you're hoping to not go to equity anymore to finance this. And I understand things happen. But in your mind, when do you think the monthly free cash flow self-finances this growth, 2 years out, 4 years out?

Well, in our business is a very CapEx -- and when you're in the data center business, it's a very CapEx sensitive business, right? So it just depends on your model. My model is to keep growing. So what we're going to do, and that's -- this is why we're doing what we're doing. We're going to go after debt finance. We're almost there as a company after we get another one of these on the ground, we should be able to do that all day along with our model and the customers we're bringing on, that shouldn't be an issue. So that's how we're going to fund this business going forward because we're always going to be building facilities. Because you want to. We're like me, right? We want to have as much assets out there as possible because the value there, prime example, I build that site in Iowa, and I build 10 meg there in 4 years, 3 years. Say those contracts are 5-year deals. In 5 years, say that customer moves out, I've got 10-megawatt available there. Every data center, big brick-and-mortar data centers that I've owned and sold over the years, not one of them have gone the other way. And every one of them is at capacity. So once you own the power and the infrastructure, it's golden.

Okay. That's very reassuring. So I'm walking away from this thinking and I would like to take your offer up on just looking at three or four different forecasts.

That's what we do, you call us any time.

Within a year or 2, you're generating enough cash flow where you're making a margin of, let's say, at least 15%, and you're financing hopefully under 110% debt. Am I in the ballpark on that?

Yes. For a 15%, we're in the wrong business.

But yes. Okay, good to hear that I'm thinking more should be 25% or better, right? .

9% we're financing that. But we should be in the 70% range. margins.

Our next question is coming from Nathan [ Frankovitch ] from [indiscernible].

I think you touched on this a bit earlier, but if you could just add a bit more color on the success of the high-power 1 to 2.5 EDCs? Can you just talk a bit more about why customers would want these versus what we see a lot of other companies talking about with the mega data centers at 100 megawatts plus? Like what kind of companies would want these and what the end use is for?

Sure. So obviously, number one, these big facilities you see are these big huge training facilities, right, that you're building. These are huge facilities that are outlay out there, right? So now with inference, you have to be where the data is, right? These aren't training. These are inference, right? So they have to compute where the -- what we call the eyeballs are. So where the data is performing and transmitting, it's got to be close to that. So that's why you're starting to see the Googles of the world and the [ Caruso ] and everybody else, deploying these inference sites. And you have to do that somewhat close to the data. So that's why you'll see us going into these Tier 3 markets. Prime example, Corpus Christi. Corpus Christi, I can still get 5 milliseconds to Houston. Amarillo, the same thing, Waco. Those markets, you can still get that. And in those markets, they need inference as well. So the local hospital, the local government, gaming, all that stuff that you're starting to see that the data needs to be where the eyeballs are. And we've seen it for years, but now it's really starting to kick in where the amount of massive data and compute have to be done localized. And that's why before back in the day, a cabinet was 10 -- max 10kW. Well, the type of compute you have now, it's 100kW cabinet. You can't put that in a normal data center. So you have to build for that category. So that's why you see all these new builds going up for this new inference. And you're going to see a lot. And keeping my eye close on Google, because I would like to get into Google and say, "Look, we can do this for you. You're doing it anyway. You don't want to be in this game. We do it. We do it well. You just want to be an OpEx model, not a CapEx model, we can do that". They just want to compute, they don't want to have to maintain facilities. So obviously, it's going to grow and everything, you guys are one of the best at it. Obviously, you guys know this business better than most. And you see where it's going as far as inference and how you can deploy a lot quicker. So the [ neo ] clouds of the world are -- want to get their GPU out burning, and this is an even faster way for them to do that as well. If that makes sense.

Yes. That's helpful. And I appreciate that. And then I guess, I think you already spoke to the revenue per megawatt on colo, but in terms of the economics of scaling, can you just touch again on EBITDA margins based on that CapEx and revenue? EBITDA margin per megawatt?

Sure. So our EBITDA margins for full year is 17% and adjusted EBITDA is 27% full year consolidated.

Great. But then specifically to the colocation business, do you have that broken out?

Yes. For a high-powered colocation, our EBITDA margins are about 80%. GPU as a Service is around there as well.

Next question is coming from Caroline [ Gangi ] from [ Aramis ].

So it sounds like just kind of sum this up, so in my head. You're going to be going from 25 to 40 megawatts this year to the next year. And you made the comment that you can sell significantly more systems if you had capital and you mentioned you would hit the debt markets for that. Do you think that you would be able to find a strategic investor? And then finally, I think part of the confusion with the stock is that the company is kind of misunderstood. It's a lot of moving parts. You're changing business models. Divest these certain businesses. When do you think you'll get coverage on this -- for the stock equity coverage?

To answer that last question quickly. I think soon because what's going on now is people are actually starting to see our deployments, they're starting to see the revenue come in. And obviously, once this Hydra Host deal hits hopefully in the next 30, 40 business days, that will really drive. We have talked to a bunch of analysts that were in the middle of with right now that have come in and they've done a deep dive now they understand -- even the analysts didn't understand what we do. And they're helping us get the story out. So I would predict, hopefully, we see something in the next 30 days. And there's going to be some other drivers that are coming up that are going to push them to come see us. So I'm looking forward to that as well.

Okay. Got it. And then my other question about, as strategic investor. Is that something that's possible instead of having to raise money? .

Absolutely. So that's why we're -- I want to say we're still on our proof-of-concept model, but we're really not, right? We're executing now. So now is the time, and this has been my job to get in front of the NVIDIAs of the world that could backstop us on something like this because they like what we're doing. They see that inference obviously, is the way to go. They want to sell their GPU. They also have a lot of GPU customers that are sitting on. They want to move those so they can buy more. So that would be one of my strategic partners, but also these folks at back the big data centers, right, [ Blackstone ] and all the big guys that digital bridge. They all have stakes in these big data centers, right, [ Vantage ] and [ DataBank ]. Well, what's going to happen with those data centers. They're going to look at us as a hub and spoke. They obviously are doing training molecules, but they are going to want those inference modules out there. So when I say hub and spoke, you've got the main data centers in Dallas. But you don't have anything in Corpus Christi. You don't have anything in Waco in the suburbs and these outer smaller Tier 3 markets, Tier 2 markets. So you put these out there and they're going to buy them because now they have their hub and spoke model, where they can sell their services there. And that customer, 90% of the time that customer is in that pod is in their core data center. So I see -- just like my first company, EdgePresence, we did this before and [ DataBank ] funded us. [ DataBank ] funds us $35 million to prove that concept out. Unfortunately, we are bought before we could prove that out. But that's where I see this partnership coming and it makes complete sense and it's -- it would make complete sense to do it that way.

We reached end of our question-and-answer session. I'd like to turn the floor back over to Mr. Recker for any further closing comments.

Well, I want to thank everybody. And hopefully, you've got a good vision of where we're going, and we are running 150%. And I think you should be proud of us. We'll keep going and we're excited to see some good announcements this week as well. So I look forward to talking to each one of you. Please call me any time with questions. I'm here all the time. Thank you so much for your support. We look forward to talking to you soon. Thank you so much.

Before we conclude today's call, I'd like to provide Duos safe harbor statement that includes important cautions regarding forward-looking statements made during this call. This earnings call contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. Forward-looking terminology such as believes, expects, may, will, should, anticipates, plans and their opposites or similar expressions are intended to identify forward-looking statements. We caution you that these statements are not guarantees of future performance or events and are subject to a number of uncertainties and risks and other influences, many of which are beyond our control, which may influence the accuracy of the statements and the projections upon which the statements are based and could close Duos Technologies Group Inc.'s actual results to differ materially from those anticipated by the forward-looking statements. Rates include, but are not limited to, those described in Item 1A and duals annual report on Form 10-K, which is expressly incorporated herein by reference and other factors as may periodically be described as Duo's filings with the SEC. Thank you for joining us today for Duos Technologies Group's First Quarter 2026 Earnings Call. You may now disconnect.

Thank you.