Duos Technologies Group, Inc. (DUOT) Earnings Call Transcript
August 17, 2026
Earnings Call Speaker Segments
Good afternoon, everyone, and thanks for being with us on this call. We put out our earnings press release earlier today, and by Wednesday, August 19, 2026, we'll file our 10-Q for Q2 2026. You'll be able to find copies in the Investor Relations section of our website. I'd urge everyone listening to review the press release along with our 10-Q filing so you can get a fuller picture of some of the specifics we'll cover on this afternoon's call. Broadly speaking, the second quarter marked yet another significant step as we transform into a company focused on data centers and AI infrastructure. During the quarter, we kept executing our strategy — expanding our edge data center platform, growing Duos Technology Solutions, and pushing forward several key strategic initiatives meant to underpin long-term revenue growth and profitability. Adrian will walk through the quarter's financial performance, but I want to take a few minutes to talk about the key operational developments and strategic progress achieved during the quarter. Starting with the Rail business, I'm happy to share that on August 5 we successfully closed the sale of Duos Technologies, Inc. Following the transaction, Duos Technologies, Inc. now operates as an independent privately held company under the Duos TI brand, with Javier Acosta serving as President; this deal marks the completion of the strategic repositioning initiative we announced earlier this year. With this divestiture complete, we can now direct our capital, management, resources and operating focus entirely toward scaling Duos Edge and Duos Technology Solutions. We think this streamlined structure will let us accelerate execution, sharpen organizational focus and better position the company against the opportunities we see across the AI infrastructure market. We will provide certain transition services for a period after the closing, but investors should see this transaction as the completion of our shift from a rail technology company into a focused AI infrastructure and edge data center platform company. From here on, our attention is centered on executing our deployment strategy, expanding customer relationships and converting our growing backlog and pipeline into long-term recurring revenue streams. Separately now, New APR Energy sold substantially all of its assets during the quarter. As a reminder, we held a 5% stake in the APR parent company, and the sale valued our interest at approximately $60 million; we received $50.4 million in cash, with the remaining $10 million subject to a 12-month holdback that is recorded as a receivable on our balance sheet. Together with the Rail divestiture, these transactions essentially complete our full transition to a data center operation, and they also materially strengthened our cash position to execute against the opportunity ahead. Looking beyond 2026, we believe the strength of this balance sheet and the focus that comes from operating as a pure-play AI infrastructure company positions us to keep building momentum into 2027 and beyond. Alongside these strategic and financial steps, we've also continued to strengthen our leadership team to support the next phase of our growth. I'm happy to announce that Dipan Patel has joined us as Chief Operating Officer. He comes to us with a wealth of experience fueling expansion in the digital infrastructure sector, having worked with SBA and, most recently, Telstra InfraCo, and he'll be instrumental in driving execution throughout our edge data centers and technology solutions platforms. Additionally, our hunt for a new CFO is now in its final phase — we're seeking someone with substantial public market experience — and we can't wait to reveal further information once that process wraps up. We feel this hire makes us even better positioned to capitalize on the growth opportunities that lie ahead; with those announcements out of the way, I'd like to shift focus to the primary growth avenues for our business, Duos Technology Solutions and Duos Edge AI. All right; grab your popcorn, since things are about to get truly thrilling. Let's begin with Technology Solutions. This segment keeps building momentum and stays a vital piece of our expansion plan. Revenues for the quarter came in at roughly $3.23 million, largely fueled by ongoing expansion in data center deployments and the confidence these operators place in our capacity to protect and supply the essential equipment required to keep their projects on schedule. We also grew our backlog to $25 million, which shows sustained appetite for what we offer from enterprise clients, contractors, data center operators and AI infrastructure participants. The potential here is still substantial, since this business lets us bring in revenue with fairly modest capital needs while simultaneously backing our own infrastructure rollouts. We continue to feel optimistic about the opportunities emerging in today's market. Our expansion is still powered by the strategic partnerships we've established over a fairly brief span, and we're dedicated to providing solutions in a marketplace confronting extraordinary timing pressures, demand levels and supply limitations; we've locked in strategic MSA contracts, demonstrated our delivery capabilities and keep building momentum alongside our manufacturing and vendor allies. Nowadays, data center operators, contractors and hyperscalers require more than conventional order takers. They require partners capable of pivoting fast and steering through the biggest boom this industry has ever witnessed. That's precisely what our Technology Solutions team delivers to the market. Turning to Duos Edge AI, the demand landscape for AI infrastructure is still remarkably robust. One of the most notable milestones since our prior earnings call was the signing of a 5-year colocation agreement with Axe Compute, Inc., a neocloud AI infrastructure platform, to supply 10 megawatts of critical IT load capacity at our Columbus, Georgia campus; this deal carries a value exceeding $111 million in contracted revenue across the initial term and should go operational in the fourth quarter of 2026. We see this agreement as a key endorsement of our approach to building owned high-density AI infrastructure in appealing markets where power availability, deployment speed and operational flexibility matter greatly to customers; it also highlights the rising demand we're witnessing from large-scale AI and cloud customers in search of scalable infrastructure solutions beyond traditional Tier 1 markets. Also, as was shared earlier today, this partnership has taken a significant leap ahead. Alongside Axe Compute, we've unveiled fresh service orders totaling 55 megawatts of AI data center capacity spread over numerous U.S. sites as part of a 5-year deal — growth that directly extends the 10-megawatt rollout we intend to complete at our Georgia site. These contracts amount to an anticipated $500 million plus in total base payments across their original 5-year period. Within these expansions, the two sides have signed a linked agreement envisioning combined cash equity investments from Axe Compute reaching as much as $140 million into the project, contingent on necessary approvals, financing, finalized technical design and additional conditions. Axe Compute and Duos will co-own the new data centers, with Duos retaining 51% and Axe Compute retaining 49%; for Duos, such an arrangement offers a non-dilutive financing approach that lets us bring more data centers online more quickly. Early project readiness is set to kick off in late 2026 and carry on into early 2027, depending on construction, commissioning and performance testing; that appetite is clear throughout our pipeline and customer engagement efforts. Over the quarter, we held an open house across a number of markets, among them Lubbock, Dumas, Hereford and Abilene, giving potential customers, community leaders and strategic partners a chance to witness our deployments in person.
To illustrate, we just shared news about a chance to work with Zero Latency, or 0LAT, spanning as many as 15 sites plus 225 cabinets — that tells you how much interest exists throughout that portfolio. On top of those market development pursuits, our work with Nistar keeps moving forward. During this quarter, we still anticipate that deployment work will advance according to the mutual deployment timeline we set together. Keep in mind that this arrangement amounts to roughly 2 megawatts of contracted capacity, and it stands as yet another meaningful endorsement of our edge data center platform along with our capacity to serve customers who need high-density AI compute. Notably, we expect the Nistar deal to deliver recurring colocation revenue once capacity is brought online and customer utilization increases. Although recognizing revenue depends on final deployment and when operations actually begin, we see this project as one more substantial piece in building toward a bigger portfolio of recurring infrastructure revenue. Combined with what we announced in Columbus and the Axe Compute expansion, these deals confirm our edge data center approach and bolster our assurance about growth ahead; next, let's discuss HydraHost. The HydraHost deployment still ranks among the biggest opportunities this company has ever had and remains a foundation stone of our growth plan. Throughout the quarter, we kept advancing deployment work and customer onboarding. Last week we toured our Columbus facility, with our customer serving as host. They came away highly impressed by our deployment quality and speed, and by how far we've come within only a few weeks. Our attention stays on adding more capacity and helping customer utilization grow as deployment moves along. We expect revenue recognition to rise once systems go live and GPU capacity enters service. Financially speaking, we keep gaining from the agreement's solid contractual base, which includes customer deposits already collected and further funding milestones we anticipate completing under the contract's terms. Equally significant, we think this relationship opens meaningful expansion possibilities past the first deployment. Rising demand for AI training, inference, and high-performance computing workloads keeps generating chances for more capacity, and we stay involved in conversations about potential future expansion scenarios. All told, we're pleased with the progress so far and still believe this relationship puts Duos in position to take part meaningfully in the fast-growing AI infrastructure market. On the subject of capacity expansion, we keep carrying out our nationwide deployment plan. Our 2026 target remains about 25 megawatts, and we are 100% on track to hit that milestone. We also keep assessing ways to speed up deployments wherever customer demand and power availability make the economics attractive. As mentioned before, our approach isn't merely to add capacity — it's to place capacity in spots where power, connectivity, and customer demand come together to generate long-term value. The demand backdrop stays very favorable, and we think our modular method gives us the flexibility to scale efficiently while keeping a disciplined stance on capital deployment. Because of that, we still have faith in our capacity to keep growing our edge data center footprint, which supports commitments we've made to current customers as well as chances that may arise later. The patterns we're observing, in our view, keep bolstering both our business model and the possibilities for growth over the long haul. When we look past 2026, we anticipate that contracted backlog combined with a pipeline that keeps expanding, plus extra capacity being brought online, will keep fueling growth as we move into 2027, and our attention will stay fixed on turning that visibility into lasting long-term shareholder value; at this point I want to hand things to Adrian Goldfarb, our CFO, who is going to walk through our financials for the second quarter of 2026. Adrian?
Thanks, Doug. This quarter was the most consequential stretch our company has ever been through, and the transformation we set out to accomplish is complete. During the quarter, we closed the sale of our legacy rail business on August 5, and that operation is now reported within discontinued operations for each period shown. We also wrapped up the APR chapter. Beginning in the first quarter, the asset management agreement was amended so that fewer services would be provided, and the revenue tied to it is winding down, with just small amounts expected through the third quarter. The staff who ran that agreement, together with the full related cost structure, have been redeployed elsewhere; in May, new APR sold essentially all of its assets, and that deal established the value of our 5% interest at about $60 million. Cash received came to $50.4 million, and the remaining $10 million is subject to a 12-month holdback and appears as a receivable on our balance sheet. Against a carrying value of $7.2 million, the transaction generated a gain of $53.2 million. That included a $13 million zero coupon seller note and $2.82 million in additional consideration. Our edge data center and AI infrastructure strategy is now the sole operating focus of the business; and when I walk through results today, I mean continuing operations unless I say otherwise. I'll now take you through our second quarter 2026 financial performance and highlight the primary drivers behind our business. Total revenue from continuing operations for Q2 2026 increased 30% to $6.18 million, compared with $4.77 million in the second quarter of 2025 as currently presented to exclude the divested rail business. The mix shows what the pivot has accomplished. Technology Solutions generated $3.23 million, our largest revenue line versus a 0 figure a year ago. Related party services and consulting revenue totaled $2.91 million, which included $2.71 million of onetime accelerated recognition tied to the remaining APR deferred revenue. Go-forward AMA revenue will be minimal as it tapers off through the third quarter, and hosting revenue is only beginning to build. For the 6 months ended June 30, 2026, total revenues reached $8.32 million, versus $8.68 million during the same period last year. That modest headline decline is the pivot working as planned. AMA revenue fell $4.2 million amid the wind down, while Technology Solutions contributed $3.8 million starting from nothing. Cost of revenues for Q2 2026 fell 9% to $2.73 million, compared with $2.99 million for Q2 2025. Across the 6 months, cost of revenues dropped 32% to $3.82 million from $5.65 million during the same period last year. That reduction reflects the wind down of the low-margin AMA pass-through cost structure. Gross margin for Q2 2026 climbed 94% to $3.45 million, or 55.8% of revenue, versus $1.78 million, or 37.3%, for Q2 2025. Across the 6 months, gross margin rose 48% to $4.5 million, or 54.1% of revenue, from $3.03 million, or 34.9%, during the same period last year. This improvement in margin quality is structural rather than seasonal. This aligns with how our revenue mix keeps shifting toward technology solutions and infrastructure services. In Q2 2026, operating expenses totaled $3.4 million, up 2% from the $3.32 million posted in Q2 2025. For the 6 months, operating expenses hit $7.63 million, compared with $5.11 million in the same stretch a year ago. That first-half increase comes from deliberate spending — investments we elected to make — such as hiring to support growth, costs associated with operating as a public company, and stock-based compensation, all of which were put in place ahead of the second-half revenue ramp. Operating income in Q2 2026 was approximately $50,000, versus an operating loss of $1.54 million in Q2 2025 — meaningful progress toward profitability as a data center infrastructure company. For the 6 months, our operating loss was $3.13 million, compared with $2.07 million in the same period last year, reflecting the first-quarter investment phase that came before revenue; the increase largely relates to noncash compensation. Including the $53.2 million gain on the sale of investments, income from continuing operations before income taxes was $53.6 million for the quarter. After a $5 million income tax provision, net income from continuing operations was $48.7 million, versus a $1.6 million loss in Q2 2025. The loss from discontinued operations narrowed to $0.8 million, from $1.9 million. Consolidated net income for Q2 2026 was $47.8 million, compared with a consolidated net loss of $3.5 million in Q2 2025. Basic earnings per share from continuing operations were $1.61, and diluted earnings per share were $1.37, versus a loss of $0.14 per share in the same period last year, also from continuing operations. For the 6 months, consolidated net income from continuing operations was $45.6 million, or $1.70 per basic share and $1.41 per diluted share, compared with a net loss of $5.6 million, or $0.21 per share, in the same period last year. As we've noted on earlier earnings calls, returning to positive adjusted EBITDA has been an important milestone for this company. I'm pleased to report we reached it ahead of schedule. Adjusted EBITDA for Q2 2026 was positive, at roughly $0.5 million, excluding the onetime gain on the sale of investments and stock-based compensation. We expect adjusted EBITDA to remain positive through both remaining quarters of 2026, with meaningful expansion in the fourth quarter as GPU deployment accelerates. Now to the balance sheet. The company ended Q2 2026 with $112.3 million in cash, compared with $15.5 million as of December 31, 2025, plus stockholders' equity of $207.4 million. That increase reflects the $50.4 million we received from selling substantially all the assets of New APR Energy, our March public offering, and our $55 million registered direct offering, which closed in June with one large institutional investor; we carry essentially no debt, with nothing borrowed beyond a small insurance financing balance and a $13 million zero coupon seller note, of which $2.82 million remains. Both the seller note tied to Columbus and the GPU senior debt facility we intend to put in place land in the back half on purpose, matched to the assets those funds support. Operating cash flow is the number from this quarter that makes me proudest. Continuing operations delivered $11.9 million of cash from operating activities across the first half, an improvement of roughly $20 million compared with the same period last year. Our customers are the ones footing the bill for the growth we're experiencing. On the balance sheet sits $18.8 million of long-term deferred revenue, a total that takes in the prepayment one customer made under our GPU program. Set against the $50.4 million gain on investment, investing outflows reached $77.1 million: staged growth capital, $68.8 million of deposits connected to our GPU equipment program, and $5.8 million of deposits for real estate and edge data construction — assets set to begin producing revenue in the second half. Once the quarter had closed, we completed the purchase of our Columbus, Georgia data center for $30 million, and the capital was arranged efficiently: $15 million of cash along with a $13 million zero coupon seller note, plus $2.82 million, which is repaid only as additional power arrives at the site. Columbus is being outfitted to support 2,304 NVIDIA B300 GPUs. All told, our financial standing allows us to put capital to work in a meaningful manner behind the growth we anticipate in the second half and beyond. Turning now to our outlook. I want to be clear on our method for guidance. What we guide to is what's contracted, deposited and scheduled — not what's possible. Our outlook builds in no acceleration in GPU energization and nothing from transactions that haven't closed.
When we closed out Q2, our bookings stood at roughly $43.5 million worth of revenue, and every dollar of that should be recognized within the year — that covers contracted backlog along with awards we anticipate in the near term. On top of that, about $1.1 million in contracted Technology Solutions deferred revenue that was recorded in 2025 is going to be recognized as revenue in 2026, which gives the company's performance additional support. Given those committed contracts plus near-term pending orders that are either already performing or on the schedule to be executed across the rest of 2026, we are reconfirming that we expect total revenue in 2026 to be above $50 million. Let me take a moment to briefly explain the bridge from the $6.18 million of Q2 revenue to our full year target. The main driver is our GPU-as-a-Service business, and we look for it to contribute about $26 million as the deployment comes online and utilization ramps in the second half. Beyond that, we anticipate roughly $25 million from our Technology Solutions backlog — a solid base of committed revenue that includes $2.9 million presently recorded as deferred revenue to be recognized in the second half. Our confidence in this outlook holds given the accelerating demand for our edge data center model, continued customer expansions, new hosting deployments and continued capacity build-out. Turning to profitability, we expect positive adjusted EBITDA in both the third and fourth quarters, with fourth quarter adjusted EBITDA falling in the range of $8 million to $10 million. The way to understand 2027 is through our fourth quarter exit rate. In the fourth quarter of 2026, we expect recurring infrastructure revenue — GPU-as-a-Service, colocation and hosting — of approximately $17 million to $18 million, which represents an expected annualized recurring revenue exit run rate in excess of $70 million, contracted under multiyear agreements and carrying gross margins well above 70%. For full year 2027, our early framework calls for total revenues of at least $160 million, a full year of the GPU program, a full year of our contracted colocation deployments now in development and continued growth in Technology Solutions; that framework includes only announced and contracted programs. Additional site acquisitions and partnership structures in our pipeline will be incremental. We expect adjusted EBITDA margins to expand very significantly in 2027 as the recurring revenue base scales against a substantially fixed cost structure, and we will provide formal 2027 guidance with our third quarter results. Six months ago, this company was a rail inspection business with an asset management side agreement. Today, it is a funded effectively debt-free data center infrastructure company with $112 million of cash, positive operating cash flow, 2,304 NVIDIA B300 GPUs being installed in a facility we own and a contracted path to an annualized recurring revenue exit rate above $70 million. The pivot is complete. Now we execute. Doug, I'll turn it back to you for your final comments.
Thanks, Adrian. I said this would be exciting material. All right. Looking forward, our view is that Duos is moving into the next stage of its evolution, supported by a sturdier balance sheet, a leaner operating structure, and building momentum in both Edge and Technology Solutions. Those forces, in our view, leave us well placed to carry our growth path forward through 2027 and past it. With 25 megawatts of contracted capacity set to be deployed in 2026, and with over 75 megawatts now under contract after the Axe Compute expansion, alongside a rising backlog and mounting need for AI infrastructure, our attention stays fixed on execution and on turning the chances before us into recurring revenue over the long term and value for shareholders. My thanks go out to our employees, customers, partners, and shareholders for standing with us. On that note, let's take questions. Operator?
Regarding the quarter and what looks like a bunch of positive announcements. Could we begin with, say, today's news about Axe Compute? Was that 55 megawatts of IT load, or was it gross? And does that come on top of the 10 you previously signed with them? So maybe it's 65 altogether with Axe Compute, or perhaps I've got that wrong? Or is the total 55?
Hi, Brett. This is Doug Recker. I appreciate the questions. Yes, it's gross. That 55 megawatts is gross. So it's on top of the 10 meg we've already signed. So the 10 meg being deployed in Georgia, and then there's another 50 megawatts plus contracted recently, in fact this week.
Regarding where that capacity is headed, have you fellas locked down those locations already? By secured, I mean perhaps pinpointed and funded those spots? Or does that remain sort of undecided?
Yeah. No, here's how we approach it. What we do is head to the market and locate several locations. At present, essentially across our locations, there are six to seven under consideration, with a handful sitting in LOI status. Then we'll select whichever of those moves ahead most quickly. So most of our locations, just so you're aware and it's very — it's helpful for everyone to grasp, none of these sit behind the meter. These are spots where the electricity utility already has power transmission running down. Which means it's already present at the location. There's no need for me to pursue requests with the energy provider to confirm I'm able to obtain it. These are powered locations prepared for immediate deployment.
Alright, fantastic.
And to give you a bit more detail on that, those locations happen to be located in Texas. A handful of them can also be found in South Carolina, Iowa, and Alabama. So the facilities we currently have under LOI and are actively working to acquire already come with electricity service.
So next, let's walk through the joint venture. Is it accurate that they will essentially put up $140 million upfront in exchange for a 49% stake? That $140 million ought to cover the bulk of the data center CapEx. That means you're essentially obtaining a 51% interest in a data center portfolio that's already fully built out via this transaction?
You've got that exactly right. Keep in mind the other essential piece — since everyone will want to run the numbers, I'd like to address that for you here. Keep in mind, as we build, we're currently evaluating our figures and landing below $6 million a megawatt; therefore you can work out that we must put in roughly $30 million for each site.
So when you mention it on a per-site basis for each EDC rollout, is that right?
Exactly. That applies to the 2 sites where we're doing the Axes deployment.
Okay, so regarding this Zero Latency nonbinding term sheet — am I getting this right? These might be the cabinets with lower power density, and you're basically signing a lease for those sites so you can concentrate more on the high-power EDCs?
That's right. That's right. So essentially, what we're talking about here — picture in your head a bare metal provider. They're evaluating every one of our 15 sites. This is a 10-year deal. A 10-year deal; so essentially, if you multiply 15 by 15 cabinets, they're taking up the entire pod.
Yes. Congrats on all the strides you folks have made. My question is this: has anything shifted on the competitive front?
Up to now, our specialty, clearly, as you're able to see, spans anywhere between 1 and 20 meg; that's the space we play in. Demand is beginning to climb sharply. If we walked through our entire funnel, we'd likely be sitting here another 30 minutes. So within that specialty, what we're also noticing is enterprise clients stepping forward saying, look, we need two meg. They're moving away from the enterprise data centers they started with; so what you'll begin to witness is competitors entering the market doing the same thing. I mean, we aren't inventing — this isn't rocket science for us. We found a market and it's on the verge of exploding, and that's visible essentially just from our funnel. So I expect others to show up, but the crucial thing is, do they understand how to deploy modular, correct? We hold the advantage; so we've been deploying modular for the last 9 years. And second, what matters greatly to our business that we must concentrate on is we hold a patent. I'm certain everyone knows about that, the patent named the clean room. When you deploy GPU, they're incredibly sensitive to dust, pollen, all of it. So absent that clean room, there's — nobody is putting $40 million worth of GPU or $100 million worth of GPU into a cluster sitting in a modular environment. So that's among the primary key differentiators we currently have in the market.
That’s excellent to hear. Also, congratulations on all that you folks have accomplished, and I want to wish all of you the best of luck.
Thanks, Ed.
Congratulations on everything you've accomplished. My first question concerns what was said about the growing demand funnel and the contracted backlog. Could you assist us in assessing the size of that? Roughly what period do you believe would be required to work through it?
Sure. The initial orders we placed are cluster #1, followed by the Georgia expansion for the extra 10 meg. That should arrive by the close of the year. We remain on schedule for it; and regarding the second portion we committed to, we anticipate it in the middle of the first quarter. Should everything align perfectly, we would certainly see it at the start of the first quarter, though we're allowing ourselves some buffer there. Therefore, we'll state the end of first quarter, and that's quite robust. Considering it represents under 6 months of delivery, that's robust.
Thanks for that detail. And I apologize if I overlooked this, but can you take us through the markets that are still the most appealing for your company, especially when it comes to growing the portfolio? Have the political headwinds we've seen lately changed the approach in any way? My guess is not in a meaningful way, considering how big the sites are that you're bringing online.
Yeah, that's exactly correct. That's the other secret sauce, as I mentioned. We're under 20 meg. So whenever you enter a market that's under 20 meg, you're really not hitting the radar, correct? So our focus is on what we call stranded power. So there's a lot in South Carolina. There's a lot in Iowa. There's a lot in markets where pricing is still around $0.04 to $0.07 a kilowatt hour; so our focus is on those markets where we find the stranded power, or the second strategy is there are bit miners out there that weren't successful that actually had powered sites, and we're looking to take a lot of those over. And those are in the range of 5 to 20 megawatts. Remember, we didn't disclose our funnel, but in our funnel, we have a ton of use cases for two to seven meg. So we're going to start hitting those as well.
Apologies for that, everyone. To start, I'll add my congratulations to the ones you've already gotten. Doug, I'm interested — can you elaborate a bit more on what's in the pipeline across the Tier 3 and Tier 4 markets, past that 25 megawatts plus the 55 for Axe? Going forward, what's the main thing holding things back? Is it power? Is it capital? Is it customers?
Excellent question. Excellent question. What we're seeing is that Tier 3 and Tier 4 markets suit us perfectly; we remain inside 130 miles; we remain below five milliseconds wherever we operate. Fiber is still obtainable. Getting fiber into these sites requires only short builds. But the real thing we notice is that's where the stranded power sits, correct? That's where there's electricity we can use immediately. As for the market and backlog, I'd say — I can tell you confidently that just within our local funnel, in-house here, not counting inbound calls, which is a separate log, we're well past 100 megawatts across 5 to 10 meg tranches in our funnel for this year alone. So clearly, we can't achieve it this year. But these customers are reaching out asking, hey, can you deliver it within the next 6 or even 9 months. And those are the ones we're prioritizing right now. Because what matters most in this business is deploying fast, but you want to do it correctly. Make one mistake in this business and you're finished. And I've been doing this 30 years. And let me tell you, you must do it correctly because these customers depend on your power 24 hours a day, and it's their livelihood and it's our livelihood; so we're doing it correctly. We're not tossing a bunch of stuff against the wall here; we're doing it correctly, and we're targeting the right customers, which we — when we say right customers, they're Tier 1 credit. They're highly reputable companies. They've existed for a long time. So we're now in a position where we can be selective about who we take, which is great because in this model here, you've got to be cautious, correct? We've got to be cautious about who we sign, and we're very diverse across our customer base. So no single customer is consuming all our bandwidth here.
Great. And Doug, regarding how quickly you can deploy, in Columbus, what was the real timeline starting when the site was chosen until customers were actually getting billed?
Yes. This is going to astonish you. Last week on Thursday, we actually had a customer come take a look, and they were astonished. Honestly, I was astonished as well. So at the start of July, we acquired this building; and just consider that, at the start of July, and right now it's, what, the 17th, 18th of August; we've put in more than 7 megawatts of new infrastructure. By new infrastructure, I mean you take a building that's only the shell. You bypass everything inside that building other than the fiber build-out, and we installed everything entirely fresh, all of it from generators to cooling plant right through to PDUs to cabinets to all of it, lighting, raised floor, we installed 17,000 square foot of raised floor. We completed all of that in under 60 days.
Wow, okay.
I challenge anybody to do that.
And if I could fit in just one more for Adrian. During your prepared remarks, you brought up a few of the larger expenses from the first half of the year. I'd like to know whether we ought to treat that increase as a one-time event. And once we get into '27, should we expect OpEx growth to be somewhat quieter? Would that be right?
Without a doubt. You nailed it. We've just finished an extremely complicated stretch, roughly the opening two quarters of the year, when, clearly, the APR sale, the rail business divestment, involved a great deal of intricate accounting. And then there was the change from adjusting the staff and all the rest. So every one of those costs sitting in there is very much one-time by nature. We were around — we wrapped up last year while APR was still with us, and we had roughly 100 people. And today, following the rail business divestment, we're at approximately 25 full-time people. Sorry, I just got disconnected there. What I meant was that SG&A is now increasing very, very gradually. And those costs will truly be tied more to particular opportunities heading that way. But we've done — Doug and I have been focused on ensuring we streamline our operations. So yes, very much one-time.
Doug, I've got a few things to ask. On that Axe transaction — is there anything you're able to share regarding, I'm guessing, the identity of the final customer? And I realize you mentioned a $140 million deposit coming from them. If they're handing over that cash, my assumption is they're a fairly strong credit customer. But can you provide any detail about who that end customer is?
Sure thing. They're a Tier 1 hyperscaler. They came to the site in person last week. We're not able to reveal who they are, but what I can say is that they are exceptionally credible and creditworthy. In fact, our GPUs were funded on the basis of that customer, so they're very solid. Once again, though, this fits within the Axe deal as well. So we wanted to confirm, obviously, as we enter this SPV, that firstly, the money was there. And secondly, whether this customer is credit worthy and whether they'll remain for 5 years, correct? The term is 5 years. So we performed our due diligence, and we feel very strongly about it. But Justin, I also want to mention that Dipan is here with me in the room, and I'd like him to quickly walk you through what this SPV is, just so it's clear, because I know I'll receive a ton of calls. And I still welcome the calls, but this may help everyone understand what that SPV is. So Dipan, go ahead and give Justin a quick background on what that SPV looks like.
Appreciate that, Doug. Hey there, Justin. Good morning, everybody. This SPV serves as the mechanism through which we'll finance building out those data centers. Even though we're going to act as managing partner of the SPV and hold a majority stake, the cash infusion is coming our way from Axe in exchange for equity. After that, we'll build these data centers out, backed by a lease from Axe plus their offtaker.
Understood. And perhaps one more question. The transactions you've completed, I believe, have run 5 years, based on what I see among some competitors who, as I understand it, are committing to larger agreements, and I recognize that your market differs and that you pursue, as you mentioned, deals below 20 megawatts. So what is the lifespan of your pods? And how do you approach, I suppose, the duration of these agreements? Since the payback period looks fairly appealing up to now, at least regarding the 5-year contracts you've executed here.
Right, Justin, the useful life of our pods and our facilities extends considerably beyond 20 years. The sole component we would replace after 10 to 12 years is the batteries within the UPS system; however, the infrastructure as such lasts considerably beyond 20 years.
All right. That sounds good. Perhaps one final question from my side. If I've calculated correctly, then taking into account this SPV, the cash sitting on your balance sheet, and what I believe you mentioned about possibly needing around $30 million per site for this new deal, it seems like you'd largely be covered for these deals. That said, could you discuss the path ahead and funding — how do you approach, I suppose, financing the extra 100 megawatts you referenced within the pipeline? How are you thinking about financing down the road?
Sure. Essentially, another thing this accomplishes — and the reason I opted to pursue this deal — is that it clearly provides us with the money to purchase infrastructure shortly for these projects, ensuring we reach it by the first quarter. That means we must place those orders at this point. Beyond that, it will generate revenue that we're able to borrow against. So in essence, debt financing will come after this. Diluting our investors isn't why we're in this business. Thus, choosing this type of financing, or this partnership, was a sound move for us. Now, whenever I seek capital, I've got that solid — this SPV — which we can borrow against. So I essentially have this infrastructure, totaling $140 million, that I can borrow against, first of all. Secondly, it lets me purchase time at present to get my revenue going, as we discussed, from the GPU and all of that. Therefore, that $4.4 million will begin shortly, and after that, it will provide us with greater credibility and additional money on our balance sheet.
Each month.
Each month, every month. So, is there anything I overlooked...
No, no.
Sorry, Justin — did I manage to cover everything?
Can you hear me? Have I actually joined?
You have, indeed.
This time, please keep me on the line. I'm hoping the connection holds up; it's truly impressive. I believe you aren't receiving sufficient praise, since you accomplished something that rarely occurs — you stated an intention and then followed through. What you've got going, all these wonderful developments, is honestly remarkable, and I find it genuinely thrilling that, in my view, the bulk of the hardest portion now lies in the past. Yet I keep encountering situations where confusion appears to persist. What I'd like to do is pose a question that recasts this so my brain can grasp it far more simply: rather than a data center company, picture us as essentially a business that constructs apartments and leases them out. That means this operation has two pieces. As you put it, our model — you describe it as our model, and that's how I interpret it — was that we used to do this at $1 million per unit. Then all at once, a unit required $30 million. That capital wasn't available to us, and no cash was flowing in, so debt was out of reach, which was absolutely the correct decision, because doing otherwise would have created an opening to damage shareholders. And by the way, I'm grateful you did the right thing for us. But that spot between a rock and a hard place — has it passed? Obviously, were a huge opportunity to arise and dilution appeared to be the proper course, I wouldn't demand that you declare yourself forever finished with it, yet I sense you've been punished disproportionately on the short side; and perhaps you're unaware, but more than 5 million shares of your company are held short; and I'm fairly confident that's due to your not having turned a profit and your having raised capital repeatedly. So clarifying this matters a great deal; and then I'm certain it will occur tomorrow, because you just reported $1.61 for the quarter, stemming from a onetime one-off sale. And given how much trading is driven by algorithmic activity, we could witness some goofy behavior tomorrow. And the goofy part is that you deserve it regardless, but it may stem from the sale that occurred; so expect the unexpected, I'd say. But what I'm really asking is whether this is simply the situation where we needed a couple of these apartments constructed so we could secure a tenant we'd already locked in, under a contract, to begin renting the space from us. And once we combine the tangible asset — we can HELOC it just as you would with your own home — then we have this contractually obligated high-margin revenue that will cover the cost of capital for using debt on new projects. So that's the situation we've been in — that's how I've come to understand the model. I believe it's much simpler to reason about this since, unlike an apartment, if things got tough and our tenants disappeared, well, I don't figure the spot matters all that much, but we could hoist it with a crane and relocate it somewhere — somewhere more advantageous, correct?
Yes. You're answering my questions; but yes, so when we enter a market and say I put in 10 megawatt, suppose that customer leaves after 3 years, right? First off, my infrastructure is nearly fully covered by them. Second, the power is so sought after. At this moment, if I had 20 megawatt today, if I had 10 megawatt, if I had 4 megawatt available right now, I've got roughly 16 customers waiting in line to grab it. That's how robust this market is at present. I'm revealing my secret sauce here, but I'm just being straightforward with you. So the other wild thing is, if you consider it, Nico, if you consider the customers sitting in enterprise data centers today, and this was the presentation I gave last week. If you consider the enterprise customers who are actually in a data center who are accustomed to taking a 20 kW cabinet or a 15 kW cabinet, now they're swapping their components over to AI. Now suddenly, they require 60 kW. Those data centers, those legacy enterprise data centers are unable to supply the cooling, not merely because they can't obtain the power, but they lack space for the generators. They're landlocked. They can't make that happen. So Fortune 100 companies are phoning us, asking us, could you deploy a meg with us? Could you deploy two meg with us? That's where your market is going to shift. So we're concentrating on these Neo clouds at the moment. But I'll tell you what, our market has just flung wide open to that sector. That sector, I adore because the data centers that I've previously owned and sold, even the one just down the road, Colo5, when I sold that to Cologix, those customers are growing. They've got nowhere to go. So where will they head? They'll head to where the network exists and the power exists in a dependable data center. And it truly doesn't matter whether they're 130 or 140 miles outside of town since the networks are constructed more robustly these days.
So, the original approach — the standard unit, that wholesale model — the thinking was that having a lower-powered unit situated near a school or hospital was preferable from a latency perspective. And that shifted — to be clear, the Iowa project amounts to contracted revenue tied to that output power, and it simply got relocated to Georgia, since the infrastructure existed there to start generating cash flow sooner. That was the entire situation. To be clear, it isn't as though Iowa was abandoned; we still hold a parcel of land there. We were simply able to prepare the 10 megawatts for them at that location. Then you mentioned there are an additional 10 that were contracted within that same building — is that right?
Yes, that's right — we're adding another 10 megawatts onto the building that already exists.
And that's the 5-year 111?
Yes.
What's the thinking behind 10 megawatts priced at 176 across a 3-year term, while another 10 megawatts goes at 111 over 5 years? Are these somehow distinct arrangements?
In the initial case, GPU capacity is part of the package. Keep in mind, our earliest rollout at that site includes the GPU component. For the subsequent rollout, the GPU isn't ours. That deal is pure Colo, which is precisely the business we operate in. That's the approach we follow.
I’m grateful to each of you for being here. Just keep in mind that reaching out to me is always an option—drop me an email. If anything comes up, I’d be glad to discuss it; thanks so much for taking part in today’s call. I value all of you, and we’ll be in touch before long. Thanks again.
As we wrap up this session, I want to share Duos' safe harbor statement, which covers key warnings about forward-looking statements delivered on this call. This earnings call includes forward-looking statements as defined under the Private Securities Litigation Reform Act of 1995. Terms like believes, expects, may, will, should, anticipates, plans, along with their negatives or comparable wording, are meant to flag forward-looking statements. Be advised that such statements do not promise future results or occurrences and are exposed to various uncertainties, risks, and other forces, a great many of which lie outside our control; these could affect how accurate the statements and the projections behind them turn out to be, and could lead Duos Technologies Group Inc.'s actual results to differ materially from what those forward-looking statements anticipate. Among these risks and uncertainties are, without limitation, the ones set out under Item 1A in Duos's annual report on Form 10-K, which is expressly incorporated herein by reference, plus other factors that may be described from time to time in Duo's filings with the SEC. We appreciate your participation today in Duos Technologies Group's Second Quarter 2026 Earnings Call. You may now disconnect.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Duos Technologies Group, Inc. transcript - plus 131,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Duos Technologies Group, Inc. earnings transcripts and 131,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.