Operator: Thank you for standing by. This is the conference operator. Welcome to the Gorilla Technology Group, Inc. NASDAQ GRRR First Quarter (sic) [ First Half ] 2026 Financial Results Conference Call. [Operator Instructions] The conference is being recorded. Before we begin, we would like to read the forward-looking statements. Today's call includes forward-looking statements made pursuant to the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. These statements reflect management's current expectations and projections about future events and are subject to known and unknown risks and uncertainties that could cause actual results to differ materially. Forward-looking statements often include terms such as expects, believes, plans, anticipates, may, should and similar expressions. For a discussion of important factors that could affect Gorilla's results, please refer to our filings with the SEC, including the most recent annual report on Form 20-F. Except as required by law, Gorilla undertakes no obligation to update or revise any forward-looking statements made on this call, whether as a result of new information, future events or otherwise. I would now like to turn the conference over to Jay Chandan, Chairman and Chief Executive Officer; and Bruce Bower, Chief Financial Officer. Please go ahead.

Jayesh Chandan: Thank you very much. Good afternoon, everyone, and thank you for joining us. Now the first half of 2026 marks, for me, a very decisive step forward for Gorilla. The revenue increased 99%, nearly 100% year-on-year to $78.4 million, effectively doubling in the first 12 months. But more importantly, the momentum strengthened as the half progressed. Now Q2 revenues reached well over $50.1 million, which was a net increase of roughly 78% from Q1 and 138% from Q2 last year. Now we had originally expected, as we had promised to the market, about $33 million, which we upgraded to $44 million. We've exceeded that by another $6.1 million to nearly by 14%, which principally means that all the deliverables and certain milestones were completed earlier than anticipated. Personally, that's what execution looks like. Now the challenge we've also had is that the operating performance and progression at the same time was also equally significant. Our reported operating loss narrowed from $41.1 million in Q1 to approximately $2.2 million in Q2, which was a reduction of 95%. A substantial part of the first quarter result was share-based compensation that has already been recognized and more than 80% of the H1 share-based compensation was absorbed in the Q1 itself, and the quarterly charge declined by approximately, what, about 78% in Q2. In plain, simple English, Q1 carried the overwhelming majority of the burden. Q2 showed a much clearer picture of the operating momentum beneath it. Now our cash efficiency also improved considerably. Whilst the revenue increased by approximately 100%, operating cash consumption declined by approximately 65% from $12.5 million in H1 of 2025 to $4.3 million in H1 of 2026. Operating cash usage also fell from 31.8% of the revenue to just 5.5%. Now these are very material important numbers, which we need to take into consideration. Now the company also recorded an overall increase in cash of approximately $79.8 million during the first half, principally reflecting financing support and the expansion program together with customer collections. So we ended June with roughly around $179.4 million in cash, approximately 82% above our Q1 closing balance. That capital is not just sitting there for like in a bank and trying to get some interest rate, it is there to be deployed. What we are doing is that we are purchasing infrastructure, securing capacity, preparing sites, building teams and funding the deposits and working capital required to deliver projects of a scale Gorilla has never previously undertaken. And just FYI, we are preparing currently about 5 different sites in parallel, and that takes a humongous effort. At the same time, we also understand that the cash balances will move between the reporting periods. Investors should distinguish between cash being consumed by an underperforming operation and capital being deliberately deployed into contracted projects and revenue-generating infrastructure. They're not remotely the same thing, however convenient it may be for some people to pretend otherwise. Now this investment phase also explains the current gross margin profile as well. Now our gross margin had dipped, reflecting a revenue mix weighted towards hardware, initial deployment and project mobilization. What I need to make sure is that Gorilla also deployed more than $14.1 million into property and equipment. Currently, that number is $29.4 million. We are also building the installed base first. As the infrastructure is commissioned, customer workloads migrate, utilization increases, and we expect the revenue mix to broaden towards compute, monitoring, managed services and all other associated services. Hardware, personally, guys, does not begin producing its full financial results the moment it leaves the factory. It must be delivered. It has to be installed. It has to be powered. It has to be tested. It has to be accepted by the customer. And more importantly, then the utilization happens. More importantly, we want to make sure that we are moving very, very quickly. Now in terms of updates, I think the market has been asking me for updates for a long time. For Yotta Phase 1, for example, the testing has been completed. The equipment deliveries are underway and deployment have commenced. Testing will commence by the end of this week, early next week. Yotta Phase 2, the equipment is currently being manufactured with completion expected over the next 25 to 30 days. In Indonesia and Batam, we're working very closely with our OEM and infrastructure partners. We have, as everybody knows, secured substantial data center space, and we're targeting approximately another 200 megawatts of capacity with an initial ready for service in the middle of 2027 and the broader deployment expected to be in the second half of 2027. At Korat in Thailand, the land has been cleared. We're advancing with the financing, the GPU procurement, the infrastructure requirements while engaging with prospective offtakers with the objective of moving into discussion into firm customer contracts. Now to be absolutely clear, megawatts are not just capacity -- they're not just capacity, they are revenue. More importantly, the capacity must be commissioned. It has to be contracted. It has to be utilized. Now our accountants remain very stubbornly unwilling to accept that electricity as a payment. So unfortunately, we have to make sure that the customers pay us at any given point of time. Alongside these major infrastructure programs, our established security and network intelligence operations remain an important part of Gorilla. They provide not just the customer relationships, they also provide a better delivery experience, cash collections that support our broader expansion. Now for people thinking that we are pivoting, we're not. We're not abandoning the business that brought us here. We are using it as a platform to build something substantially larger. Now for Q3, we are planning revenues between $48 million to $50 million compared to the previous plan of $36 million to $40 million. That represents a significant jump of about 20% to 39% than the earlier planning range. For Q4, our operating plan is revenues exceeding well over $60 million to $70 million. Taken together with the H1 revenue of $78.4 million, our revenue outlook for 2026 now stands at least $200 million, which is up from the $137 million to $200 million range we provided at the beginning of this year. Now reaching the upper end requires additional execution, including further deliveries, customer and workload activation. We intend to earn the upper end. I want to make sure that we're not simply announcing this, but we want to make it more and more useful as we go through the quarters. Now looking further ahead into 2027, we're targeting revenues of about $450 million to $500 million. Now that's an ambitious target that represents a quarterly revenue of roughly over 120 -- $112 million to $125 million. Now we are not expecting the calendars to produce the growth for us. The target depends on all of the capacity that's being installed now, the commissioning of the additional projects, the conversion of the prospective demand into contracts and the migration of utilization of the customer workloads. Now there has been no shortage of personally barking from the sidelines. Unfortunately, that does not deliver GPUs for us or neither does it commission data centers or collect dollars from our customers. Our answer to all this would be execution. Now Gorilla has entered the second half with substantially greater revenue scale, dramatically improved quarterly performance, stronger liquidity and a growing portfolio of major international projects. We have more work to do. We are maintaining absolute delivery discipline. We're managing capital very carefully. We're improving utilization, converting opportunity into recognized revenue. And make no mistake, the direction of the travel now is unmistakable. So we're no longer explaining what Gorilla intends to become. We're beginning to demonstrate it. Thank you very much. Bruce, over to you.

Bruce Bower: Thank you, Jay. I think Jay hit on all of the highlights, but there are a couple of areas I want to expand on or emphasize. So the first is, of course, the first half revenue of $78 million, 99% revenue growth. As you can hear from the guidance figures, Gorilla is in hyper growth mode in terms of revenue. So we're happy with these figures, and we expect to see this kind of growth trajectory continue. Also, as Jay mentioned, the gross profit for the first half showed an investment into the business and also reflected a mix. As the mix improves, we expect to see an expansion of gross margins. The mix will improve in a couple of ways. The first is as Yotta 1 and the other GPU-as-a-Service projects go live in the second half of the year. Those projects generate gross margins of 75% or more. So of course, there will be an expansion in the overall gross margin of the business as a result. And then that will flow through to the operating -- there will be significant operating leverage coming from that as well. So that will flow through to the underlying EBITDA numbers as well. We mentioned some of the expense items. I would just like to highlight that this was not a normal first half. There were significant for -- there were foreign exchange -- significant foreign exchange movements in the markets due to the Iran conflict and [ Freedom Day ] tariffs and tariff wars even as we have a significant exposure in foreign currency in the Middle East and in Southeast Asia, which were the 2 regions hardest hit by geopolitics and by tariffs. This did produce significant volatility in our underlying numbers. We expect that as the situation seems to have calmed down that, that will also result in a calming down of some of the below-the-line expense items that we incurred in the first half. Also, as Jay mentioned, there was a stock-based compensation item. This is something that had been previously disclosed in all of our filings, which should come as no surprise to somebody. And then my attitude would be that this is out of the way. So again, the second half P&L should be quite clean from that perspective. A couple of things that I want to highlight from the balance sheet side. One is that we finished the quarter with the second -- the first half going into the second half with $175 million (sic) [ $179.4 million ] of cash. We had $13 million of conventional debt, sort of traditional debt bank loans and then $107 million of long-term debt from a 5-year convertible that we placed in June. We also did a convertible in July. That gives us really enough cash that we are going on offense. This is the time where we are paying for equipment and for deliveries of equipment and setting them up, and that will be converting into revenue in third and then substantially in the fourth quarter. We have paid for the items for Yotta 1 for the first deployment with Yotta out of our own balance sheet, our own funds and then a facility, a small facility that is tied directly to the project level. And then Yotta 2, we have paid again substantial prepayments out of our own balance sheet, and then we intend to fund the balance of the payments from a larger project finance facility, where I will share more details when we can. But I think everyone on the line will be very happy with the terms that we've managed to get. And then a couple of other things. When I talk about investment, you can see already, as Jay mentioned, in the CapEx investment. So $14 million of investment went into the first half. That number is, of course, going to be many, many multiples of that in the second half. But this is where the investment in the business is going to be showing up in the future, and that leads to revenue growth and to margin expansion. In spite of the large CapEx, I would say that the business is actually performing well on the cash flow front. Operating cash flow was minus $4 million in the first half. This was a tightening from minus $15 million (sic) [ $12.5 million ] in the first half of last year. I would expect to see the operating cash flow numbers continue to improve. We're expecting large customer collections in September and October to the tune of over $20 million. And then the SG&A and everything else from the -- all the other operating costs in the business will not be overwhelmed by that. So that should result in a breakeven or operating cash flow for the entire year. And last but not least, I want to remind you, we have the guidance figures in the press release, $200 million plus is our guidance for this year, $450 million to $500 million is the revenue guidance for next year. How we make guidance is we take what is contracted revenue where we have an amount and a date. If we have a contract, but maybe the timing is not exactly firmed up or the amounts aren't exactly firmed up, we don't include it in the guidance. So again, we try to be underpromising and overdelivering to the market. We try to trade only on -- based on what will not disappoint you. So when Jay talks about the delivery schedules for Yotta 1 and Yotta 2 and then going out to the project in Batam or NeutraDC, as we call it, where we have the delivery schedule firm, that has been included in guidance. Where we do not have the delivery schedules firm, that has yet to be included in guidance, but stay on the lookout as those delivery schedules firm up, then the guidance might change as a result. We don't provide a gross margin forecast for next year at the moment. That will depend really on the timing of when these projects initiate. But I would just remind people that Yotta 1, Yotta 2 and NeutraDC, we expect on those projects an average gross margin of 75%. So if we're talking about $450 million to $500 million of revenue, then the gross margin on that would be substantial. The majority of that would be coming from GPU-as-a-Service. So I would expect you would see gross margins in the sort of 40% plus range for next year. Again, that is not official guidance, but that's just working out the figures. In terms of a financing update, I mentioned -- so we have an offer from -- already on the table for financing for NeutraDC. We also have -- we also have used the proceeds from the convertible for the initial part of that. For Yotta 1, we've used -- and Yotta 2, we've used substantially all the proceeds from the convertible to pay for prepayments or deliveries. And then we're funding the balance of the deliveries from 2 different project finance facilities that we either are arranging or have arranged. So again, I think you'll see for the future growth and the future CapEx of the business, you'll see more of a focus on project finance or debt overall, and we'll have more details in the coming days about how that works. With that, I turn it back to Jay.

Jayesh Chandan: Thank you very much, Bruce. We're now open for questions.

Operator: [Operator Instructions] And the first question comes from Brian Kinstlinger with Alliance Global Partners.

Brian Kinstlinger: A lot of progress and a lot of things to talk about. But Bruce, you mentioned guidance, at least for next year and any year is based on what's in hand where you know how much is coming in and what projects and when. So for 2027, what is on the delivery schedule, which phases will be active? And which programs? Is it all of the first 2 phases of Yotta and the first of NeutraDC? Or just maybe take us through what is assumed versus what is not assumed?

Bruce Bower: Yes. So the first phase for the Yotta contract, we have delivery schedule and we're taking deliveries now of the equipment. So that is assumed. That is $100 million annualized roughly of incremental revenue. The -- then for the Yotta 2 contract, as we call it, there -- it's split into 2 different batches. The first batch, we have the delivery schedule. So that is assumed. That is $250 million of incremental revenue. And then the last is for the NeutraDC project, we mentioned in the initial press release that there was a deployment expected of 300 servers around the September time frame. And then by the end of the year, a subsequent 700 servers. The 300 servers, the delivery schedule has firmed up. So that is included in the guidance. That would be 30%, about $260 million. So that would be an incremental $75 million to $80 million a year. So -- and then the balance comes from existing contracts or contracts that we've won and not yet announced, and that's how you get to the $450 million to $500 million mark. So there is the 700 servers from the first phase of NeutraDC and which are not included in the guidance. The second phase of NeutraDC, which is 875 servers, correct me if I'm wrong, is also not included in the guidance. And then the second phase -- part 2 of the second phase of Yotta is not included in the guidance at the moment. So obviously, we want to get the delivery schedule firmed up. GPUs are like spice in Dune, they're very valuable commodities. So it's better to have them in hand before we start talking about schedules and timing and amounts. So again, this is why we take the ultra-cautious approach in formulating guidance.

Brian Kinstlinger: Great. And then a follow-up -- of course...

Jayesh Chandan: Sorry, Brian. See, for us, the confidence is from the capacity and the projects we're putting now. Like I think Bruce eloquently said, right, we have delivery scheduled for Yotta 1. In fact, there are a whole bunch of them on the plane right now as we speak. They're underway. We are commencing all our deployment, which effectively means the power drops are there, the connectivity is there. And the Yotta Phase 2 equipment is also being completed and will be approximately delivered. The deliveries will start coming in. So we have a fixed time line. That means we have to deliver the first set by the end of this month, the second set mid of next month. And then -- so between now and end of November, Yotta 1 and Yotta 2 Phase 1 will be complete. And then you've got the big elephant, which is the NeutraDC, that particular one actually -- the first 300 servers have to be deployed by the end of October. Testing will take probably between 20 and 30 days. So we're looking at going operational end of November, first week of January -- December, sorry. Now on top of that, we were also working towards closing the remainder of the 700 plus the 875 servers, which will obviously -- once that is done, we will absolutely revise the targets for next year. But again, the challenge we have today, just to address what the challenges are, the GPUs are in shortfall. The CPUs are in shortfall. Memory and storage is short. Then you've got the cabling and everything else. On top of that, electricity seems to be of a bit of a problem. But what we are doing is we're making sure that whatever we have committed today is based on the operational milestones, which we already have in place, whether it's equipment, capacity, customer contracts signed, workloads migrated and all of the utilization that is being done.

Brian Kinstlinger: Great. That's super helpful. A follow-up. I think the NeutraDC, those are data centers that were already built, if my memory serves me. But on Yotta, do you have any construction you have to do?

Jayesh Chandan: No, none whatsoever. Yotta is a phenomenal constructor. They build it themselves. So they have done it all by themselves. And by the way, all of the Yotta data centers are fully ready for service. All the floors are ready for service. All we're doing now is getting the power drops in and getting the GPUs to be tested.

Brian Kinstlinger: Great. And then on the project financing, last quarter, you had mentioned you had offers and you made similar offers on the table and you had similar comments here. What has been the biggest obstacle in securing the project financing?

Bruce Bower: Brian, I'd be very careful. I didn't say we had offers. I said, in one case, we have arranged and in the other case, we are close to completing. The obstacles has been, first of all, us fighting for the best terms possible. There were some initial offers that were not conducive to shareholders. So we said, no, we're not going to accept that. And then the second thing has been more recently when there has been a shift in some of the delivery schedules. The shift actually was one where they wanted us to pay. The vendors wanted us to pay more quickly. So we had to make people hurry up and meet different delivery schedules and thus we do something. But in general, I'm very happy with where we are. I'm very happy with where -- not just with the project finance story, but also with how we are with the debt markets overall. And then I would say, we said earlier that we are pursuing a credit rating. So I'd stay tuned on that front.

Operator: Your next question comes from the line of Alex Latimore with Northland Capital Markets.

Alexander Latimore: Glad to see everything evolving well here. I was curious about the terminal value of the GPUs. Are you looking to sell them after 5 years? If so, what would that residual value be? And then also alternatively, is there an opportunity to keep operating the GPUs for a sixth year?

Jayesh Chandan: That's a great question. Alex, good to hear from you. I hope all is well. So yes, so the current -- if you look at various sources today, B300s and GB300s, they still have residual value at the end of the fifth year. You're looking at roughly around between 20% and 25%. But that's today, Alex. We don't know what's going to happen in 5 years' time. But at the end of the day, we will continue to operate those GPUs. We will also -- we've already received offers, just FYI, from various either institutional investors or from data center operators saying, look, I'm happy to sign an agreement with you at the end of the fifth year, we'll take it off and here's -- we will get it evaluated by a top-tier firm like Ernst & Young or PwC or KPMG, and then we will put an assigned value to it. But the Gorilla's intent is to continue working. As the models evolve, you're evolving from training workloads to inferences, edge will become a meaningful kind of -- will make a meaningful entry towards the beginning of next year. We are looking at deploying at scale in these regions. The other things you should also look at is sovereign AI. Each of these countries where we are present today, whether it's the Middle East, parts of Europe, parts of Asia, they are very actively sourcing GPUs. They're looking to secure the future, especially the governments and so on and so forth, and we can deploy it at that point of time. So net-net, either it's revenue-based incremental working towards the next -- I don't know, the next 3 or 4 years after or there's an immediate liquidity for sale to an existing data center provider.

Bruce Bower: Alex, if I may add to that. So the -- for accounting purposes, we'll depreciate the equipment over 5 years. So that's the assumed life just -- the major reason is that some of the contracts -- most of the contracts we're signing for 5 years. But I would emphasize that, first of all, what Jay mentioned that there is a couple of ways to monetize them later. I would also point out that this has been a story that we've heard discussed numerous times, what is the useful life of the GPU. Some other players in the market are putting A100s into service for 7, 8, 9 years running now and still generating revenue and a very decent yield on cost. So obviously, the A100s don't rent for the same thing that they would 5 years ago, but it's still a very healthy return and very favorable economics. So all the evidence is pointing to the fact that the service life might actually be much longer than 5 years.

Alexander Latimore: Awesome. Another question. For your CapEx forecast for India and Indonesia, does that include maintenance CapEx? Or is that additional? And if it is, how should we think about the sequencing or the cadence of that maintenance CapEx?

Bruce Bower: Yes, I'll take that. So the CapEx forecast we've given is for the upfront CapEx. The upfront CapEx really covers networking equipment, GPUs, servers, cabling, setup, everything that you would need to get up and running. The maintenance CapEx, we are taking as an expense. There are a couple of reasons for that. The first is that most of it will be labor actually. So we're doing an RMA service where we're -- we have people that are going to be servicing the equipment and then also repairing it. And then secondly, a lot of the equipment will be covered by warranty by the vendor. So we will spend a little bit to repair or to swap out spare parts, et cetera. But the big CapEx that would be needed in case a machine -- an overall server breaks would not be with us. It would sit with the vendor. So basically, when you look at the forecast and our gross margin, we put in the gross margin, all the cost of spare parts and labor. It's not broken out separately as maintenance CapEx.

Alexander Latimore: Awesome. One more. Could you describe the debt financing for Indonesia? Just a few questions to run through. What is the interest rate? Is the customer financing portions of it? If so, how much? And then what percent of the financings are complete there?

Bruce Bower: We have an offer that we disclosed in the press release about the project for 70% of the project. The balance will come from the Gorilla balance sheet and then from customer prepayments. So so far, how it works is if we're deploying 30% of the project upfront, so 300,000 (sic) [ 300 ] servers, then that comes out of customer prepayments and out of Gorilla's pocket. And then the debt portion would fund the 700 servers to come at the end of the year. So that's what I can share about the financing arrangements for that project. So the first deliveries are coming basically out of Gorilla's pocket and then drawing on customer prepayment.

Operator: Our next question comes from the line of John Roy with Water Tower Research.

John Marc Roy: Congratulations, gentlemen. I wanted to maybe take a step back real quick and think about what could go wrong next year. What do you think is your biggest risk? Is it people? Is it power? Is it building facilities? Is it acquiring equipment? What are you most worried about?

Jayesh Chandan: Great question, John. Good to hear from you again. So let me classify that into 3 principal risks, okay? One is the hardware timing. The second one is the -- I would categorize site and power readiness. And the third one is customer acceptance with workload migration. Now we've been ordering early. Just for FYI, all of the equipment, which are supposed to be delivered in end of September have been delivered now. So it's about 5 weeks early. Yotta 2, it's about 8 weeks early in terms of manufacturing and so on. So we're ordering early, testing before the deployment, taking deliveries with our OEM and our integration partners. Then more importantly, we're making sure all the sequencing matters. Second, we're gating the deployment against what I call power -- confirmed power and readiness, site readiness. Engineering is working actively 24/7 on this. Networking and installation work streams are running in parallel. Our teams are sitting on the sites in different parts of the world. Now unfortunately, electricity has an inconvenient habit of being very essential. So on top of that, customer testing and workload migration has to begin before full commission. Now that allows us to resolve all of the integration issues progressively. So what we are doing is we're making sure that the paying workload works better. So we have scheduled buffers in each of these phases, phased deployment plans. And then more importantly, the ability to resequence work where appropriate. We also have to build our own internal processes. You will see that we have actually gone -- we've almost doubled our size in terms of human resources. And we're also making sure that there's contingency plans, which is both for the operational preparation as well as all of the hardware as well. So all of the RMAs, all of the RFSs, all of them have to be done well before the schedule. So that is something we believe are some of the important risks we are looking at in H2 ramp for ourselves. But we do have contingencies for every single one of them.

Operator: [Operator Instructions] Your next question comes from the line of Bharath Nagaraj with Cantor Fitzgerald.

Bharath Nagaraj: Previously, you referenced hiring across -- aggressive hiring, I should say, across Thailand, India, Taiwan, et cetera, targeting 1,000-plus people, I think, Thailand alone maybe. What's the current headcount? And what's the monthly OpEx run rate exiting Q2 at this stage? That's the first one.

Jayesh Chandan: So good to hear from you again. So we have, I think, on a full-time resource basis, I think we're about 300-plus people. On a contractor basis, we're already at about 300, 350. We're going through the -- as you can imagine, we're going through various phases now. We are going through land assessment, power readiness, water readiness, EPC certifications and so on and so forth, both in Thailand and in Indonesia. That will involve probably another 400 to 500 people, but we are going through the whole RFP process right now as we speak. As we are going forward, we would see that number specifically expand significantly. So I can tell you at the peak, let's say, by mid-2027, we'll be at about 2,000 people. Currently, we're at about 300 plus another 350, so about 650 people. So yes, we are expanding rapidly. Bruce, do you want to take the second half?

Bruce Bower: Yes. It's about $2.7 million a month is the SG&A. So that is more than just the people costs. We have -- we expect that to go up over -- by the end of next year to sort of $38 million to $40 million annualized range. Now as Jay mentioned, there's a significant headcount expansion, but just keep in mind the geographies, right, India, Thailand, et cetera, not -- it's not Silicon Valley, right? So that explains the -- why the cost increase would not be so large. And then we think that, that's -- in order to operate projects generating hundreds of millions of gross margin, I think that that's a worthwhile trade-off.

Bharath Nagaraj: Understood. The second question I have is on the GPU spot prices, given how volatile they are. Are Gorilla's contracted take-or-pay agreements fixed price? Or is it indexed to the spot? For example, if things change and supply catches up next year and you have a multiyear agreement, is Gorilla taking the margin risk? Or how does it work?

Jayesh Chandan: That's a very good question. So we are basically -- all our hardware are fixed cost today. There is no, what I call, index or pass-through pricing, which transfer some of the volatility to the customers. All of our agreements, for example, the power agreements are fixed. There is no change in the power prices. All of our prices for water and the connectivity are fixed as well. So we're not trying to either pass on the risk to the customer or keep the risk to ourselves. Everything has been defined very clearly. Our customers have also been very understanding. So there's a higher charge upfront. And we are basically telling, look, this is the risk today. This is the cost of memory. It's gone up 40%, 50%, 60% over the last 4, 5 months. Here is the upgraded cost. And we're making sure that we are intending to protect all of the project economics. We're not just relying solely on unhedged spot pricing or anything like that. That said, we have gone to every single vendor. Personally, we haven't -- we've sat down with every single one of them and made sure that the prices are fixed. So whatever they deliver between now and December are all fixed, on a fixed price basis, it does not carry any risk for us.

Bharath Nagaraj: Okay. Very clear. Just a couple more for me, if that's all right, just quick ones. I know you gave us attribution -- project level attribution for your guidance for 2027. I just wanted to understand what kind of utilization assumptions underpin the guidance, like maybe even for '26 and for '27 as well. And then a separate question is around Egypt, Taiwan and Thailand, like what kind of contribution has it made in H1?

Jayesh Chandan: That's a great question. So Bharath, I'm going to break it down into 2. So there is no project level guidance or attribution -- unit attribution because these are take-or-pay, okay? The customer chooses to use it, the customer doesn't choose to use it. It's basically an identified customer, not customer demand, okay? So the mix for us is commissioned GPU capacity. Utilization is 100%. That's how we consider it. Obviously, there are some RMA issues. So the customer has asked us for a 99.95% delivery. But more importantly, we are making sure that all of these recurring compute service revenues are part of the established business. So when I tell you, I'm billing, just to give you an example, $1 for this customer, that is not going to go down to $0.9. It's not going to go up to $1.01. It is actually $1, and that will be something which we will be billing for the next 5 years. Now we model a phase commissioning. The commissioning is what will -- there will be an onboarding customer curve. As the utilization will increase as the workloads will migrate. So what would happen is the customer testing will happen. The first 300 servers will go live, the next 700 will go live, then the next 875 will go live. That is the onboarding curve, which we have. But in terms of the payment, the customer is actually paying us a flat fee for that. Now Yotta is obviously a big contributor as the phases go through. India and Indonesia are poised to become very aggressive. It's going to be a huge part of our revenues going forward. But at the same time, on Egypt, Taiwan and Thailand, we have already had, as Bruce alluded to, we've established already a security, network intelligence, public sector operations and so on. We're not giving up on that. In fact, we're bidding for some very large projects as we speak, and we're going through the motions of closing them as and when. So we will make the necessary announcement there. They remain for us a very important source of revenue, customer collections, but more important, technical credibility to help support our AI infrastructure. Now for people who don't understand what we are, how we actually do this, all of our network intelligence, all of our established security, all of our video, our data intelligence platform, our building management solutions and so on and so forth are part of our data center. Our SOC and NOC, which we have built for Egypt has actually become a part of our business. And now we are providing the same for our data centers. At the same time, we made an investment in a company called Astrikos in Bangalore in India. They are helping us integrate our SOC, NOC and our BMS solutions and providing us even more robust technological infrastructure, which we are personally using for some of our data centers. So if you look at Korat, for example, as and when we build it, almost 90% of all the technology provided will come from Gorilla or Gorilla white label solutions.

Operator: And that concludes our question-and-answer session. I would now like to turn the conference back over to management for any closing comments.

Jayesh Chandan: Thank you very much. I mean, thank you, everybody, for being part of this. I want to thank everyone who actually stood by Gorilla, when the numbers were smaller, our explanations occasionally required a map sometimes. People question where the certain countries were. Sometimes it took a calculator and probably even a strong shot of whiskey. In the next 12 months, now what we've done after that, we've doubled our first half of our revenue. We've reduced our operating loss by approximately 95% from Q1. Now I'd love to take credit, but unfortunately, the people who did it -- who did the actual work are actually listening to this call. So I'm not taking credit for that. Now my wife recently asked me something. She said, hey, what are you thinking about? I said power distribution. Personally, I don't think it's a -- romance is dead. It only requires a substation. And that's what we're building. So as CEO, I provide the optimism. Bruce patiently explains that, hey, the optimism is still not recognized under the IFRS. I've asked him twice. Unfortunately, he doesn't blink. So I'm asking all of you to judge us by the contracts we've signed, equipment we've delivered, revenue we're recognizing and cash we're collecting. To all our shareholders, customers and partners and all of our extraordinary employees, both old and new, I really thank you. We intend to make your patience personally look less like faith and more like excellent judgment. I know sometimes the market can complain and they'll say, you know what, oh my God, the CEO is trying to make a fool of you. That's fine. It's okay. Gorilla is only getting started. You can say only once. And when all goes to plan, and I'm saying this very clearly, when all goes to plan, not if. My family will eventually invite me back to the dinner table, which I left. I promise you that we will be talking about GPUs and megawatts at the dinner. So until then, patience. Thank you very much indeed for your time, and thanks for your support. Cheers.

Operator: Ladies and gentlemen, this does conclude today's conference call. Thank you for your participation, and you may now disconnect.