E2E Networks Ltd — Q4 FY26

E2E Networks Ltd Q4 FY2025-26 Earnings Conference Call https://www.youtube.com/watch?v=2m6Kh_drb9I Published: 3 weeks ago

0:01 1 second Ladies and gentlemen, good day and welcome to E2E Networks Limited Q4 and FI26 earning conference call. As a 0:10 10 seconds reminder, all participant line will be in the listen only mode and there will be an opportunity for you to ask question after the presentation 0:17 17 seconds conclude. Should you need assistance during the conference call, please signal an operator by pressing star then zero on your touchstone phone. Please 0:25 25 seconds note that this conference is being recorded. I now hand the conference over to Miss Rashi Katri from Go India Advisor. Thank you and over to you Rashi. 0:39 39 seconds Uh thank you and good afternoon everyone and welcome to the Q4 and FI26 earnings call of E2E Networks Limited. We have on 0:47 47 seconds the call Mr. Taran Dua, managing director of E2E Networks and Mr. Nithan Jen, CFO of E2 Networks. We must remind 0:56 56 seconds you that the discussion in today's call may include certain forward-looking statements and must be therefore viewed 1:04 1 minute, 4 seconds in conjunction with the risks that a company may face. I will now request Mr. 1:09 1 minute, 9 seconds Tarun to take us through the financial and business updates subsequent to which we can open the floor for Q&A. Thank you and over to you sir. 1:20 1 minute, 20 seconds Yeah. Hi everyone. uh hi to all our team members, all our investors and we hope 1:27 1 minute, 27 seconds that you have had like a good last financial year and uh [clears throat] so 1:33 1 minute, 33 seconds I'll uh do a brief uh coverage of like all the progress we have made during the last quarter and last year and then I 1:42 1 minute, 42 seconds will hand it over to Nan to talk about the financial highlights of this quarter and the overall year and then we'll open the floor for questions. 1:51 1 minute, 51 seconds So in this year like we have like a lot of milestones as a company that and a as a team that we have achieved. So like uh 2:01 2 minutes, 1 second we have been able to use our P platform for large scale GPU clusters and we have helped like uh LLM teams like operate 2:11 2 minutes, 11 seconds like trainings on large clusters over these platforms. So with this like we have demonstrated a full stack 2:19 2 minutes, 19 seconds capability of operating uh both bare metal and container based uh large scale 2:26 2 minutes, 26 seconds deployments for Nvidia powered uh GPU infrastructure at a uh bigger scale and 2:33 2 minutes, 33 seconds uh we have like not only demonstrated uh that like we have the ability to operate the infrastructure but also 2:42 2 minutes, 42 seconds successfully monetize the infrastructure at a much larger scale than we had proven like last year. And uh uh this is 2:51 2 minutes, 51 seconds a testament to the strength uh strengthening of our technology and talent uh which has been used to build 2:58 2 minutes, 58 seconds like very deep in-house capability across the entire five layers of the stack and we have done like major 3:07 3 minutes, 7 seconds software improvements to improve the reliability performance and scalability of our GPU infrastructure in the last 3:15 3 minutes, 15 seconds many quarters and uh uh going forward like we feel that like we have been saying this a lot that like look the 3:23 3 minutes, 23 seconds future is all AI the is like a decadal story now uh given how the markets uh 3:31 3 minutes, 31 seconds all across the world are reacting to uh the AI infrastructure uh usage being increased like very very drastically. So 3:40 3 minutes, 40 seconds I think like we have been validated in uh kind of like putting up like speculative GPU infrastructure. So where 3:50 3 minutes, 50 seconds the world is divided between halves and have nots, we are the halves in terms of like having access to speculative GPU 3:57 3 minutes, 57 seconds capacity where like capacity is running out everywhere. So that allows us to kind of like uh continue our growth. 4:08 4 minutes, 8 seconds So uh with regards to like where we see like medium-term and long-term uh kind 4:15 4 minutes, 15 seconds of like growing in terms of like the world shifting to GPUs from the CPUs from rule-driven software to AIdriven 4:23 4 minutes, 23 seconds software. So we continue to explore like how we can build more infrastructure under our management. So where we have 4:32 4 minutes, 32 seconds explored equity, where we have explored debt, now we are also exploring like a variety of private trade asset models to 4:39 4 minutes, 39 seconds kind of like bring in rapid expansion to our GPU capabilities. 4:44 4 minutes, 44 seconds Uh so so we have consistently met the industry 4:51 4 minutes, 51 seconds benchmarks for performance for like inference or training or GPU deployment uh under our management and uh we expect 5:01 5 minutes, 1 second our uh cluster 1 B200 1024 uh to go live like somewhere in the midmay 5:08 5 minutes, 8 seconds and in a couple of months we expecting to kind of like be able to deploy another cluster of 024 which is which 5:15 5 minutes, 15 seconds has already been planned. Uh and uh then we are very strongly positioned to 5:21 5 minutes, 21 seconds capture uh significant uh uh uptake in the blackwell generation of GPUs apart 5:28 5 minutes, 28 seconds from continuing to run very very strong on the hopper generation. Uh apart from this like we continue to build a plan 5:37 5 minutes, 37 seconds around like uh B300, GB300 and Vera Ruben deployment and we will continue to 5:44 5 minutes, 44 seconds uh keep increasing the GPUs under our management and under our direct deployment uh both through uh partnerships through uh direct 5:53 5 minutes, 53 seconds acquisition and by exploring like various financing models that can help us procure more GPUs. So we'll continue 6:00 6 minutes to focus on that. We continue to focus on our end-to-end project management capabilities to deliver AI infrastructure to uh AI services and AI 6:09 6 minutes, 9 seconds product companies, AI digital natives uh enterprises in the BFSI and other sectors. 6:16 6 minutes, 16 seconds So we'll continue to work towards that. 6:19 6 minutes, 19 seconds So okay so I think like that pretty much covers it that uh so we 6:28 6 minutes, 28 seconds so this has been a good year and a good quarter. So uh so pretty much like our strategy has been validated over last 6:36 6 minutes, 36 seconds many quarters. So where we significantly continue to add to our technology head in terms of like having deep tech talent 6:45 6 minutes, 45 seconds inhouse to be able to uh do stuff. So we gave a certain guidance in terms of like 6:54 6 minutes, 54 seconds where we would be uh by the end of this quarter like a couple of quarters back. 6:59 6 minutes, 59 seconds uh I think like we have more or less like met and exceeded that target and uh we hope to continue to do uh better in 7:09 7 minutes, 9 seconds the future. So uh given that the market is growing very very positively and uh it's a it's a broad trend like there can 7:17 7 minutes, 17 seconds be again like we have mentioned in the past that there are constant drops but ultimately this is a market where uh the 7:25 7 minutes, 25 seconds GPU utilization is going up the demand for tokens is through the roof like so India being a country of more than like 7:33 7 minutes, 33 seconds a billion uh people on phones like and of course even if they were not on the phone like there is like the inherent 7:40 7 minutes, 40 seconds ability to treat every possible piece of information that is going in and out of people in and out of machines in and out 7:48 7 minutes, 48 seconds of enterprise systems uh to be described in the form of tokens. So tokens can be representing voice text any other form 7:55 7 minutes, 55 seconds of data whether invisible spectrum, invisible spectrum, audible voice, inaudible sound waves, ultrasound or 8:04 8 minutes, 4 seconds whatever. So like everything is tokenized data for AI and I think like India as a country generates like a lot 8:12 8 minutes, 12 seconds of data. So we continue to believe very very strongly in the long-term India story of uh AI uh factories being built 8:22 8 minutes, 22 seconds here and India becoming the AI factory of the world. Now with that I would like 8:28 8 minutes, 28 seconds to kind of like hand over for more uh numbers driven financial highlight talk 8:35 8 minutes, 35 seconds to Nan and then we'll open it up for question and answer. 8:42 8 minutes, 42 seconds Nan over to you. Thank you Damon. Good evening everyone. 8:49 8 minutes, 49 seconds Thank you for joining us today. I will walk you through our financial performance for the fourth quarter and fully FY26. 8:57 8 minutes, 57 seconds Q4 FY26 again witnessed a good quarter for E2 network. We have demonstrated our ability to scale AI infrastructure 9:05 9 minutes, 5 seconds rapidly achieve high utilization across GPU cluster and convert capacity into strong revenue growth. Most importantly, 9:14 9 minutes, 14 seconds we have been able to drive our EBIDA margin in four. So let me start with the quarterly highlights for the SIM 4. 9:22 9 minutes, 22 seconds Revenue stood at 956 million which is up 186% yearonear and 37% quarter on 9:29 9 minutes, 29 seconds quarter. IDA stood at 581 million with IDA margin expanding up to 60.7%age. 9:36 9 minutes, 36 seconds Profit before tax turned positive at 86 million compared to a loss of 75 million in Q3. 9:43 9 minutes, 43 seconds Tax stands at 64 million. This performance reflect 120 million swing in profitability sequentially driven by operating 9:52 9 minutes, 52 seconds leverage and a strong execution margin expanded by 413 basis points sequentially. At the same time 10:01 10 minutes, 1 second depreciation increased to 513 million in Q4 reflecting ongoing infrastructure investments. Despite this, we achieved 10:09 10 minutes, 9 seconds positive EBIT and positive PPD for the fullear year. FY 3026 revenue stands at 2456 10:18 10 minutes, 18 seconds million which is up 50% yearonear. IDA cost 1263 million which is up 30.6%. 10:26 10 minutes, 26 seconds However, we reported a pack loss of 156 million which is driven entirely by the depreciation on our GPU infrastructure 10:34 10 minutes, 34 seconds investment. Our core business remains strongly cash positive, operationally profitable at IDA level. As utilization 10:42 10 minutes, 42 seconds continues to ramp up, revenue will progressively outpace depreciation, improving profit reported profitability. 10:51 10 minutes, 51 seconds To conclude, Q4 demonstrate our infrastructure investments are translating into revenue and we are firmly on a path for growth. 11:01 11 minutes, 1 second Thank you for your time. Now I would open the floor for the questions. Thank you sir. 11:08 11 minutes, 8 seconds Ladies and gentlemen, we will now begin with the question and answer session. 11:11 11 minutes, 11 seconds Anyone who wishes to ask a question may press star and one on their touchstone telephone. If you wish to remove yourself from the question Q, you may 11:20 11 minutes, 20 seconds press star and two. Participants are request to use handsets while asking a question. Ladies and gentlemen, we'll 11:27 11 minutes, 27 seconds wait for a moment while the question Q assembles. 11:31 11 minutes, 31 seconds Our first question come from the line of Bhavy Gandhi from Bajage Alternate Investment Management Limited. Please go ahead. 11:39 11 minutes, 39 seconds Yeah. Hi, thanks for the opportunity. 11:41 11 minutes, 41 seconds So, first question is regarding the asset light model that we are looking out for with L & D. Uh could you throw some light how will the asset light 11:49 11 minutes, 49 seconds partnership fractify what sort of arrangement we have? First of all like of course uh we have a MOU with L&T to 11:57 11 minutes, 57 seconds monetize the GPU infrastructure that they are building. So now that is still in exploratory stage. So as the GPUs get 12:06 12 minutes, 6 seconds deployed and like we start into start to work into monetization like as we know more and more like we will talk more and 12:14 12 minutes, 14 seconds more about that. That being said that like this is not a exclusive arrangement. So we will continue to 12:21 12 minutes, 21 seconds operate at an arms length with L&T and we will continue to explore other partnerships uh of similar or different 12:29 12 minutes, 29 seconds nature as well. So there is like like a long gap between like say pure debt 12:36 12 minutes, 36 seconds versus pure equity. So there are like number of structured uh possibilities in terms of like figuring out that how to 12:44 12 minutes, 44 seconds finance the GPU in partnership with a lot of different types of partners with different differing objectives and uh we'll continue to explore all those 12:53 12 minutes, 53 seconds partnerships and as and when like we kind of like uh reach a conclusion on any particular partnership we'll obviously keep everyone informed. 13:02 13 minutes, 2 seconds Sure. But will it be like margin accurative or because because the capex is going to be on other partners. 13:07 13 minutes, 7 seconds It's too early to say Baba like how would the numbers look like? So obviously it's very very early to say like how those things would look like. 13:15 13 minutes, 15 seconds Okay. 13:17 13 minutes, 17 seconds Unless we unless the structure is finalized like uh uh these would like uh uh see obviously we all work for a 13:25 13 minutes, 25 seconds profit so like we will not do anything which doesn't earn us the profit. So uh 13:31 13 minutes, 31 seconds that being said so uh like how to put like numbers to that is not something we can do today. 13:39 13 minutes, 39 seconds Got it sir. And the on and the MR would you like to guide for the next year because you've been doing uh this capex uh from your own pocket also this year. 13:46 13 minutes, 46 seconds So if you can Yes. So we'll we'll continue to we'll continue to kind of like focus on growth 13:54 13 minutes, 54 seconds uh just the same as rest of the market is growing like we want to be very very growth focused. So that being said that 14:01 14 minutes, 1 second I think it would not do justice to do a MR guidance for like one year in the future. I think like the future is changing very very rapidly week on week. 14:12 14 minutes, 12 seconds So uh it is obviously impossible to give like a week on week guidance on MR and uh it would be better to kind of like 14:20 14 minutes, 20 seconds watch it quarter on quarter than kind of like predicting four quarters in advance. 14:25 14 minutes, 25 seconds If you if you can just throw some light on the asset turn for the capex that we are doing at least some sort of uh understanding we can get in terms of 14:33 14 minutes, 33 seconds capex like I think we want to uh get the framing away from things like asset 14:40 14 minutes, 40 seconds terms. So essentially like don't look at us as a asset monetization business. 14:45 14 minutes, 45 seconds Look at us as a technology business. So like I think there was like a recent conversation uh like around the tokens 14:55 14 minutes, 55 seconds becoming more valuable. So same set of tokens that were being generated by uh say open source AI or clos force AI like 15:04 15 minutes, 4 seconds what used to produce like say X amount of value uh we are increasingly seeing that like as the accuracy efficiency 15:13 15 minutes, 13 seconds capabilities of AI increases then the generated tokens also become more valuable uh for businesses who have 15:20 15 minutes, 20 seconds figured out how to utilize those tokens uh in their business. So which means that like uh uh kind of like setting us 15:29 15 minutes, 29 seconds uh setting up ourselves for looking at the uh value of our outcomes based on the size of our assets doesn't do 15:37 15 minutes, 37 seconds justice to uh the business. So ultimately we want to get away from saying that okay this infrastructure 15:47 15 minutes, 47 seconds uh produces only x amount of value. So we will continue to explore options for how to produce more value from the same 15:54 15 minutes, 54 seconds infrastructure by figuring out like where the tokens are more valuable. 15:59 15 minutes, 59 seconds Got it. Fair enough. Okay sir. That's it from my end. I'll get back in the queue. Thank you so much. 16:04 16 minutes, 4 seconds Thanks. Thanks for thank you. Our next question come from the line of Kesha from Nasha. Please go ahead. 16:14 16 minutes, 14 seconds Yeah, thanks for the opportunity to congratulate set of numbers. So like we the black Yeah. Hi sir, am I audible? 16:26 16 minutes, 26 seconds Yes. Yes, please go. Hello. 16:28 16 minutes, 28 seconds Yeah. So like we procured the blacks like way back in the early quarter and you know data not license. So what's the 16:36 16 minutes, 36 seconds reason behind you know the develop are we facing some some kind of procurement issues or something from the 16:44 16 minutes, 44 seconds global supply chains have been impacted somewhat so it's always uh like the delays are always like for what of a 16:52 16 minutes, 52 seconds horseshoe nail so uh it's kind of like we have been working very diligently towards like making sure that like everything is planned out and everything 17:01 17 minutes, 1 second gets delivered on time but then uh you can't control every single component and sometimes like some components can be 17:08 17 minutes, 8 seconds the most uh critical ones from the point of view of like getting the entire deployment done. So we are targeting like the first deployment to go live 17:17 17 minutes, 17 seconds before uh mid of May. So keeping our fingers crossed over there. 17:24 17 minutes, 24 seconds Got it. So I think like you know this quarter that GP demand demand has been strong globally and you know people think that all the GPUs are sold out 17:32 17 minutes, 32 seconds including the old ones. So how are you seeing this trend in the domestic market? So are you also experiencing the same or is it different situation in India? 17:41 17 minutes, 41 seconds Uh okay I think you're I don't think I have understood your question but let me try to answer what I have understood. So 17:48 17 minutes, 48 seconds like uh see obviously it's a uh it has turned into a somewhat into a uh uh much 17:55 17 minutes, 55 seconds better market than it was like a couple of quarters back uh especially for like people who have the GPU infrastructure 18:02 18 minutes, 2 seconds in place. Uh that being said that as a company as an Indian company we are very very focused on India first. So wherever 18:10 18 minutes, 10 seconds we get an opportunity to support Indian companies we definitely like to do that. 18:16 18 minutes, 16 seconds Now uh uh with that in mind like basically like we always prioritize India first and then uh we are also 18:23 18 minutes, 23 seconds happy to support the uh global infrastructure needs today. I I I hope the answer is in the right direction towards what you were asking. 18:33 18 minutes, 33 seconds Yes. So that's definitely like I was just trying to ask like you know demand has been strong globally that the GPS gold got sold out. So I was trying to 18:42 18 minutes, 42 seconds understand there is fresh we are seeing like demand both India as well as globally. Uh it is strong both ways and 18:50 18 minutes, 50 seconds uh we definitely try to prioritize India. Uh so but then like wherever the demand comes from eventually like we are happy to fulfill it. 19:00 19 minutes Got it. And one last question from my side that employee cost increase uh you know this quarter. So if you could give some color on that like what's the key drivers behind this size. 19:11 19 minutes, 11 seconds See as we grow like we kind of like start figuring out like more and more interesting problems that need to be solved in this space. Uh obviously our 19:21 19 minutes, 21 seconds goal is to go toward uh what you call uh higher value tokens and that obviously 19:29 19 minutes, 29 seconds requires the uh application of uh uh quite a uh high level of talent. So like 19:38 19 minutes, 38 seconds uh uh as we obviously grow the base effect would still be there. So but on the other hand like we want to maintain 19:46 19 minutes, 46 seconds a balance of not losing out on future opportunities because uh we didn't invest today on what was required 12 19:53 19 minutes, 53 seconds months later or 18 months later or 2 years later. 19:57 19 minutes, 57 seconds Got it. And one last thing like if we could also provide the GP utilization levels rate if possible like for this 20:03 20 minutes, 3 seconds quarter. See broadly like uh across our entire infrastructure not just GPUs I think we are looking at like 80% plus 20:12 20 minutes, 12 seconds utilization uh so certainly uh uh across the quarter now I would say like only in the March 20:20 20 minutes, 20 seconds month we are looking at like 80% plus kind of utilization certainly less than 85%. 20:25 20 minutes, 25 seconds So uh so there is like uh uh elasticity in terms of like uh basically like how we can increase the utilization further over there. 20:35 20 minutes, 35 seconds Got it. So I will get back into the Thank you so much and come back again for the Sure. Thank you. 20:41 20 minutes, 41 seconds Thank you. Next question come from the line of Barat Gulati from Dalal A and Rocha. Please go ahead. Yeah. Hi. 20:50 20 minutes, 50 seconds Hi sir. Uh just wanted to get a breakup of what kind what is our MR breakup currently? Can you give a split between 20:57 20 minutes, 57 seconds India mission enterprise and also a split between what would the GPU contribution of that MR be as compared to CPU? Yeah, 21:06 21 minutes, 6 seconds see I think like uh broadly the way things are progressing like in another couple of quarters I think GPU 21:13 21 minutes, 13 seconds contribution would exceed um closer would be closer to uh 85 90% uh over 21:20 21 minutes, 20 seconds coming quarters. Uh that being said like all these numbers are fairly dynamic. 21:26 21 minutes, 26 seconds So in the sense that like uh they vary very very rapidly from a week to another week. So I wouldn't like to kind of like 21:34 21 minutes, 34 seconds uh uh and the sample size today is fairly small. So like what's the split is something that like we don't want to 21:42 21 minutes, 42 seconds worry about today. So like as we grow the cluster sizes to a much larger number then it would start making sense 21:50 21 minutes, 50 seconds to kind of like talk about like okay hoppers you've got a couple of thousand of them uh then uh uh blacks you've got 21:58 21 minutes, 58 seconds a couple of thousand of them that's the point at which like it would start making sense to say uh okay what's happening on hoppers what's happening on 22:05 22 minutes, 5 seconds black wheels today the entire universe of GPU CPU storage all of that stuff that we have is like fairly fairly small 22:13 22 minutes, 13 seconds to uh kind of like put a very very strong split around those numbers today. 22:19 22 minutes, 19 seconds Fair enough. So but could you just give a MR split between the India mission and our enterprise or our theme clients you know just trying to get an idea of what 22:28 22 minutes, 28 seconds kind of visibility do we have in that 374 million MR and also just to add on to that before that our MR for the previous two months if we average it 22:36 22 minutes, 36 seconds would come to like a 290 million kind of MR. So just trying to understand that spike that come in the came in the last month. What was the reason for that 22:43 22 minutes, 43 seconds spike and also how sustainable is this 374? 22:47 22 minutes, 47 seconds It's broadly an increase in it's broadly an increase in overall utilization uh that has led to the increase in the uh 22:54 22 minutes, 54 seconds spike in the March month. uh from the perspective of uh uh like uh uh overall 23:02 23 minutes, 2 seconds split like I think like uh the India uh kind of like 23:13 23 minutes, 13 seconds yeah so basically like the overall government business like uh across this portfolio has not exceeded more than like say uh 35 40%. 23:24 23 minutes, 24 seconds All right. Right. So, so could you is there some particular reason for that because ideally our missions were our India mission was supposed to start in 23:31 23 minutes, 31 seconds the month of Jan we are in May we are in a March as of this quarter's reported numbers so what kind of ramp up are we 23:39 23 minutes, 39 seconds seeing in that and you know just any kind of timelines and what kind of revenues in terms of MRS do we expect from that the way we have always looked at India 23:48 23 minutes, 48 seconds AI mission is that like look we are trying to obviously like we said that we are supporting India we are supporting India mission we are supporting Indian 23:55 23 minutes, 55 seconds companies and we definitely give a preference to all those workloads. But that being said like so whatever uh 24:03 24 minutes, 3 seconds capacity is not opten so uh we are not trying to kind of like force anyone to opt any capacity from us. So we are very 24:11 24 minutes, 11 seconds very happy to sell it to uh outside the scope of either the AI mission or the uh government workloads. So we have enough 24:20 24 minutes, 20 seconds customers across multiple segments to kind of like uh solely rely on one revenue driver growth. But that being 24:28 24 minutes, 28 seconds said like we continue to work very closely with the AI mission uh for the coming generations of GPUs as well and 24:35 24 minutes, 35 seconds uh we are hopeful that like we will continue to collaborate continue to work together uh continue to solve problems. 24:43 24 minutes, 43 seconds So just trying to understand that India emission was almost all our hopper and hopper series GPUs were given to India mission. We're seeing 80% utilization 24:52 24 minutes, 52 seconds but India no we have not we have not scaled up we have not scaled up that to that extent. 24:58 24 minutes, 58 seconds So majorly like uh whatever was the scale required by the a loties of AI mission we have provided that scale but 25:06 25 minutes, 6 seconds like we have not insisted on uh them having to increase their scale uh without like uh having the need for that. So we are quite okay with that. 25:16 25 minutes, 16 seconds So in terms of our revenue currently what kind of clients are we currently catering to and what's the pipeline? You know you said midMay we expect Blackwell 25:24 25 minutes, 24 seconds to get deployed. Do we have some firm orders? This is this is a very very small base like so it's like uh kind of 25:30 25 minutes, 30 seconds like okay uh there are GPU customers like everywhere in the world there are GPU customers in India there are GPU 25:37 25 minutes, 37 seconds customers outside India like every possible segment that you can think of like whether it is enterprise whether it is BSSI whether it is education everyone 25:45 25 minutes, 45 seconds needs GPUs so it's practically like uh basically like almost first come first serve today so like the nature and 25:54 25 minutes, 54 seconds profile and putting like a uh stat on like okay this much of our business comes from here this much of our 26:01 26 minutes, 1 second business comes from there is like kind of like very very premature I think like we need to grow the base of GPUs to a 26:09 26 minutes, 9 seconds couple of tens of thousands before these numbers start to make sense so today at a very very small base like none of 26:16 26 minutes, 16 seconds these numbers would make sense to kind of like track and then kind of like say that okay uh what's what's the kind of number this quarter it will next quarter. 26:27 26 minutes, 27 seconds So, so would it be fair to say that until we don't build a huge base of GPUs, our MRS will be lumpy in nature? 26:32 26 minutes, 32 seconds Is that what you're trying to iterate at? 26:34 26 minutes, 34 seconds No, I think the demand has secularly changed in the GPU world. So, like from the era in which like the lumpiness was there because our base was even smaller. 26:45 26 minutes, 45 seconds I think we have come a long way over there. uh that being said the lengthiness will continue to decrease as we continue to build more and more 26:53 26 minutes, 53 seconds volume of GPUs under our management. So out of this 374 million what exactly would be long-term MRS that at least we 27:01 27 minutes, 1 second are seeing a visibility for the next 8 to 10 months or a year or so. 27:07 27 minutes, 7 seconds See we are not very very focused uh uh on uh so 27:15 27 minutes, 15 seconds so we have a balance of like basically like thinking about like okay what needs to be like 6 months one year uh month on 27:22 27 minutes, 22 seconds month which is preemptable. So we are doing like kind of like a whole series of them. So like it's a week-on-week 27:30 27 minutes, 30 seconds effort to kind of like figure out that okay what is the capacity you want to sell hourly what is the capacity you want to sell uh weekly preemptable what 27:38 27 minutes, 38 seconds is the capacity you want to sell monthly what is the capacity you want to sell yearly that being said like wherever larger clusters are involved we try to 27:46 27 minutes, 46 seconds do at least like uh 6 months to one year visibility with our customers and increasingly for even larger customers 27:53 27 minutes, 53 seconds we are trying to look at even longer visibility be to at least like uh two years to even going up to three years. 28:00 28 minutes But that being said like our view is that like as tokens become more expensive to generate and uh kind of 28:09 28 minutes, 9 seconds like they generate more value for people uh it is better to kind of like not put all the eggs in the long-term basket but 28:18 28 minutes, 18 seconds have like a incre uh judicious mix in very short-term contracts and medium-term contracts and some degree of long-term contract. 28:28 28 minutes, 28 seconds Got it. So got it. And just one last thing so in terms of the black well that we expect to get deployed in May do you can you give any idea in terms of who will be the customer will it be an 28:37 28 minutes, 37 seconds enterprise India mission and theme or and when would that customer's revenue start to come in would it come or will there be a lag 28:44 28 minutes, 44 seconds we will we will announce that after closing that particular deal instead of like saying anything speculative today 28:52 28 minutes, 52 seconds fair enough so fair enough thank you that's it for my s thank Thank you. 28:59 28 minutes, 59 seconds Our next question come from the line of Nishan Johi from Equisense Advisor Private Limited. Please go ahead. 29:06 29 minutes, 6 seconds Yeah. Hi. 29:08 29 minutes, 8 seconds Good evening. I have one question. As you said, company is building assets using I'm sorry to interrupt you Mr. Johooshi but your voice is not audible. 29:19 29 minutes, 19 seconds Am I audible now? 29:21 29 minutes, 21 seconds Yes, please proceed. Uh I want to say that as company is planning to build assets using debt as well as equity and 29:29 29 minutes, 29 seconds subsequently intend to go for the asset like model also. So will this change our business model means instead of providing hardware we would be providing 29:37 29 minutes, 37 seconds more of service uh we'll be offering our T platforms. So will this lead to a change in our revenue model? Think of 29:45 29 minutes, 45 seconds this think of this as a expansion of the number of business models without shutting down any of the existing business models. So we will continue to 29:54 29 minutes, 54 seconds own continue to operate GPUs. We'll continue to acquire new GPUs. We will continue to expand the partnerships to 30:02 30 minutes, 2 seconds kind of like bring increasing number of GPUs under our management uh available through our Track platform uh which 30:09 30 minutes, 9 seconds operates at multiple layers where it is a choice of the customers like what layers they want to buy uh and uh kind 30:17 30 minutes, 17 seconds of like uh nothing is like uh changing from the point of view that okay we are not going to do this or we not going to do that. We are simply saying that like 30:26 30 minutes, 26 seconds we are going to do additional number of things to expand the universe of our thinking to uh do a lot more than what we are doing today. 30:36 30 minutes, 36 seconds So sir it could be an ideally additional line of revenue you mean to say absolutely absolutely. 30:43 30 minutes, 43 seconds Okay that's what my query sir. Thank you. Thank you. 30:48 30 minutes, 48 seconds Thank you ladies and gentlemen. In order to ensure that the management will be able to address all the question from the 30:57 30 minutes, 57 seconds participant, we request you to kindly limit your question to two question per participant. If you have a follow-up question, please rejoin the queue. 31:07 31 minutes, 7 seconds Our next question come from the line of Deepak Podar from Sapphire Capital. Please go ahead. Yeah, I'm audible sir. Yes, you are. 31:14 31 minutes, 14 seconds Oh, okay. Um, thank you very much for this opportunity. So, I just wanted to understand now um at a March MMR MR of 31:21 31 minutes, 21 seconds 37.4 four cross. So what is our capacity utilization and on what capacity? 31:28 31 minutes, 28 seconds Uh so this is closer to overall capacity utilization of around 80% is in the March month and and and and on a capacity base of 3,900. 31:38 31 minutes, 38 seconds Yeah. Yeah. So this is like CPU, GPU, storage all capacity put together. So the utilization is closer to 80%. 31:46 31 minutes, 46 seconds Closer closer to 80%. Yeah. And and and I think we have around 50 lined up in 31:53 31 minutes, 53 seconds next what 6 to 12 months right 48 plus some pairs uh that we are planning to deploy in this uh financial 32:01 32 minutes, 1 second year. Uh starting from May we the first uh lot of 1024 is expected to go live. 32:08 32 minutes, 8 seconds Uh so so by FI27 end we would you targeting around 6,000 of capacity. 32:14 32 minutes, 14 seconds I wouldn't want to place a limiting number over there but like you could say that like that's the minimum number. 32:22 32 minutes, 22 seconds Huh. At least at least at least that. 32:24 32 minutes, 24 seconds Okay. Okay. Understood. And that that part is already visible. So that's what we've already spoken about. 32:30 32 minutes, 30 seconds So yeah, understood. And you mentioned around 80 85% utilization by March. I mean you're talking about by March 27 of this 32:37 32 minutes, 37 seconds expanded capacity. I mean how so? So so no no this is the previous previous gone by March. 32:43 32 minutes, 43 seconds previous knowledge only. Okay, understood. And my second question is on your fixed asset. We have got around 1,500 crores out of feed fixed asset 32:51 32 minutes, 51 seconds right in last two years capex has been around close to that only. I mean um so so this entire is a depreciable asset. I mean would that be a fair assumption? 33:00 33 minutes Uh I would let like uh uh Nan put a perspective on that. 33:06 33 minutes, 6 seconds Yeah. could be dire that fix block is is the depreciable asset which constitute of the new V200 which is currently displayed as uh CWIP. 33:18 33 minutes, 18 seconds Correct. Correct. And and what is the amortization schedule for for this? I mean is it 3 to four years? 33:26 33 minutes, 26 seconds Six years. Six years. Okay. Okay. Okay. Understood. And and what would be FI27 Kix target? 33:37 33 minutes, 37 seconds So like we said like we are already planning to deploy 2048 B200. So that is 33:44 33 minutes, 44 seconds already in place uh as far as our known plans are concerned. That being said like that's we are not limiting 33:51 33 minutes, 51 seconds ourselves to those plans. So we continue to deploy uh capital under various business models judiciously to expand the GPU footprint like fairly rapidly. 34:03 34 minutes, 3 seconds Correct. So any number you you have I mean what is I mean I I think FI26 was close to a look back a look back is more important than putting a number in production. 34:14 34 minutes, 14 seconds Okay. Okay. Okay. Okay. Okay. 34:15 34 minutes, 15 seconds Understood. Um that would it from my side. I wish you all the best. Thank you. Thank you. 34:22 34 minutes, 22 seconds Thank you. 34:25 34 minutes, 25 seconds Our next question come from the line of Vun Gandhi from Finn Avenue Growth Fund. Please go ahead. 34:33 34 minutes, 33 seconds Hi Taron. Uh my question my question is on GPU rental prices and 34:40 34 minutes, 40 seconds uh are we seeing pressure on them and I asked this from two aspects. The first aspect is the technologies advancing rapidly. So we see the Ruben 34:49 34 minutes, 49 seconds architecture has been uh disclosed by Nvidia and which is much more co cost efficient for uh inference compared to 34:56 34 minutes, 56 seconds your uh even your blackmail architecture. And number two is the competition that's been ramping up across domestic uh competitors, right? 35:05 35 minutes, 5 seconds So are we witnessing any sort of pressure on GPU rentals right now or do you foresee that in the near term? 35:13 35 minutes, 13 seconds I think like broadly the trend today is that like uh there are not enough GPUs in the world that people want to buy and 35:20 35 minutes, 20 seconds deploy. So that's the current trend. Now whether this trend remains like uh uh 35:27 35 minutes, 27 seconds for how long like is hard to predict but uh the broad secular trend has been there that like basically the there are 35:35 35 minutes, 35 seconds AI believers and uh there were there were AI non-believers I think like the uh proportion of non-believers is like 35:43 35 minutes, 43 seconds slowly dwindling and uh uh it's quite clear that like basically uh so we are barely scratching the surface in the 35:52 35 minutes, 52 seconds terms of uh in terms terms of utilization of AI like we have uh looked at so many of our customers so many of 35:59 35 minutes, 59 seconds the organizations we work with including our very own so where we say that okay is it uh uh is it the case that like our 36:08 36 minutes, 8 seconds AI uh utilization is to an extent that we want it today the uh answer for ourselves that we increasingly get is no 36:16 36 minutes, 16 seconds this is going to be like more like 10x 20x 50x of where we are today in terms of how we are utilizing utilizing AI. So AI is not coming out of the IT budget. 36:27 36 minutes, 27 seconds AI is essentially coming out of your budget for the capabilities that you're building for the business. So that being 36:35 36 minutes, 35 seconds said that I don't think like the demand environment in the foreseeable future is going to uh be uh changing negatively 36:44 36 minutes, 44 seconds for long periods of time. I don't think that going to be the case uh from the visibility we have today. So uh so like 36:53 36 minutes, 53 seconds we we don't see that there is like any negative pressure on the pricing today. 36:59 36 minutes, 59 seconds So in fact like there are like I think good set of uh tailwinds which are uh slowly inching up the prices rather than like a decrease in pricing. 37:09 37 minutes, 9 seconds Got you. So from from a demand standpoint, we aren't seeing any sort of negative but but from a technology advancement standpoint let's say if the 37:17 37 minutes, 17 seconds Ruben architecture comes in live in this year and it should uh would that pressure there is a lot of workloads that will go 37:25 37 minutes, 25 seconds into production where they will continue to operate on the targeted architecture. 37:31 37 minutes, 31 seconds So typically it's like a uh what you call like a uh u uh funnel kind of a growth where uh the next layer of the 37:39 37 minutes, 39 seconds funnel is bigger than the previous layer of the funnel. So which means that like uh the uh demand for existing GPUs will 37:47 37 minutes, 47 seconds continue to remain very very strong. So demand for even the NP series has remained strong. Uh hopper series has 37:54 37 minutes, 54 seconds remained strong. uh like uh uh Karon I understand the demand has remained strong but let's say 2 years ago the GPU rental prices for Hopper 38:03 38 minutes, 3 seconds H200s are they same today or they are lower 30 40% lower than what they were quoting at 2 years ago. uh see I think 38:12 38 minutes, 12 seconds like uh uh taking the case of two years ago is slightly uh anomalous from the point of view that I think that was one 38:20 38 minutes, 20 seconds particular year in which the GPU demand went from say 1 to 10. So uh like I 38:28 38 minutes, 28 seconds think like uh that capabilitydriven uh shortages where the capability to print that many GPUs was uh had to be built 38:36 38 minutes, 36 seconds very rapidly. So that created like a I think like a uh local maxima of pricing but broadly we are seeing that like like 38:46 38 minutes, 46 seconds there is stability in the pricing uh for GPUs over an extended periods of time. 38:51 38 minutes, 51 seconds Typically we feel that like it is possible to kind of like definitely uh use and utilize the GPUs over a 7 8 year 38:59 38 minutes, 59 seconds period. So we don't think that like the GPUs are like a twoear story or a threeear story and then kind of like 39:06 39 minutes, 6 seconds they uh there is something available at a much lower cost and you are able to shift and those shifting costs are like very reasonable. So that's not the case. 39:16 39 minutes, 16 seconds Like your entire test debug cycle for doing AI on a new chipset is like a whole other set of expenses that you 39:26 39 minutes, 26 seconds don't want to do like if your job is getting done by a GPU which is uh available at a very reasonable price. 39:33 39 minutes, 33 seconds Got you. Now my second question is uh at the beginning of the call you touched upon how you've significantly improved your software architecture. 39:42 39 minutes, 42 seconds uh could you just give me a brief uh if you could just highlight a brief example of how you've done that you you iterated 39:49 39 minutes, 49 seconds that you're now trying to capture high value tokens. Is there some sort of example that you could showcase? 39:57 39 minutes, 57 seconds So this is broadly from the point of view is that like the tokens themselves today are more valuable. It's just a 40:05 40 minutes, 5 seconds point of like choosing the right set of tokens that people are going to utilize. 40:10 40 minutes, 10 seconds So uh it's not like a uh uh kind of like a strategy statement to go and capture 40:18 40 minutes, 18 seconds one particular set of high value tokens today. It's a broad futuristic outlook that like look ultimately uh we have to 40:25 40 minutes, 25 seconds go and figure out like which are the high value tokens. Now that being said like from a platform improvement perspective like we have been working on 40:33 40 minutes, 33 seconds uh things like uh uh which are required by uh people running training on larger clusters, people who are running 40:41 40 minutes, 41 seconds inference, how to support them, uh people who want to build rapidly the pipelines reliably uh across larger 40:48 40 minutes, 48 seconds number of GPUs and understanding of like how what is it that AI data scientists do from a been there done that perspective. So we have made 40:57 40 minutes, 57 seconds improvements in all those areas for productization of our feedback stack. 41:02 41 minutes, 2 seconds Right. So just as an extension to this question um see we're very well positioned to capture the sovereign AI tailwind and my understanding is that 41:10 41 minutes, 10 seconds BFSI companies would be a key client base over there. Is there any sort of strategic development that you could share where you you're partnering with 41:19 41 minutes, 19 seconds some sort of SAS companies or other software companies and trying to uh offer bundle of software plus compute? 41:26 41 minutes, 26 seconds Uh is there something you could help us with? 41:30 41 minutes, 30 seconds Yes. So as we do something like we'll definitely share that with everyone uh over here. So like as things happen like 41:38 41 minutes, 38 seconds we'll do a look back and say that okay this is what we've done. 41:43 41 minutes, 43 seconds Okay. But there's there's nothing that you can share in terms of your thought process over here. 41:50 41 minutes, 50 seconds See like uh we are we continue to drown in opportunity. So what we capitalize on we will come and inform the market. 42:00 42 minutes All right. Thank you very much. Yeah. Thank you. 42:05 42 minutes, 5 seconds Thank you ladies and gentlemen. A reminder to all the participant in order to ensure that the management will be able to address all the question from the participant. 42:15 42 minutes, 15 seconds We request you to kindly limit your question to two question per participant only. If you have a follow-up question, please rejoin the queue again. 42:25 42 minutes, 25 seconds Our next question come from the line of Abhishek from Incredities. Please go ahead. Yeah. Hi. Hi. 42:33 42 minutes, 33 seconds Hi sir. Thanks for the opportunity and congrats on a great set of numbers. Uh sir my first question first question is 42:41 42 minutes, 41 seconds about uh you know the guidance commentary. Uh you know if we summarize your comments on the demand it appears 42:50 42 minutes, 50 seconds to be really strong. We have a visibility of uh 2048 GPUs and then we 42:57 42 minutes, 57 seconds said that you know we don't want to get fixated on the guidance. So just trying to understand that the purpose here was 43:05 43 minutes, 5 seconds not to get fixated on the numbers rather than any worries in your mind about uh the current macro. Is the understanding right? 43:15 43 minutes, 15 seconds Yeah. Yeah. We are not worried about like whatever numbers we give whether we will be able to meet them or not. I think it's like uh uh we don't want to 43:23 43 minutes, 23 seconds give a underwhelming number without first exploring like over uh week after week over next 52 weeks like what we are 43:32 43 minutes, 32 seconds capable of doing. So it will become sort of like a limiter for us ourselves to say that okay we are only to meet this number. 43:41 43 minutes, 41 seconds So that's why we are not putting a number today. 43:44 43 minutes, 44 seconds Perfect. That's very helpful sir. The second question is on the pricing. So contrary uh you know to the prior 43:52 43 minutes, 52 seconds question um you know I was the recent articles are suggesting that the GPU spot rates for April have gone up 44:00 44 minutes substantially higher. So for an H100 the prices are up anywhere by 25 to 30%. Um 44:08 44 minutes, 8 seconds can you just we we are seeing that we are seeing that we are looking at that. Uh that being said like we uh obviously like we try to 44:17 44 minutes, 17 seconds capture some of that but then that being said like we are also trying to support uh the Indian companies and India as a 44:25 44 minutes, 25 seconds country. So where we want to have a balance between long-term and short terms. So shortterm of course like we 44:33 44 minutes, 33 seconds want to make more money and long-term of course we want to work with the kind of customers who will sustain and grow on 44:40 44 minutes, 40 seconds our infrastructure over a long period of time. 44:44 44 minutes, 44 seconds Uh perfect. So just last one data point from Nitpin sir. So uh the depreciation for the next quarter should be up by 44:52 44 minutes, 52 seconds around uh you know 25 crores a quarter right? Is that assumption right? 44:57 44 minutes, 57 seconds I'll let Nathan answer the questions around the depreciation. Uh Nan over to you. Uh sorry could you repeat again? 45:05 45 minutes, 5 seconds Uh sir the the depreciation for the next quarter should be up by should go up incrementally by 25 crores a quarter. Is the assumption right sir. 45:16 45 minutes, 16 seconds Why do you expect that? Uh because uh because of the B200 cluster. 45:24 45 minutes, 24 seconds B200 cluster would be that we are saying the timeline of midway. So that's what there would be increase in that depreciation but not to that significant effect that's what you are telling us. 45:36 45 minutes, 36 seconds Okay perfect sir that is very helpful. 45:38 45 minutes, 38 seconds Thank you for taking my question and best wishes for the next year. Sure. Uh thank you. Thank you. 45:46 45 minutes, 46 seconds Our next question come from the line of Neil Mun from Pico Capital. Please go ahead. Hi sir. 45:53 45 minutes, 53 seconds Yeah. Hi. Hi Neil. 45:57 45 minutes, 57 seconds Uh hi, thank you for the opportunity. So my first question is building on this partic 46:04 46 minutes, 4 seconds um the capacity utilization is uh I'm sorry to interrupt you Neil but your voice is breaking is it better now? 46:16 46 minutes, 16 seconds Uh yes please proceed. 46:18 46 minutes, 18 seconds Yeah. So, uh the first question is basically building on participants question is that the 80% utilization 46:26 46 minutes, 26 seconds that we're talking about I'm assuming this is excluding the the new B200s that have Yes. This that is excluding the B200 obviously. 46:34 46 minutes, 34 seconds Yeah. So uh when we talk about the 80% utilization uh with respect to the remaining uh capacity available have we 46:43 46 minutes, 43 seconds kept these aside for PC's or demand is still ramping up in that sense? 46:49 46 minutes, 49 seconds No. So the demand is still ramping up over there. So like uh uh it's like all 46:56 46 minutes, 56 seconds sorts of different sorts of capacity. So like they all work in different combinations. So an ideal utilization 47:02 47 minutes, 2 seconds number uh uh could be like uh uh uh kind of like uh it depends on like basically 47:10 47 minutes, 10 seconds like where do you end up but that's not a limiting factor on uh the MR from a very straight line perspective. So for 47:18 47 minutes, 18 seconds instance like the same uh particular piece of hardware could like under different circumstances be at like wildly varying rates. 47:30 47 minutes, 30 seconds So uh what you my assumption these are all averages like essentially. 47:35 47 minutes, 35 seconds So these are all important from a look back perspective. So what this is our uh assumption on that like look we are like 47:42 47 minutes, 42 seconds uh doing what is like uh thought of like an 80% capability. Now 80 could be 80 80 could be maybe 85 80 could be even 75 maybe even 70. 47:54 47 minutes, 54 seconds Okay. And so with respect uh to the demand scenario in B200 have we closed entire mission? 48:04 48 minutes, 4 seconds Uh we we have not closed any deals on the black fill capacity which is going up as yet like obviously once we close a 48:12 48 minutes, 12 seconds deal we will kind of like inform uh everyone over here. So the reason for this is is the reason uh ramp up or is 48:21 48 minutes, 21 seconds the reason that we are still closing PC's or in that sense. 48:30 48 minutes, 30 seconds So there is a lot of interest in the blackmail capacity that we are building. 48:34 48 minutes, 34 seconds So uh kind of like we are working towards with multiple customers to figure out like what sort of problems 48:42 48 minutes, 42 seconds they need us to solve for them and what is it that they are looking at from us and we obviously want to work with the 48:49 48 minutes, 49 seconds customers that are uh uh best placed to utilize what we have to offer. 48:57 48 minutes, 57 seconds Uh please please rejoin the queue if you have a more question. Thank you. Thank you. 49:06 49 minutes, 6 seconds Next question come from the line of Hardik Gandhi from HPMG share and securities. Please go ahead. 49:13 49 minutes, 13 seconds Hello. Hi sir. Thank you for the opportunity. Hi. Hi. 49:15 49 minutes, 15 seconds Congratulations on a set of good numbers. Uh thank you. 49:19 49 minutes, 19 seconds Just wanted to push on a question asked by the previous participant. I think so majority of the investor group just 49:26 49 minutes, 26 seconds wants to know whether the revenue uptick is due to any lumpy nature or are you seeing any sustainable growth in the 49:33 49 minutes, 33 seconds long run because in the last December and March we had we suddenly dropped sorry but we suddenly dropped the 49:41 49 minutes, 41 seconds numbers we s we stated that the startups were no longer wanting our uh ecosystem 49:47 49 minutes, 47 seconds right so uh we just want to know from a long-term perspective or at least from a So let me let me try to answer that 49:56 49 minutes, 56 seconds question as best as possible. So see like the lumpiness is a part of the nature of the business. So the lumpiness 50:03 50 minutes, 3 seconds will keep going down as we keep increasing our uh GPU base. Now that being said like we are currently seeing 50:12 50 minutes, 12 seconds the kind of customers who are buying these GPUs have sustainable long-term workloads. So these are not of the 50:19 50 minutes, 19 seconds short-term nature for most part. So we do expect to have sustainability uh for these workloads uh over the 50:27 50 minutes, 27 seconds medium term. So I don't think like there is going to be a massive amount of lumpiness in the near term. Uh 50:35 50 minutes, 35 seconds medium-term of course like as the number of GPUs grow like I think the uh effect of lumpiness would be far more muted uh 50:43 50 minutes, 43 seconds than it has shown up in the past in the early days of us building the capacity. 50:50 50 minutes, 50 seconds Right. Right. And the second question from uh from my side was to understand again on this part itself that what percentage of revenue are we allocating 50:59 50 minutes, 59 seconds to a long-term MR sorry long-term contracts so that our MR it's of a dynamic nature so like 51:07 51 minutes, 7 seconds basically like it is not like possible to predict uh yeah we are not predicting I'm just saying that from a from a like from a uh 51:16 51 minutes, 16 seconds we are not setting any hard lines over there we are not setting any hard lines over there like there are I think like uh far 51:23 51 minutes, 23 seconds more number of variables to consider than it would be possible to describe over here that like what to pick up when 51:30 51 minutes, 30 seconds in terms of like customer interest. So like uh I don't think it would be like I would be able to give like a very fair 51:40 51 minutes, 40 seconds answer to that question. So like we are definitely interested in having some percentage of long-term contract. what 51:47 51 minutes, 47 seconds that percentage will come out to be is like something on a we need to look at and on a look back basis when the size 51:54 51 minutes, 54 seconds of the size and scale of the uh GPU installation is sufficient. So I think in the short term there is no clearcut answer to that. 52:03 52 minutes, 3 seconds Right. Just a small data point again. Uh what percentage of revenue did uh come from uh customers outside India versus 52:11 52 minutes, 11 seconds uh domestic demand for the last quarter? 52:18 52 minutes, 18 seconds Uh I'll let like uh Nathan take your stab at this question. 52:23 52 minutes, 23 seconds So the international revenue for the last quarter is roughly around uh 351%. 52:29 52 minutes, 29 seconds 35 to 37% is the international customer revenue. 52:34 52 minutes, 34 seconds Okay. Yeah. Thank you. Thank you. That's that's really helpful. I appreciate the time. Thank you. 52:42 52 minutes, 42 seconds Our next question come from the line of Shirinasu K from Trust Investment Advisor Private Limited. Please go ahead. Yeah. Hi. Hi Shinasu. 52:51 52 minutes, 51 seconds Uh hi sir am I audible? 52:54 52 minutes, 54 seconds Uh yes yes loud and clear. 52:57 52 minutes, 57 seconds Yeah so uh sir what percentage of uh uh pour revenue came from inference 53:03 53 minutes, 3 seconds workloads versus training and uh any uh what would be the feature split up of this mix at least directionally. 53:12 53 minutes, 12 seconds Um so I think like there is a lot of uh confusion about like what consists like of inference revenue versus training 53:20 53 minutes, 20 seconds revenue. I think these are not like super hard line. Sometimes the same set of GPUs are used by the customers uh for 53:29 53 minutes, 29 seconds kind of like bursting up the inference workloads when the inference workloads are high and the uh at a time when the inference workloads are low they get 53:37 53 minutes, 37 seconds converted to training. So the workload management is far more flexible today to kind of like point to like okay what 53:45 53 minutes, 45 seconds percentage of influence what percentage of training today and uh uh I think again uh like at a much larger scale like these numbers would become clearer. 53:56 53 minutes, 56 seconds So today of course like it's hardly possible to kind of like super differentiate between okay who is using 54:03 54 minutes, 3 seconds that 1 GPU 4 GPU 8 GPU 16 GPU cluster either for training or for inference. 54:10 54 minutes, 10 seconds But that being said like majority of revenue is still like uh closer to training than to inference. 54:20 54 minutes, 20 seconds Okay sir. And uh uh you said the 6,000 GPUs uh by I27 at least minimum, right? 54:27 54 minutes, 27 seconds So what would be the upside? Can is it uh could you deploy 8,000 an upside 8,000 to 10,000? 54:35 54 minutes, 35 seconds I don't want to put any numbers right now. So like let's look at those numbers as I look back like uh over 54:44 54 minutes, 44 seconds uh coming period of time. So rather than like uh putting putting out a number today. 54:52 54 minutes, 52 seconds Okay. Thank you and best of luck. Yeah. Thank you. 54:59 54 minutes, 59 seconds Thank you. 55:01 55 minutes, 1 second Our next question come from the line of Rohan Nagpal from Hoys Capital Management. Please go ahead. 55:09 55 minutes, 9 seconds Hi. Hi. 55:10 55 minutes, 10 seconds Thanks for the opportunity. Hi. Thanks for the opportunity and congrats on a good set of results. Um my first question was um so you said that there 55:18 55 minutes, 18 seconds is a uh there is a variety in terms of um your current MR being um short-term, medium-term, slightly long-term. Uh 55:26 55 minutes, 26 seconds could you provide some um uh some like directional commentary on how much of that MR is 6 months plus or one year plus. 55:38 55 minutes, 38 seconds So I think like one is like we are still on a very small base of GPUs. So kind of 55:45 55 minutes, 45 seconds like we are not doing any hard split in terms of like where we want to be or where we are today and it's fairly uh 55:53 55 minutes, 53 seconds dynamic and rapidly changing uh both with new capacity coming up online. So all these numbers are subject to very 56:02 56 minutes, 2 seconds rapid and very wide amount of change. So these do not add to any understanding for anyone today. 56:12 56 minutes, 12 seconds Um I mean if uh if if there is a 20 to 25% increase in the spot price then um 56:19 56 minutes, 19 seconds capacity that is contracted out for 6 months or one year will not be um in a position to take advantage of any increase in realization. Right? So I 56:27 56 minutes, 27 seconds think it would be somewhat beneficial for someone looking at it from the outside to get a sense of um some amount 56:36 56 minutes, 36 seconds of revenue uplift or exposure to um increased realizations. 56:45 56 minutes, 45 seconds Uh so like that's a very granular set of understanding that you are taking on a 56:52 56 minutes, 52 seconds very small uh size of the uh uh infrastructure installation base compared to the global peers. So I think 57:01 57 minutes, 1 second like we would like to uh avoid this conversation for today and as we grow in size then we we kind of like talk more 57:10 57 minutes, 10 seconds about that okay like uh now these numbers actually mean something and they are making sense to uh someone on the 57:17 57 minutes, 17 seconds street. So like today these numbers would not make much sense. 57:22 57 minutes, 22 seconds Okay. Sure. Uh thanks a lot. Um and then the second question was um so you talked about higher value tokens. Um so could 57:30 57 minutes, 30 seconds you provide some um color on which industries are sort of taking advantage of higher value tokens uh from your 57:39 57 minutes, 39 seconds vantage point? Uh not not really. I think it is more it's more closer home than that in terms of like okay what's a higher value token? 57:48 57 minutes, 48 seconds So like asking a question like okay uh what is closer to a search query is like 57:54 57 minutes, 54 seconds a reasonably low value token. uh someone using the tokens to solve a problem that is paying the bills for a company is 58:03 58 minutes, 3 seconds like a completely different value conception for those tokens. So uh answering a basic transactional support 58:10 58 minutes, 10 seconds query is like a reasonably low value token. Uh being able to uh answer a more 58:17 58 minutes, 17 seconds complicated query or being able to confirm a more complicated transaction with a customer on call is like a higher value token. So like there are like 58:27 58 minutes, 27 seconds infinite number of variations of how people use AI and uh the key is to find 58:34 58 minutes, 34 seconds those customers who are utilizing the higher value token where they are not worried about like okay what's my ultimate per hour GPU price but like 58:42 58 minutes, 42 seconds what's the cost of uh my solving a particular problem and generating a higher ROI. So again we are not like 58:50 58 minutes, 50 seconds directly involved all the time in identification of higher value tokens and kind of like being able to uh 58:58 58 minutes, 58 seconds strategize in a very very static manner that oh they should go and sell to this customer they'll probably pay higher for this particular GPU like it's a broad 59:06 59 minutes, 6 seconds industry trend where we are seeing increasingly the demand coming from customers uh who are not price sensitive 59:13 59 minutes, 13 seconds because of their ability to generate or consume higher value tokens. 59:18 59 minutes, 18 seconds uh through a combination of their own software or proprietary or open source software that is available uh extent in the market today. 59:28 59 minutes, 28 seconds Um so in that case um what is uh what is E2's role in um this high value 59:36 59 minutes, 36 seconds tokenization? I mean that would just be an outcome of if the person who is renting a GPU or uh taking on GPU 59:43 59 minutes, 43 seconds capacity is able to extract higher value they'll be willing to pay more right so the price discovery should this take 59:51 59 minutes, 51 seconds care of who has who is generating higher value token mean I wasn't too clear on how um E2 is sort of working towards 59:59 59 minutes, 59 seconds higher value how you are able to quickly get to that point of being able to generate and utilize those tokens for your business 1:00:06 1 hour, 6 seconds like so it's it's a fundamentally uh how to build on various parts of the uh software and hardware stack to make 1:00:16 1 hour, 16 seconds it available quicker I think that's where the uh role is played by to you 1:00:23 1 hour, 23 seconds understood okay that's it from my side thank you very much okay thank you 1:00:29 1 hour, 29 seconds thank you ladies and gentlemen due to the time constraint that was the last question for today I would Victor in the conference over to the management for the closing remarks. Over to you team. 1:00:41 1 hour, 41 seconds Yeah. Thank you uh to all our investors, all our team members, all our customers, all our ecosystem partners for all the 1:00:48 1 hour, 48 seconds support uh you have all extended to E2 network over uh last many years. Uh it's been a overwhelming amount of love for 1:00:57 1 hour, 57 seconds us uh that we are seeing in the uh community and uh overall community and uh uh we wish you all like a uh great 1:01:07 1 hour, 1 minute, 7 seconds day ahead and uh thank you for patiently listening to our call and uh we would like to uh kind of like once more uh 1:01:17 1 hour, 1 minute, 17 seconds thank everyone and uh end this call. 1:01:24 1 hour, 1 minute, 24 seconds Thank you. 1:01:27 1 hour, 1 minute, 27 seconds Thank you sir. Ladies and gentlemen, on behalf of E2E Networks Limited, that concludes this conference. Thank you for joining us.