E2E Networks Limited (E2E) Earnings Call Transcript

Ladies and gentlemen, good day, and welcome to E2E Networks Limited Q3 FY '26 Earnings Conference Call hosted by Go India Advisors LLP. [Operator Instructions] Please note that this conference is being recorded. I now hand the conference over to Ms. Soumya Chhajed from Go India Advisors. Thank you, and over to you, ma'am.

Thank you, Danish, and good morning, everyone. We welcome you to E2E Networks Limited Q3 and 9 Months FY '26 Earnings Con Call. We have with us on call today Mr. Tarun Dua, the Managing Director; Mr. Nitin Jain, the CFO; and Mr. Ronit Gaba, the Company Secretary. I must remind you that the discussion on today's call may include certain forward-looking statements and must be viewed in conjunction with the risk that the company may face. Thank you, and over to you, sir.

Thank you, Soumya, and thank you, Danish, and welcome, everyone, to the Q3 earnings call for the financial year ending March 26. Let me start by reintroducing E2E Networks for the new business. So E2E was founded in 2009 and is amongst one of the largest AI/ML cloud GPU players in India currently. We have been a champion of contactless computing since inception. And many of the current large unicorns have used E2E Networks compute infrastructure when they scaled up themselves from start-up to unicorn stage. Our infrastructure services platform, including TIR for cloud GPU, including our sovereign cloud platform, which can work on-premise as well as like remotely, like all of that supports like, all of that supports like majority of the workloads for CPU computing and GPU computing, including training, inference, or DBs, model endpoint deployment and allows for scaling up and a number of other services that are expected from public and private clouds today. Now today, majority of our workloads belong to the AI/ML domain. And we are working with really scaled up customers using our platform and the bare metal platform to essentially allow for high-performance training and high-performance inference workloads, which are scaled up to hundreds of GPUs. Now some of the latest developments related to E2E that we wanted to cover in the call. So India AI orders, as you are already aware, they have gone into implementation and majority of those workloads are expected to go live around end of January 2026. The Jarvis Lab acquisition of assets was completed in this quarter. While we don't expect like a very substantial contribution to the revenue in Q4, this is a technology-centric move that rapidly positions us to scale up in the global market. Now there has been substantial progress on the MRR target we set for ourselves for March '26. I think we are like 70%, 75% or even 80% there. And we continue to hold our target of like being able to hit somewhere close to INR 35 crores to INR 40 crores of MRR around March 2026 or thereabouts. Now all our partnerships are going great, like we have continued to strengthen all our partnerships. And like we have already seen some progress in terms of actual conversions happening amongst the enterprise customers through one of our largest partnerships with L&T, and we continue to expect like more conversions happening in the near and the medium-term future. Now we continue to build more Blackwell capacity as we have announced from time to time. And most of the new capacity is going to operate from our Chennai facility. Like once again, I would like to briefly touch upon our sovereign AI platform. So we remain like one of the unique businesses in India, who has built like infrastructure software product, sitting here in India with a team, which has the experience of building these products all the way back since like 2014 onwards since when we started the development of our cloud platform. And so once again, I would like to reiterate the levels of sovereignty that we as a nation are trying to achieve. Of course, there are like a number of government initiatives, which intersect with our initiatives on the push towards sovereignty. So like -- one is, of course, like the first level of sovereignty that I think like we have done a very good job of as a country, localization of all our data in India. Now second part is that like is -- are the physical buildings, the data centers where the data is being stored, do they belong to primarily Indian companies, like we do that checkmark by having partnered with L&T. So third part is that like are you dependent majorly on open source and your own proprietary software built in India? Or are you dependent on software which could have compliance implications from outside the country? And fourth is, of course, that like for cloud providers who are based out of India, especially based out of some of the Western countries, they have to comply with global laws that apply to even to the data that is stored in India on their cloud platforms. So obviously, as a cloud-based out of India and headquartered in India, like we provide a complete sovereignty over there to our customers. Then the other levels of sovereignty are which obviously, there are initiatives in this country going on, especially the semiconductor mission, the quantum mission, the AI mission. So the models of sovereignty is being built with the help of India AI mission. The semiconductor sovereignty is being built with the semiconductor mission. The quantum sovereignty is being built with the quantum mission. So overall, we kind of like find the theme of sovereignty is something that like we have been working on without using the word for a long time, and now we have started using the word. And we would continue to do that to expand our services to AI natives, the enterprises, government customers and AI natives like which are global in nature and spread all across the globe. So we continue to deliver on the promise of sovereign AI for all our customers. So with that note, I would like to hand over the call to our CFO, Mr. Nitin Jain, to give some more specific inputs about the progress made in this particular quarter. Over to you, Nitin.

Thank you, Tarun. Good morning, everyone. Thank you for joining us today and for continued interest in the E2E Networks. I would walk you through our financial performance for Q3 '26. Q3 FY '26 has been a strong operational quarter for E2E Networks, marked by robust revenue growth, improved operating leverage and the continuous investment for future scale. Our operational revenue stood at INR 700 million, registering a growth of 68.3% year-on-year and 59.8% quarter-on-quarter growth. The growth is driven by higher capacity utilization, increased enterprise workloads and early traction for strategic and government-linked contracts. EBITDA for the quarter stands at INR 396 million, up by 60.9% year-on-year and 120.2% quarter-on-quarter. EBITDA margin came at 56.6%, reflecting strong operating leverage. From a profitability standpoint, we reported a PAT loss of INR 57 million. On a quarter-on-quarter basis, PAT has improved by 58%, indicating earnings are already trending positive as the revenue scale on the newly added capacity. The primary reason for the loss is on account of higher depreciation, which has increased by INR 476 million, reflecting commissioning of large GPU deployed in FY '25 and '26. Finance cost increasing as we draw our term loan for the infrastructure expansion. On a cost structure basis, our total expense for the quarter stands at INR 304 million, which is in line with the business expansion. We continue to maintain tight control on the cost and the inherent scalability of our business model allows incremental revenue to flow through at higher margins. Our monthly revenue run rate has touched INR 280 million in December 2025, with India AI mission contracts already initiated, L&T enterprise engagement gaining traction, we expect a strong revenue momentum to continue. To summarize, Q3 demonstrates strong top line growth and operating strength. Our focus remains on maximizing the GPU utilization, scaling enterprise and sovereign cloud workloads and delivering sustainable profitability as our capacity matures. We are confident that E2E Networks is well positioned to bring a long-term sustainable AI leader from India, supported by strong balance sheet and growing revenue visibility. So now I would hand over to the moderator for question-and-answer session.

[Operator Instructions] Our first question comes from the line of Keshav from Niveshaay.

Congrats on the good set of numbers. So sir, could you please help us understand like E2E's approach to the cloud reliability? So like in terms of any downtime or service disruption, like what redundancy measures are in the place that we have taken, and like how do we typically compensate our customers?

See, we typically compensate our customers in case of any service disruption or downtime due to our reasons based on the SLA. And typically, what happens is we have a range of customers. So where there are customers who implement like a full single point of failure, disaster recovery scenarios, test their backups regularly. So ultimately, we are a self-service public cloud, where obviously, we provide [Technical Difficulty] to our customers across multiple locations to kind of like do a multi-location installation with disaster recovery and no single point of failure. And then obviously, like in case like anything happens, like our entire team is there to support the customers to the best of their ability. So like we bring in everything according to like how all the public clouds like throughout the world do.

And sir, like did we have any update on the software business? Like are there any POCs that are currently running on the software side?

So like no immediate updates in terms of like basically like we continue to do what we have been doing. We continue to build our software. We continue to talk to a lot of customers, enterprise customers, and we keep validating the requirements. And then we continue to kind of like demonstrate our capabilities. And hopefully, in the medium term, we'll start seeing some action on the software licensing front for the sovereign AI cloud, AI cloud platform that we are packaging currently. And we continue to heavily invest into that.

And sir, what is the expected time line for Blackwell to go live and from which quarter we can expect the revenue to kick in?

So like we are expecting in the -- hopefully, before the end of Q4, we should be able to go live with the Blackwell that we have already procured, and we continue to expand the Blackwell capacity.

Our next question comes from the line of Bhavya Gandhi from Bajaj Alternates.

Just wanted to understand how is the payment cycle with respect to India AI mission? How frequently have they started paying in terms of service received by the customer?

India AI Mission has actually recalibrated in the latest documents that they have produced that the payment cycle, they have shifted from a quarterly to a monthly. So we hope to see the benefits of that soon. So we are working towards kind of like finalizing and figuring out like basically like how those payment cycles would eventually work for us. But like I believe that like that shift is inevitable from quarterly to monthly. So that's good for us.

And when can we expect the next lot of Blackwell GPUs to arrive because I think we are buying in terms of 1000 watts...

We have already received in our Chennai location 1,024 B200 GPUs. And we are in the process of kind of like deploying those. And hopefully, in the Q4 itself, like we expect to see the deployment completed and then we can start seeing like some action and traction in terms of orders coming in from various stages.

And so can we expect like order win could be possible in Q1 FY '27 onwards for these 1,000 Blackwell GPUs?

Keeping our fingers crossed, like we are hoping as soon as possible.

And have we ordered the next batch of Blackwell GPUs for further expansion?

So that's an ongoing process. So like there are multiple types of GPUs in Blackwell, so we are working on expanding [Technical Difficulty] our capacity also, and we will continue to expand the capacity on other SKUs as well.

And just one last thing with respect to EBITDA margins, if you can just provide some light when can we touch the 70% EBITDA margin? If you look on a Y-o-Y basis, there has been dip in the EBITDA margins. So if you can just highlight what was the...

I think like quarter-on-quarter now, like we have expanded the EBITDA margins. And I think it's a matter of establishing the scale at which point like we will be able to hit the expected number as per the business, which is closer to 70 than closer to 55 to 60 today.

And would you like to guide for next year MRR figure?

So I guess we'll wait for the next call to do that guidance, I guess.

Our next question comes from the line of Bharat Gulati from Dalal & Broacha.

I just have a couple of questions. Firstly, can you line out our CapEx for -- what will our CapEx end at for '26 and then '27? And then if you could also give us a CapEx guidance for '28, if possible?

So I guess like we'll need to kind of like wait for a couple of weeks for us to kind of like delineate all these things. So we are kind of like working with our very strong partners and suppliers to kind of like build up the capacity. So we don't want to prematurely announce things today. So let us wait for a couple of weeks, and we will come up with like basically like what all we are doing.

So any guidance you can give for '26 and '27, a broad guidance also, what we'll end at '26 at least what?

So we don't want to give the guidance today. So hopefully, like, again, like give us like a couple of weeks or like maybe by the time we are talking about the next quarter, like we'll put together the guidance.

Sir, on the 5,000-plus GPUs that we've mentioned in our PPT, so these GPUs that we'll be deploying will be Blackwell. So do we see orders coming in?

Majorly, the expansion is from the Blackwell variety of GPUs.

Yes. So sir, do we see orders coming on also in Q4? Or will this roll over into Q1 of FY '27 then?

Let's see, we'll continue updating everyone as we kind of like go into the deployment phase. So we've just kind of like received the hardware in the last couple of days. So I think like give us like some time, we will update everyone.

All right. And just sir, one last question. In terms of the L&T order that we've won in this quarter, are there any other such orders in pipeline for the coming near term?

See, like we obviously continue to work on with a lot of customers through L&T and directly as well. So obviously, like as and when the orders fructify, we will inform everyone. So that's the goal.

Our next question comes from the line of Varun Gandhi from Fident Asset Management.

Tarun, could you help me understand how much of India AI mission contribution is happening to our reported MRR, if any? And additionally, with the latest B200 chips, what would be the new peak MRR that we expect to achieve?

Okay. So like mostly in the last quarter, like the India AI mission contribution has been muted. I think like in December, it would have been the most value. But like we expect, obviously, Q4 to be massively different as the -- both the customers won through India AI mission continue to scale up in January. So that MRR percentage number should be like much higher. So yes. So basically, like that's an ongoing process. Now second, like we don't want to kind of like give any kind of guidance today, like I think like over next couple of weeks in terms of capacity expansion, what we are thinking and maybe...

Sorry, I just need what is peak MRR capacity that you would expect to achieve? No guidance as such? Just the peak MRR.

There is no upper limit for us. Like we don't want to limit ourselves by saying that like this is what we are going to only do. So like we don't have an upper limit in our mind.

All right. Secondly, on the India AI mission contracts, these are primarily focused on training workloads right now, which we know are typically....

Majority of India AI mission workloads are focused on building the LLM models, sovereign LLM models for India. So I think like majority of that workload is going to be training workloads. So that's the main focus of India AI mission as of today.

So what is the likelihood that the same LLM developers would also contract with us for inference workloads because the implication here is that inference will have stable revenue.

Very, very likely. We are already working with the LLM builders to kind of like run their inference workloads with us.

Got you. And my third question is...

But some of them are already running their inference workload with us and many of the new ones will hopefully are already working with our customers through our partners for the inference workloads, and we continue to see the acceleration of this trend.

So if someone is contracting for training workloads, there's a very high possibility that they would also -- probability rather that they would also contract for inferencing workloads.

Absolutely, Absolutely.

Also, Tarun, in December, we faced a major outage in the Mumbai servers. So just wanted to seek some clarification on your end and if there's any major client loss due to the outage? And what steps have we taken to prevent such prolonged outages in the future?

Sure, sure, sure. So there are a couple of points over there. So basically, like our plan was always to kind of like migrate the customers from Mumbai, which is a much smaller location than our Delhi NCR and Chennai locations. So the plan was to definitely migrate those customers out. Now we are working on the plan to migrate them on to newer and much stabler infrastructure. On the software and operational side, like we have made quite a few changes in terms of like increasing the investments into site reliability engineering and working with like some of the back-end vendors to kind of like strengthen our overall infrastructure. From an impact point of view, I think like over a period of time, it would become clear like what is the impact we are seeing. But as I said earlier that like this was like a substantially smaller location overall for our infrastructure. So we don't see -- we don't expect to see a material revenue impact over the medium term and the long term. But that being said, like we are very, very conscious of working very hard towards making sure that like these kind of incidents are completely avoided, and we are able to do better in terms of like providing support to our customers before an incident happens in terms of helping build an architecture that will withstand any kind of failures on a single zone for us in the future.

Understood. And if I could squeeze one last question. Since we spoke in the previous quarter, the AI accelerator industry has witnessed a significant update with the Google's CPUs becoming the centerpiece of conversation for better cost efficiency. And NVIDIA's hardware mode is now being contested -- and the broad expectation is that ASICs in general, the adoption would pick up materially. Now my question to you as the management of E2E is, would you be also open to adopting ASICs within your infrastructure if the industry trends towards that direction? Or would E2E remain an NVIDIA exclusive partner? And I ask this because both choices have clear implications on business dynamics, including CapEx, ROICs. And I also understand that it's very premature to state any conclusive comments, but all I see is your thoughts on the topic.

Fair, fair, fair. So one is accelerated computing is here to stay. Accelerated computing is going to be majority of the workloads going online in the future. Second part is that like what is it that the customers are demanding today? What is the road map of our customers what is it that our customers are looking at? I think that is more relevant to basically like what we end up adding the capacity on. So like in the near future, we again continue to see that like basically NVIDIA will be majority of our portfolio, a vast majority of our portfolio. And that is based on the inputs that we continue to get from our customers. See basically like if the customer feedback anytime changes where a customer -- when customers start demanding that like, look, can you consider X or Y or Z vendor, like obviously, we are very, very happy to provide what the customers want. So what we are seeing today as of today is what we have spoken about.

[Operator Instructions] Our next question comes from the line of Krupa Desai from Electrum PMS.

My first question was how much CapEx have we done for the procurement of 1024 Blackwell series? And how have we funded that?

Okay. I'll let Nitin answer this question.

So for the Blackwell series, roughly the CapEx would be in the range of INR 600 crores to INR 650 crores, part of which is funded through a term loan, which would we have taken from the Axis and HDFC Bank.

Okay. And sir, last time, you had guided that INR 35 crores to INR 40 crores would be the peak MRR for the current 3,600 GPUs. So do we still stick to that guidance?

Yes, more or less.

And sir, on that, we had guided that at INR 35 crores to INR 40 crores MRR, INR 65 crores to INR 70 crores would be the EBITDA margins. So do we expect in next quarter, we can hit that?

We are definitely hoping for that.

And one more, what was the utilization rate this quarter?

I think at the December end, we are touching the utilization roughly around 60% to 65%.

Okay. And sir, the per hour rent for Blackwell series would be much higher than the current. So how much could that be? Any ballpark number?

Broadly, what we are seeing in the international market today is like anywhere between like $3 to $4 approximately for slightly longer-term contracts. So like we are also hoping to kind of achieve those kind of numbers.

Our next question comes from the line of Neil Munot from PICO Capital.

Sir, am I audible?

Yes, yes. Please go ahead.

So other than the enterprise client that you've got from L&T, can you give any sense of [Technical Difficulty] because...

I'm so sorry, Mr. Neil, but your voice is breaking. We can't hear you properly.

Is it better now?

Continue your question, we'll see.

Yes. So sir, I was asking other than our enterprise clients that we got through L&T, can you give us a sense of our other enterprise client conversions, like not exact numbers, but since we are 6 months [Technical Difficulty] from start-ups to enterprises, a broad sense.

No, we continue to work with like a lot of enterprises and AI native companies, and we continue to see quite a few conversions. Now when the conversions are substantial, obviously, we'll continue to inform.

Okay. And sir, the new capacity of Blackwells that you've acquired, are they solely for the India AI mission or we are seeing tractions for POCs out of the India AI mission?

I think like a majority of those would go to India AI mission for sure. So -- but like we continue to build capacity even beyond India AI mission as well.

And sir, the revenue potential from these 1,000 Blackwell GPUs that you sent, what is like the peak sense that we could generate from this? Any number?