Chennai DC shifted from 'AI unicorn interest / IaaS launch' hype (Q3FY25) to operational revenue in Nov (Q2FY26).
- 5 year data center — answer hedged.
- 5 year data center 2 — answer hedged.
- Edge data center government — answer hedged.
Given that there are multiple things which are happening in Chennai, new plant or new setup coming in Kolkata and then multiple edge data centers. Next 5 years, what kind of revenue, EBITDA and capital employed will go into data centers?
Resham, I will say that you are asking a young person. It's a very mature question, number one. And two, you can on an average stake in next -- if everything goes as projected, data center will become the face of the company. That's how we are projecting going forward. And if not in 5 years, at least by 2030, we should have a capacity of no less than 250 megawatts at least in locations no less than 100 countrywide, including as data centers and we are definitely will be looking for revenue of no less than, I would say, almost about $0.25 billion, $250 million per year if not more out of this capacity.
Continuation on data center 5-year revenue/EBITDA projections.
So I think the way things are going today, it's very difficult to point out because the industry is very, very dynamic, Resham. If you see what happened over the last 15 days in the industry from US announcing $500 billion for AI infrastructure and then the announcement of the largest data center in Jamnagar and then going to what we all heard about from China, which is DeepSeek, that is how dynamic and disruptive this industry is. Ultimately, the infrastructure when we talk about for any such application, it is data centers only. In such a disruptive industry, it gets very, very difficult for us to become very predictable in nature on seeing what we can feel in what it would be in next 4 to 5 years. But in most conservative sense, as EPG put it up that we should see at least 200 to 250 megawatts come in by Techno Electric and at least $250 million of revenue flowing in from data center and its allied services. But let me say that the way this industry is unfolding, it has immense opportunity and one has to keep unlocking them and growing with them. And ultimately, it may end up sounding and shaping very different than what we all envisage today.
These orders that we have for the edge kind of data centers and the other government contracts. In terms of margins, how do we expect them to pan out given the nature of your contracts and the environment that you see for procurement, etcetera, could be a longer execution timeline?
Edge data centers are concessions to us for 20 years, extendable by 5 years more. So these are revenue-sharing agreements with the realtor and margins will be all guided by the tariffs or leasing rentals on which we can deploy these facilities and for what applications they get deployed. So as Ankit has already mentioned to you, these are bit growing up, disrupting but rewarding situations. So the margins are difficult to say in a concession because they will be all measured as IRR of the investments in this case, which we hope it will be not less than 20%.
We have followed the company for many years. And in the past, we have been very careful about ensuring that we take higher margin projects. So this seems to be slightly more open-ended and the question of pricing and all that to some extent doesn't stay in our hands. So just wanted to check how you guys are looking at that decision.
It is more rewarding than our EPC. That's what I'll say.
For Chennai, which of the services do you want to sell and maybe for the next round for Kolkata, which of the services do you want to sell?
So when we talk about any data center, we will build our portfolio of services around them. So for all the data centers that we come up with whether it is Chennai, whether it is Kolkata or whether it is the edge data centers, we will be having the operational bandwidth to provide all these services from all of these data centers.
How are you picking and choosing these services and customers?
So we would like to address all sorts of customers. We are not limiting ourselves to any particular type of customer or a customer which prefers a particular type of service. So for instance, in Chennai today, we are in discussion for colocation service. We are also in discussion for bare metal services. Similarly, in Calcutta, we are in discussion for bare metal service to a content delivery network while we are in discussion with a couple of customers for private cloud services out of a data center in Gurgaon and Mumbai. So they are different depending on the customer and their requirements.
As you mentioned we are also working on Infrastructure as a Service. How much is the additional cost which is required for a per megawatt basis to get the server set up? And how much is the rental which we can expect for the same?
So generally, Infrastructure as a Service can be set up in 2 models, which can be largely capex-based or opex-based. We would possibly adopt a more opex model for infrastructure as a service rather than capex.
Just in continuity of the previous participant's question, wanted to understand the rentals that we are projecting from the IaaS services. And wanted to understand by when we may be planning to bring up the IaaS service on a full-fledged basis?
So saying that any kind of value-added services or services out of Infrastructure as a service, it's very difficult to today estimate a revenue out of it or a margin out of it because these opportunities are today limited with us, and they will take time to build. And at least for the initial year or 2, they will be contributing a few percentage only to the colocation services.
Last year, there were certain supply chain kind of challenges whereby getting the material itself was not easy. Are you seeing some of those challenges residing and receding? And how is the execution looking like from your side?
Those challenges will remain and it continues, number one. But probably Techno is the oldest player in this segment and our relationships with all the suppliers is of a partnership. And we are also consuming no less than 10% of the supply chain in this segment today in the market. And this is likely to be 15% by next year. So we are hopeful despite all these challenges, we'll be able to get loyalty from our partners. The supply chain will remain a concern for the overall market, but not challenging I will say. But largely the land parcels where we got delayed in the first half of the year was the land parcels are becoming difficult for the concession [baggers 0:41:58] to arrange and hand over to us for the deployment of the facility. That is delayed in the first 2 quarters. Now all those are available, except 1 or 2 still to be given by Power Grid and a private player. But now we mostly have those land parcels and hence, you see the growth in the output also.
Continuation on BOT contracts and margin comparability vs EPC.
Yes, because these are BOT contracts, sort of a BOT contract over a period of 20, 25 years, one has to look at it more from the perspective of cash flow and IRR. We are hopeful of achieving high teens or early 20s in IRR for these projects. So there is not real upfronting of any margin as an EPC that we can really compare it to a transmission EPC that we do additionally.
On the data center business, what is the kind of capital that we have employed already? And going forward, what is the capital allocation plan for the Chennai data center?
Yes. We have deployed INR450 crores by now. That is for Phase 1 and maybe another balancing requirement of INR25 crores plus/minus more in closing out. The next phase we will take up only after this phase is fully deployed. So you can say that we will be spending another about INR500 crores to INR600 crores over the next 2 years, say, by March '26. Chennai data center, we will see another INR600 crores. But that will start only second half of next year.
The revenue accretion from this capital that we've employed already of INR450 crores, so that revenue accretion we're expecting from maybe H2 FY '26 or FY '27? Is that the correct understanding?
It starts off '25-'26. Q2, you can say it will start flowing.
On the non-hyperscale data center, like the edge data centers like Gurgaon, Bombay, and the RailTel ones. What is the kind of capital allocation plan there? Is there a plan in terms of capital allocation? Or it is like as and when we see a good opportunity in the edge data centers, we'll go for it?
No. So let me clarify this for you regarding Chennai first and then coming to the edge data centers. As mentioned, Chennai data center should start seeing revenue start flowing from second quarter of financial year '26. Going forward, one can expect INR200 crores to INR250 crores of capex behind Chennai data center annually for expansion of the phases until we have commissioned the entire 24-odd megawatts. Having said that, you'll see revenue start flowing from the second quarter of financial year '26. In regards to edge data centers, obviously, it depends on how things shape up in the industry, but we target to at least put in -- as contract requires, we have to set up 20 locations every year. But on all practical basis, we assume we will be starting anywhere between 8 to 12 locations every year, and that is due to the contract that we have with RailTel. But the locations that we choose will dictate the capex involved because each location may not be of the same size data center. Each location may be different. For instance, Gurgaon was a smaller one with 200 kilowatts, while Mumbai is a larger one, which is going up to about a little more than 0.5 megawatt. Similarly, the ones which we are discussing in Gandhinagar and Indore may go up to anywhere between 1 megawatt on an average. Similarly, Bhopal may be around 1 megawatt again. So each location is differently sized. Not all locations are the same. And depending on the number of locations we pick up in a year, the capex would be dictated accordingly. But in all possibilities, we will see at least INR150-odd crores to INR200 crores of capex annually once the RailTel project comes in a smooth operating procedure.
On the T&D EPC business, what was the order inflow number for Q3 and the 9 months FY '25?
This year as I said the total order intake will be about INR3,500 crores. We have already booked orders worth about INR750 crores by now or INR1,800 crores. And this quarter, it is around INR1,100 crores. We are also L1 in another INR1,600 crores which will be converted into orders and concessions won by the developers already. So it's in the formality of their acquiring SPV land parcel and releasing of the orders by power grid. Lately, they are believing unless they have a land parcel, they will like to delay issuing LOI. But they are committed to issue so. So it will be about INR3,500 crores this year and we should close the year with an order backlog of INR10,0000 crores plus.
This INR3,500 crores is essentially entirely T&D EPC?
Yes, absolutely EPC.
What about the non-T&D part of it like the power gen or the smart metering business?
No. Then we are executing FGD order. As you know about unexecuted part is still about INR1,250 crores that will take another 2 years to complete. Then also we have a distribution side business on the digitizing or modernizing the distribution network about INR300 crores we are doing for DVC and we expect more business in this space. Similarly, we are deploying smart meters, that is another business of INR2,500 crores. Additionally, we have our own TBCB projects with a capex of about INR700 crores plus minus.
From Q2 FY '26, Chennai data center will start contributing to the revenue. What is the anticipated contribution that will be there in FY '26 from that data center to our top line?
Honestly, we have not considered it. If something happens it will be upside. You can take it. But on an average, we do expect that we should at least target to have about INR100 crores with ranging from anywhere INR50 crores to INR100 crores minimum for the year.
This year we have a guidance of INR2250 crores on a consolidated basis. So are we sure to achieve that guidance, we have to do around INR800 crores in Q4. Are we confident enough to do that number? And secondly for next year, you guided for EPS of INR50 but this time in opening remarks you mentioned INR45. What is the reason for this decline?
No, I have not mentioned INR45. Firstly, I mentioned this year we should be doing an average of INR200 crores. And generally, Q4 is about 30% of the top line. So definitely, it will be INR750 crores to INR800 crores somewhere as a Q4 and 35 will be EPS and 25, 26 I said our top line will be about INR300 crores per month. So you can expect a top line of about INR3,500 crores, INR3,600 crores and EPS of INR50. That's what I reaffirm.
I think these are standalone numbers. What about the consolidated top line?
Sir, in consolidation there is not a very large difference because they are only meant for our own concessions. They are only accounting requirements per se. So it is not going to be very different, I will say, at least for next year.
What is the margin guidance?
It will be same about EBITDA of 13% to 14%. You can take 14% plus minus.
Question already answered.
My question has already been answered. Thank you.
Regarding data center, the services which you want to offer -- what's our view in terms of leasing out, what sort of services? I heard colocation, I heard infrastructure as a service. I just want to understand that better.
So as I was mentioning Infrastructure as a service we include bare metal services, we include cloud services, private cloud services, white glove services, amongst others.
Is there a strategy to get a reference for each of the services with a good customer or is the profitability greater?
So profitability is not that we are compromising when we are providing any of these services. All of them are equally or more profitable than each other. There is no preference of the management when it comes to choosing which service we rather choose over the other.
Question already answered.
My question are already answered.
Recently, we have heard that DeepSeek's entry into this industry will have an adverse impact on the data center companies. Are we seeing any negative impact on our business?
Actually, look at DeepSeek's and its development, I would say it's actually a very, very positive development instead of negative. I do understand where the negativity stems from, but one needs to appreciate that in a very disruptive kind of technology, which has recently got innovated, such kind of disruptions are a way of life, and we'll see more of it coming in news time-to-time. And these disruptions are actually very, very good for any industry because the more affordable and accessible the technology becomes, the cake size increases, the market size increases, and therefore, the demand always remains at peak. So there is never that any disruption causes lack of demand. So if tomorrow's infrastructure requirement reduces and prices for the end consumer improve, we'll see more market participation, and therefore, automatically, the cake size of the industry will increase. That is what happened in the telecom sector when Jio entered. It never -- while they were able to run at a bare minimum infrastructure and reduce the cost of telecom services to the end customer, but the entire market size went possibly 4x.
We have been guiding INR50 EPS for FY '26 and INR75 EPS for FY '27. And in one of the calls, even we mentioned that we are expecting INR25 to INR30 EPS only on the data center business by FY '27, FY '28. So this would be over and above our guidance of INR75 for FY '27?
No. That data center will become part of this INR75 going forward. On the EPS, yes, but not value accretive will remain in the subsidiary separately. One is revenue part, and one is value accretive because businesses are valued differently whether it is transmission or AI or data centers. What I'm saying is the value accretive part of the asset is not part of EPS. EPS is purely revenue-based.
Question on DeepSeek AI advancements on data center business.
Sir, congratulations for putting a good set of numbers. Actually, my question was regarding data centers effect of DeepSeek AI advancements on data center business, but I think already you have covered already.
We have like 2 data centers which we are working on, the Chennai data center and the Calcutta data center. On a per megawatt basis, how much is the rental which we are looking at either on a monthly basis or on an annual basis? And how much margins we foresee from that?
Yes, you can presume a rental of about -- if we are looking at a pure colocation service without any infrastructure as a service, then in that case, we can expect a revenue of about INR10 crores to INR12 crores per megawatt annually of the commissioned capacity.
In terms of the margin profile for the same, how much would it take to operate it?
So EBITDA margins of about 80%.
Can you be a bit more elaborate on the opex part? When we say opex, so will we be taking the service on rent itself?
Yes, you can possibly take the hardware and the technology, which is built on the hardware for these kind of services on more of a lease model or more like a subscription model from technology service providers and then provide it at a higher margin to the end customer.
In terms of our edge data centers, is it just a build-operate-transfer model or we will also be managing it for RailTel?
No. So it's a build-operate-transfer model where we transfer the asset back to RailTel after 25 years. For the first 25 years, we have to operate and maintain these assets as well as bring customers for these assets, provide them with O&M services or any other service that we want to, and then the revenue gets shared with RailTel.
For the IaaS services that we are planning to provide since we will be sourcing the hardware and the tech, do we have any plan in terms of will we be sourcing it from outside India or from India itself?
So see, most of these -- today, most of the technology is available in India. And there are global players as well as some domestic players, which are providing these platforms. So depending on the end customer requirement, we may source from either of them because some customers have a preference for domestic players given neighbouring country regulations and rules and some customers have a preference of global players given the majority of the technology. So it would really depend as per customer preference.
In terms of the smart metering business, what is the number of smart meters that have been given operation go-live status? And is the receipt of money for setting up the smart meters moving smoothly?
Yes, it is going very smoothly. We have already achieved this set of about 4.5 lakh meters. In the state of MP at Indore for about 1.75 lakh meters, another about 2.75 lakh meters in Srinagar and Jammu in Kashmir and they are going very well. There is no challenge. Ministry of Power conducts a review meeting every week with all these states DISCOMs, and they are very serious to make sure no wrong happens in smart meter deployment and payments.