ETERNAL LIMITED (ETERNAL) Q2 2027 Earnings Call Transcript

Ladies and gentlemen, a very good evening, and welcome to Eternal Limited's Q2 F527 earnings conference call. From Eternal's management team, we have with us today Alvindar Singh Dinsa, Akshan Goyal, and Kunal Suruk. Before we begin, a few quick announcements for the attendees. Anything said on this call, which reflects outlook for the future, or which could be construed as a forward-looking statement, may involve risks and uncertainties. Such statements or comments are not guarantees of future performance, and actual results may differ from those statements. Additionally, please note that this earnings call is scheduled for a duration of 45 minutes, and we will be starting directly with the Q&A section of the call. If you wish to ask a question, please use the raise hand feature available on your Zoom dashboard. We will announce your name on the call and unmute your line, post which you can proceed with your question. We will wait for a minute while the question queue assembles. The first question is from the line of Garima Mishra from Kotak. Please go ahead.

Hi, thank you so much for the opportunity. First question, could you comment on the competitive scenario in the food delivery business? Has discounting been more aggressive in 2Q than it was in the previous few quarters? In this context, should we assume that margins of the food delivery business have peaked out and any surplus would just be reinvested towards growth?

Hi, Varimha. Akshan here. So I think generally, the comparative intensity last quarter in food delivery was higher compared to the previous few quarters. And I think partly also driven by Zomato. So I think we've been aggressively investing in growth and we are seeing good outcomes on the market. On certain investments that we've experimented with in the last few quarters, they're paying off well. So both new customer acquisition as well as in driving frequency of our existing customer base, we have increased the amount of investments we have done consistently over the last few quarters. And I think that trend is likely to continue. You know, directionally, we want to maintain margins broadly, and reinvest in growth as much as it makes sense to continue to drive the growth upwards. I think that's what the operating principle is. So hard to project or comment on margin trajectory every quarter. But broadly, I think next one or two years, we expect the margins to remain flattish.

Sure, that's helpful. Quick commerce business, of course, the new store edition targets that you've laid out are more aggressive than they have been in the past and imply a higher run rate of store editions in FY28 versus FY27. So, what is giving you this confidence? Of course, there are some, you know, things you've laid out already in the letter. But, you know, what kind of geographies would this entail expansion into? And does this automatically mean that tier two, tier three cities would just account for a lion's share of these incremental store additions, particularly those in FY28? Great.

That's right, Darima. I think focus, directionally, I think we've also mentioned in the letter that a large part of the bulk of the expansion of these new stores will be in smaller cities. So we do expect that to be the focus from here on, and we are seeing great outcomes in terms of payback of these stores in these markets as we've expanded in the last few quarters.

Great. That's it from me for now. I'll come back in the queue.

Thank you.

Thank you. Next question is from the line of Aditya Soman from CLSA. Please go ahead.

Yeah, hi. Good evening. So two questions. So firstly, on dark AI, so while you've included some clues on it being sort of a privacy-focused LLM for power users, first, is that correct? And second, can you elaborate a little more on what capex this could entail, how you see monetization? And then just firstly, is this within Eternal? And on QuickCommerce, again, just dialing up down on that store expansion, on the cost side, so is there a meaningful difference in the cost to serve a smaller town to date? And how do you expect to sort of overcome that over time? Or it's just a function of you having a higher market share in these smaller towns? Thanks.

Yeah. Hi, Ditya. So, yeah, dark.ai is definitely an eternal. I think anything in this letter, safe to assume, is part of eternal. We, I mean, on your other question on this, at this point, we don't expect, I mean, like, first of all, just to clarify, all the investments that we've been doing, this product has been in the works for the last about a year. And all the investment, it's not monetized yet, right? So all the costs here are accounted for in our other segment, right? So they're already carved out. They're not very meaningful at this stage. And we don't expect that capex or cost, at least as of now, we don't expect to become meaningful. Unless, of course, we receive a blockbuster response to this product, then there might be a scenario under which we may have to invest for infrastructure, GPUs, etc. And we'll update everyone if that happens. But I think as a base case, we think this could be, you know, high gross margin business, which could fund itself largely and get to profitability pretty soon. So that's the outlook right now. But I think very early, once we launch, we'll sort of iterate on what's the right monetization here and how much do we need to invest to continue to grow the business. So I think we'll take those calls closer to the launch time. And maybe after that, once we get some feedback from our customers. But at this point, I think nothing that we want to highlight. On your question on tier two markets, I think what we're trying to ask is if these markets are, from a unit economic standpoint, any different, better, worse off than larger markets, or is it just that we are getting into these markets so that we have the first mover advantage of market share? So I think what we've seen so far is that economics are pretty robust. As I said, paybacks are extremely good in these markets. And the other point is also true. These markets are less competitive. And I think going in first does give one an advantage of locking in the early users of those markets. So, I think both of these things are true, and that's why we want to double down on expansion into these markets at this point.

Perfect. And maybe if I can just follow up on the first one. So, on this RKI, this is likely to be open to everyone, or you limit it being open to a few customers? Or, again, that's something that you'd rather disclose when it happens?

Yeah, we'll see. I think we are still a few months away from launch. So, but we'll see how the go-to market is. But, I mean, long term, it's a product open for all. I think the reason we've said it's for power users is it could be priced in a way where it may not become a product that everyone wants to use. So, we'll see. I think we're still debating this. We don't know.

Understood. Here. I'll come back in the queue for a while.

Thank you. Next question is from the line of Jignanshu Gaur from Bernstein. Please go ahead.

Thank you for the opportunity. Just two quick questions on the quick commerce business, especially in the context of expansion into tier two cities, and in the context of increasing assortment. How do we manage or what is our view on the 1P versus 3P here? So I think a few quarters ago, Akshant, you had mentioned that we'll keep 90-95% of our SKUs as 1P and the rest as 3P. I just want to understand how to look at working capital in the context of expansion outside to smaller cities and more assortment.

Yeah, so Jignanshu, I think we are at that steady state now, where about 95% is now 1P. And so we don't expect any more incremental working capital change on that account, if that was your question.

Okay, sure. The second point is on QuickCommerce profit margin profile, and specifically the indirect cost structure. How do we understand this? Because then, again, despite the excellent growth that we are seeing, I can end the larger dark store addition into smaller towns. I think the contribution profit margin is explainable. But the sort of 15 to 20% quarter-on-quarter growth that we are seeing on indirect cost, how do we think about this going forward?

So I think a lot of the cost below contribution margin is also variable. I think you mentioned it in the past also everything above contribution is not variable and everything below contribution is not fixed. And so, for example, you know, technology server costs, they are below contribution. So as the business scales, those costs will go up. And some of those costs are priced in dollars. So exchange rate also doesn't help there. And so yeah, so I think broadly, what we should look at is the overall adjusted EBITDA margin. And a lot of it, the growth, the improvement we see there is on account of operating leverage. And as until the time we continue expanding, I think that that is what will drive most of the margin expansion. And at some point, as the expansion will become a smaller part of the business, we will see incremental improvement on account of variable costs coming down also.

Okay, fair. Helpful. Just a very small one, if you can. Are we selling anything different in a smaller town versus what we are selling in a metro like NCR? Is the customer uptake of product categories different? Thank you. Thank you, Alvindar. Thanks for the opportunity. I'll come back.

Thank you. Next question is from the line of Gaurav Malhotra from Axis. Please go ahead.

Yeah, hi. Thanks for the opportunity. I just wanted to, you know, just touch base on the, on question too, you know, where you have detailed out how you guys think about agentic commerce. So a couple of questions over there. You know, first of all, you know, have you guys opened up yourself to agentic traffic? I see in some places you have. So then the question is then why do it right now, right? When things are still in an evolving stage and you want to sort of build your own kind of a vertical agent. So just wanted to get some sense on that perspective, please.

So, Gaurav, I think it's very early, and we want to learn and experiment. So, anyways, in India, the agentic commerce adoption is, I think, close to zero. So, I think we want to be prepared for all scenarios. If it does change, then we want to be there for our customers. Anyways, we're not saying agent e-commerce is not a use case. I think that it will be relevant for some customers and we want to be there for those customers on those platforms, right? So these are all iterative things. And we should not, I mean, these things can change over time, right? So I generally agree with what you're saying. And what we're doing right now, I think, has no bearing on how we think about the market and what we want to do in the future.

So you're saying at some point in time, if you see market evolving in a particular way, you may actually block off this agent traffic also. That's an option which sort of exists, say, at a later stage.

Yeah, we'll think about it. I think once it's a relevant question right now, this question is not even relevant in the Indian context. I think things are playing out differently in US and in other markets, and we are learning from what's happening there. And we're hoping by the time I mean, we think that that market will lead how things shape up, for example, in India, because India will follow in some sense. So I think most of this debate would get settled before anything meaningful action or any meaningful change in customer behavior.

Okay, last question on food. You know, we see new players sort of entering. So, are we picking up anything as to, you know, what is it that has suddenly got everyone, not everyone, few players interested in food. It's a fairly penetrated category. So, it's surprising that people are planning to enter this category at this juncture.

I mean, yeah, it doesn't make sense to us at all as well. We are equally surprised. Okay, cool. Thank you.

Thank you. Next question is from the line of Manish Haddukia from Goldman. Please go ahead.

Hi, good evening. Thank you for taking my questions. I have a couple of questions on QuickCommerce and then one on DarkAI. On QuickCommerce, I wanted to just probe a little bit more on your comment that you now have a better read-on competition. Maybe Akshant or Akshant, if you can just elaborate on what exactly do you mean by that?

I think what we mean is that we went through a period of 12 months, last 12 months, where I would say the competitive intensity significantly increased in quick commerce, including new players entering and the market becoming five, six, seven players, right? So we have that 12-month period behind us. We know how it impacted our business. We know our strengths versus other players. We have some idea of how are other players thinking about this market? What are they investing on? What are they building? What's their strategy, etc.? So I think that all of that information and data, I think, helps us feel more confident about re-accelerating expansion, right? So like doing this last year would have been fraught with risk because we don't know how some of these things would have panned out. And, you know, payback on investments are very important for us, right? So if we were in a period last 12 months where the visibility on those things reduced. But now we feel it's, as we mentioned last time, also competition is more predictable now. And that provides an opportunity for us to become more aggressive on infrastructure expansion.

Very clear. And you call that out in the context of your growth and ability to expand in a clearer manner versus what you could, let's say, 12 months ago. But does that also mean that that better return competition also allows your gives you more comfort around your billion dollar EBITDA guidance that you have for FI29? Or like, from your standpoint, that's not really material change versus what you thought a few months ago?

Yeah, I mean, I would prefer the second option of your answer. I have no comments on that at this point. I think very early on that three-year-old journey, I think we're still two quarters in. So, don't want to jump too much into the future. I think we maintain that guidance as well.

Very clear. And then when you're guiding for growth to pick up in Blinkit from the 79%, are you specifically referring to the December quarter or that's a more generic comment for the next 12 months?

No, I think it was a comment specific. I mean, we're trying to, what we're saying here is that as we expand, as we accelerate store addition, which is starting OND, we don't expect that to, Hit the margins in the short term, right? Longer term, of course, margins will continue expanding, but we don't expect that there should be a short term margin reduction on account of accelerated store addition.

Sorry, just to confirm, you're talking about when you say growth pickup, you're talking about store growth pickup or NOV growth pickup in that shareholder letter response?

It was NOV growth.

Right. And sorry, just to push on that. So you're talking about NOV growth pickup starting December quarter on a YOY basis.

That's right.

Understood. Thank you for clarifying. Just last question on dark.ai. One, how does this fit into what you already do in District, Zomato, Blinkit? Like, just trying to understand your competence in being able to build this business versus somebody else. And second, I know that it's still early from a monetization, or how you will think about go to market here. But in your own assessment, like just trying to think from a business model perspective, this is like an application layer on top of open source models. Is this something that you think can become very meaningful? Or this is just one of the things that you know, you're trying and you'll see how it evolves over a period of time?

Yeah, so that's definitely, I mean, nascent, we don't have a point of view on whether this could become big or not, right? I mean, if it works, it will become really big. I think it's more like a zero-one bet at this point. If it works, it's a global business. It will generate extremely high shareholder value over the next few years. But if it doesn't work, it's something we'll kill, right? So that's the space we are in. Right to win, I think, in these AI businesses, I think no one has the right to win. It's a nascent market, both consumer and enterprise AI. And most of the new large businesses coming out are actually startups. So, as I mentioned in response to the previous question, this is something we've been working on for about a year. And there are one or two more initiatives on the AI side, on the enterprise AI side, that we've spoken about in the past, including Nugget. Again, those are businesses or products we've been working on for the last many years. So, we expect next 12 months to sort of tell us whether any of this will become meaningful. But these are all businesses. Any of them scales is going to deliver a multi-billion dollar market cap outcome. So we definitely want to take short at these opportunities where either we have some unique capability, which gives us the right to win, which is true for, I think, enterprise AI products that we're working on, or, you know, we have some unique insight on the consumer side that we think no one else is building and we can take it to the market first. So dark.ai is a little bit of that.

Thank you so much for walking me through your thought process. All the best. Thank you. Next question is from the line of Gaurav Rateria from Morgan Stanley. Please go ahead.

Hi, congrats on good set of numbers. My first question is on the quality of growth that you talked about, that it should be led by the MTU growth. And it could actually grow, this number can grow to a much larger number. I think this could be potentially true led by your, you know, foreign to the smaller cities, which are really playing out well. I just want to understand how are you seeing the MTU growth in the top eight cities? What is the penetration level that we have kind of already hit there? Is there an odd star metric that we can look at one of the cities where you have hit a certain ratio at which other cities can also cash up to?

Even in larger cities, we are still seeing extremely good levels of growth. And that is driven by both new customer acquisition as well as category expansion, assortment expansion. And for example, in our mature cities, Blinkit MTU is already multiple times that of Somato. And we see that we don't see any plateau there yet. Right. Right. So hard to, we don't want to, you know, speculate on, on how large can top markets be, but suffice to say that both, whether you look at NOV growth or MT growth, the larger markets are still growing at a very rapid pace.

Got it. Second question on food delivery, the margins that you want to keep stable right now. I mean, you are already in the band that you had talked about, right? But it looks like the business still has a lot of operating leverage, which can deliver margin expansion. So this strategy of keeping the margin stable is more in the near term to counter the competition, if any, or is it more to, you know, further expand the time and kind of, you know, lead to the overall market or overall pie expansion in the country?

Yeah, Gaurav, it's more the latter. I think it might impact what you said first also, that it might impact or it might help us compete with the new end trends. But primarily, I mean, we started investing, as I mentioned, not just now, but for the last few quarters and that investment levels are going up. And that's also delivering growth, as we can see, right? So what we're trying to do here is try to expand the market, expand the growth for the business, which eventually leads to a higher growth on absolute adjusted EBITDA profits. That's what we're trying to solve for, not the percentage margin. So we feel that we can drive more growth in absolute adjusted EBITDA by investing more in growth rather than optimizing for margins. And we've been doing that now consistently for a few quarters. And we think we will continue on that path till we see the outcomes here.

Got it. Last question on doc.ai. How should one think about it like more like an application layer, more like your own reasoning model? And the moment, let's say you get more confidence for launching into the production, what kind What kind of investments would it require, not in terms of quantification, but what kind of, you know, investments you need to make, you need to make investments in hardware, software, what kind of investments, nature of the investment that it will require for you to kind of scale it once it goes into the production stage? Thank you.

There will be largely, you know, hardware and compute. So, those are the investments we will need here, which we are already doing since we've been running it on a pilot for the last few months internally within Zomato. So, yeah, I think that's what it is.

Thank you. Thank you. Next question is from the line of Swapnil Poodduke from JM Financial. Please go ahead.

Hi, thanks for the opportunity. Hello, can you hear me?

Yes, hi, go ahead.

Hi, thanks for the opportunity. So the first question is on your question number 10. You mentioned about some mature cities which are operating at more than 70% of the energy coming, being a bit positive. So can you just help us understand what do you exactly mean by mature cities, if possible, if you can name them? And are these cities predominantly in North India? I'm just trying to understand that.

Mature cities are more cities which have been operating for, you know, more than two, three years now. So they will include the metros and some of our older cities, which may not be metros, but top 15, top 20 cities, I think. Don't want to name any specific city here.

Okay, but is there any skew towards North India specifically, or like, it's spread across?

It is spread across.

Okay. And just a related question to that, what percentage of our NOE now comes from the top eight cities?

We are not sharing that data.

Has that percentage moved more significantly over the last, from the last time you mentioned somewhere around 70%? If you can just call And second question is, with respect to your contribution margin in Blinkit, I can see that your NOV per store has gone up, and this happened last quarter also, which means, but your contribution has been flat, right? So, which means the uplay that you should ideally get from dark store or a warehousing cost has not necessarily flowed down to your contribution. Is there any particular reason to call out over here? Why that does not happen?

Investment in new stores? Also, Swapnil, see, NOV per store can also go up because the store sizes have increased, right? But therefore, the cost will also increase. So therefore, you know, both are not correlated necessarily. And the other factor is also the gross profit, right? So while NOV per store can increase, it's not necessary that that translates into gross profit, Because the mix can be different. So there are various factors at play. So there won't be a linear correlation between NOV per store going up and contribution.

Okay. And then you talked about that there is not much difference between quick commerce and ecommerce in your mind. Does that realistically put your TAM at around 30-35 crores transacting users on ecommerce today? And assuming all the use cases that are currently served by e-commerce will eventually move to quick commerce. Is that how one should read that?

We are not aware of the 35% and no one has disclosed that. So if that is the number, then yes, you're right. But we don't know whether that number is true.

But directionally, you're basically saying that all the use cases that can be served on e-commerce will eventually move to quick commerce.

I mean, like, maybe not some categories, like fashion, etc. That is different. But I think what we're saying here is that the empty audition can still be meaningful from here. Right now, whether the 37 million becomes 350 million, as you are saying, or 100 or 200, like, we have no point of view on that, actually, honestly.

Just a last one on your AI agents. So here, can you make out what percentage of your traffic is actually coming from these AI agents today? And is there any defensive measures you can take to block this traffic if you don't intend to give them access? I'm just trying to understand what are the measures you can take if you think that you don't want to service that kind of a traffic.

So, right now, we are friendly to the traffic. We also keep investing in whatever is the latest protocols that we need to build on our end to make these transactions more user-friendly from an agentic point of view. I think right now the traffic, though, is negligible. So while we keep doing this to make sure that we are able to also see what is happening, collect data, and prepare ourselves in case this is something which gets traction, but so far there is no traction. So, so far, we don't even have a plan to figure out how to block them if we need to. Also, we don't really have a lot of use cases where there is nuisance traffic from agents because we are more transaction platform across the board. So, we don't have like blocking traffic, which is causing problems for us. So, we don't see that in the near term, unless we need to block them to block transactions, there is a use case for us to want to block agents.

Got it. Got it. Thanks a lot, guys, for taking my question.

Thank you. Next question is from the line of Rishi Junjanwala from IIFL. Please go ahead.

Yes, thanks for the opportunity. Couple of questions. One, you know, despite the fact that, you know, our overall store count in QC has grown by almost 50% year on year, the NOV per day as well as orders per day per dark store continue to go up, grow pretty healthily at 16-17%. And I'm assuming that these numbers would be higher for stores which are which have been open six or nine months ago or anything prior to that. Incrementally, you're talking about almost nearly doubling of dark stores over the next six to seven quarters. So just wanted to understand from a metric perspective, do you believe incremental dark stores will actually ensure that, you know, these NOV per day and orders per day actually remain in the range where we are? Or there is still significant scope for these to continue to improve?

Rishi, very hard to predict that. I don't think we're solving for that also. Eventually, the idea is to make sure that we are seeing good ROC on the investments we are making. And we keep sort of updating the investors on where we think, what's the framework for that, and what's our outlook on each of those metrics which influence that ROC. And so I think this is that what we know today. Yes, we think NOV per day should go up. We gave a guidance last time that we think in steady state, it could get to about 111 lakh rupees per day per store. We are at 9 lakh today. So we do think it will go up directionally over time. But, you know, there could be some quarters where the growth in number of stores is lumpy. So, I mean, therefore, it may not be linearly upwards, but directionally, yes, it should go up. But even if that view changes, I mean, what I'm trying to say is that you can change, but we could still deliver a high ROC in the business because margins go up or something else happens, right? So, very hard to commit to this going up continuously longer term.

Understood. And the second question is on dark.ai. So just, you know, wanted to understand from a long term strategy perspective, right? I mean, you've, you're already, you know, invested significantly in three, four businesses that you're doing. Is it fair to assume that over the next three to five years, we are leaning towards trying to build businesses which can be monetized on the tech side? And incremental capital may actually get utilized on that aspect, given that even if, suppose, this dark.ai succeeds, it will probably require a significant amount of investment as well, even if monetization and profitability may end up being high. So, are we looking at, you know, more or, you know, the next Lego value creation coming from some of the tech assets or capabilities going forward?

Yeah, I mean, that'll be a great outcome, I think, if that happens for everyone. And, I mean, that's our intention always, that continue to look for new opportunities, new businesses. And good thing is that we have the balance sheet to invest, if required, into building these businesses, As long as they demonstrate a longer term return on capital. So let's see, I think we are on that path. And this whole last two, three years, the developments on the AI front have given us an opportunity to take a shot at building some new businesses, which is what we're doing, right?

So, just to, you know, delve a little bit deeper. So, you know, if you look at there are opportunities available on the services side as well in the consumption basket. So, is it fair to say that our leaning is more towards looking at the technology side of the spectrum rather than the, you know, the services side where we are not present today?

Yeah, I think so. I mean, the opportunities are, of course, like many, I think we have to play to our strengths. And whichever business I think appeals to us, you know, and we feel we have a right to win and right to build is what we are choosing, right? I don't think we have the right to do everything. For example, we mentioned in the letter, private labels and quick commerce is not something that we think is something that we have competence in, right? So we're choosing to not do that. So yeah, opportunities are everywhere. We have to focus on subjects where we think we are a little bit of experts and where we think we have some core insight, which allows us to build a successful business.

Got it. Thank you so much. All the best.

Thank you. Ladies and gentlemen, in the interest of time, we will now take the last one to two questions. The next question is from the line of Vijay Jain from Citigroup. Please go ahead.

Yeah, thank you for the opportunity. My first question, on the main Zomato app, you have the chat AI interface and you have text box there as well as an interactive bit there, right? How do users engage? Do you see any difference there between the two? And does that inform your view on how some of these general horizontal, you know, AI chatbots and AI agents would evolve over time? Do people show preference for one over the other is my first question.

Yeah, so I think the current AI bot that you see is not, we're not seeing, I mean, we're seeing some traction there, but I think we are iterating on it, so you will see a much better product in the coming weeks and months. Yes. And yes, the attempt here is to see what parts of agentic interfaces actually work for our customers. And some of the insights we have shared in response to question two, that we think people still want to browse, right? So while they can, on the current Zomato app, you can go and do a conversation to order your food, but we are not seeing too much of an offtake on that, and people still want to browse and order. So attempt to learn what the consumer behavior is, and we will continue to adapt and iterate on how we can make our interface better using the capabilities that AI offers us today.

Got it. Thanks. My second question is, you know, in OneQ, you talked about some gourmet store expansions in tier one cities. We just wanted to get a sense on how that has progressed. And in general, you know, your AOVs at an overall level here, net NAOVs are stable. And you've said that obviously new cities are at maybe 10 to 15% lower. So is it the gourmet stores in general that is, you know, offsetting the impact of newer cities because clearly newer cities are growing faster, right? So just wanted to get a sense A on the gourmet stores as a collective and B if that interpretation is correct.

We've expanded gourmet to Bangalore and Bombay as well, but we are not disclosing the separate total overall number of stores for gourmet. But we have opened in those locations. They do not form a significant percentage of the orders to basically impact the AOV like you're talking about. I think most of the offset for the AOV is assortment expansion in Tier 1 cities, which leads to higher AOVs in Tier 1, and that balances out some of the lower AOVs from Tier 2 cities.

Got it. And just related question to that. So, you know, that Pareto principle you've talked about, you know, 80% SKUs, sorry, the high volume SKUs is what you want to focus on within quick commerce, the ones that would move volumes. Is there any calibration to that assessment as, you know, you wade deeper into the long tail e-commerce?

I think we don't have a Pareto in mind. I'm not sure if we've given that guidance at any point of time. But our thought process is that as we scale up and as our supply chain gets bigger, we are able to add more SKUs at a lower incremental cost. So that is why we are able to add more assortment in tier one cities faster because we have bigger scale there. And as a tier two, tier three cities also get bigger, we are able to add more and more assortment at a much lower incremental cost.

Got it. So you would be more comfortable in general speaking going forward if you are stocking some products for longer than let's say your average stocking period has been in the past in the dark stores. I don't know if I asked that question right. I hope you understood what I was trying to get at.

Look, we already sell air conditioners, and you can't keep the one week of inventory for air conditioners. So, obviously, the holding period is much longer for those products. We are comfortable doing that.

Understood. Perfect. Thank you so much. Those are my questions.

Thank you. Next question is from the line of Abhishek Banerjee from ICICI Securities. Please go ahead.

Hey, hi. Cover up questions from my side. First, on the growth part, say, in input commerce, would you be able to give some color on the growth in Delhi NCR versus, say, tier one cities versus tier two cities on a sequential basis?

I think we're not providing that breakup, but in general, we haven't seen any awkward trends over there. A lot of our growth, you also have to understand, is also related to how our supply comes online. So that tends to also create differentials between what is happening in different cities.

Got it. So, in terms of, you know, the store expansion, you have mentioned that there is going to be a lot of store expansion in the Tier 2 and Tier 3 cities. Will the KPIS requirement for adding new stores in Tier 2 and Tier 3 be a lot lower than that 2.5 loan number that you had shared earlier, or is that a blended number?

That's a blended number, Abhishek, when the guidance was given, keeping in mind that a large part of our expansion will happen in tier 2 cities.

I understood. Now, one comment that you kind of made was that, you know, the difference, the line between e-commerce and e-commerce seems to be blurring, right? So, on the ad revenue as a proportion of NOV, from what I understand, there is still a T-shirt 250 to 300 pips differential between what, say, you earn vis-a-vis, say, some of the leaders in the space earns. So, do you think that is also going to converge over time?

Abhishek, I'm not aware of that, the number, the differential that you're talking about. So, it's hard to comment where we are versus anybody else's numbers. We are actually not aware of those numbers.

I don't know, but okay, sure. So, is it?

Your question is general around monetization of ad revenue. See, the difference between our platform today is we are able to tap into brand revenue, right? Whereas your horizontal e-commerce platforms can also tap into ad revenue from sellers themselves, which are competing on the platform, right? But as the overall assortment converges, both these numbers do tend to converge.

Understood, understood. In terms of the food, now one question on food delivery also, so we are seeing now five quarters of acceleration in growth, right? And you yourself mentioned that you have been a little more aggressive in terms of pricing in some job fees. So, is a 20, 20 to 23% kind of a worldwide growth a fair number to kind of keep in mind for second half or should we kind of be a little more vulnerable to that?

Yeah, I wish we won't be able to comment on that question as you would appreciate.

Sure, sir. Just one last question on the dark AI one. So, I am honestly not very clear on what it is, right? So is it, see, a LAMA sort of a model, right, which you are building grounds up? Or is it an application layer, which kind of utilizes a model like LAMA? If you could clarify a little on that, that would be helpful.

It's the latter. So we are using open source models and fine tuning it and building an application on top of it.

But you also mentioned that there may be need for, you know, your own compute and GPU.

So that is a background. for hosting the models.

Okay. Fair enough. Thank you.

Thank you. Ladies and gentlemen, we will now conclude this conference call. Thank you for joining us and you may not disconnect the lines.