Subex Limited (532348) Earnings Call Transcript
November 4, 2026
Earnings Call Speaker Segments
Good morning, ladies and gentlemen. This is Shashmita, and I'll be moderating today's conference call. Welcome to the Subex Limited Q2 FY '27 Earnings Conference Call. <<operator_script>> Kindly be aware that this conference is being recorded. At this point, I'd like to hand the floor over to Mr. Pratik. Thank you, and over to you, sir.
Thank you, Shashmita. Good morning, and welcome to the call for quarter ending September 30, 2026. The members of management who are present for the call are Ms. Nisha Dutt, Managing Director and CEO; Mr. Sumit Kumar, CFO; Mr. Harsha Angeri, Head, Corporate Strategy and AI; and Mr. Ramu Akkili, Company Secretary and Compliance Officer. I would like to start the conference call by going through the safe harbor clause. Such statements in the presentation concerning our future growth prospects are forward-looking statements which involves several risks and uncertainties that could cause actual results to differ materially from those in such forward-looking statements. These risks and uncertainties relating to the statements include, but are not limited to, fluctuations in earnings, our ability to successfully integrate acquisitions, competition in our areas of business, client concentration, liability for damages in our contracts, withdrawal of tax incentives, political instability, unauthorized use of intellectual property and general economic conditions affecting our industry. With this, I now hand over the call to Nisha to take it forward. Over to you Nisha.
Thank you, Pratik. Good morning, everyone, and thanks for joining our earnings call. First off, we want to apologize to all of you. I think we ran into some technical problems this morning with the inbound numbers that had been circulated. And a fair number of you ended up waiting a long time just to get into the queue. So please accept our apologies ahead of time, and thank you for your patience. With that, let's begin. We posted yet another quarter of growth in Q2 FY '27, with revenue up 7.7% quarter-on-quarter and 24% Y-o-Y. That makes 5 straight quarters of Q-on-Q growth, which I believe is among the longest streaks since the company was listed. Our EBITDA margin came in at 20.8%, while PAT stands at 16.7%. EBITDA and PBT both rose quarter-on-quarter even with the annual wage increase falling in Q2. So these results show continued progress on both growth and profitability. Even more importantly, they show that the operating discipline we have been building over the last 3 years is starting to turn into tangible financial performance. For a company that has spent the last few years rebuilding its foundation, this matters. We have talked consistently about what that rebuilding demanded: greater predictability in execution, sharper choices about where we compete, a more sustainable cost structure and an unwavering focus on customers.
Our priorities remain the same. What is shifting, though, is how well we turn that discipline into growth. In my view, this quarter offers additional proof of that progress. I am equally aware that one solid quarter, or even several in a row, does not amount to a turnaround. Whether we can make growth repeatable, profitable and sustainable is the true test, and that is still where our focus lies. Today, instead of going back over the quarterly developments, since we have already covered those in our investor presentation, I would really like to take this chance to discuss something just as important, namely where Subex goes from here. Because restoring the business — and some of you have been with us on this journey, as you know, and you know we have been speaking about a turnaround — restoring the business was never meant to be the end point. What we are attempting is to build the foundation from which the next phase of Subex can be created. So where is our future headed? I want to spend a bit of time today talking about where we go from here. Let me start with a simple observation.
When you look at telecom's past, it's largely a story about flows. Back in the voice period, value was on the move, and being able to measure those flows, build them, reconcile them, and safeguard them was essential to how the industry's economics worked. After that arrived the data period: the character and size of traffic shifted enormously, yet the core idea stayed the same. Any place value travels, someone needed to comprehend it, keep accounts of it, and guard its integrity. For over 3 years -- 3 decades, Subex has operated right at this exact crossing point. Through networks spanning over 100 countries, we've gained insight into the way economic value travels across complicated systems, plus the spots where it leaks, the spots where it gets misallocated, and the spots where it's exposed to exploitation. Today we think a third form of flow is coming into being: voice flows, data flows, and, increasingly, intelligent flow AI, which alters the very character of what travels across networks and platforms. We're heading toward a world where software doesn't merely process information or suggest actions. More and more, AI systems and agents will take decisions, start transactions, engage with other systems, and carry out activities with differing levels of autonomy. And that shifts something quite fundamental. Once intelligence starts to act, the decisions themselves are able to turn into economic events.
A wrong answer from a system today might be inconvenient, but when an agent takes a wrong action, for instance, that can carry a direct financial consequence. It could result in an incorrect settlement, authorize a fraudulent transaction, misallocate revenue, or set off a chain of actions that may prove very hard to reverse. In other words, as intelligence grows more autonomous, the question is no longer simply whether a system can detect the anomaly once it has occurred. It is whether we can establish that the decision itself was appropriate, authorized, economically sound, and executed within the right boundary. And we believe this creates a new and increasingly more important layer of value in the technology ecosystem. And that is what I would call the integrity of decision. Today, enormous investments are going into making AI systems more capable, not just by us, but even by telcos themselves. So we believe the next challenge will be making those systems trustworthy enough to act, particularly when their decisions carry significant economic consequences. And this is where we truly see the opportunity for Subex. And there are 3 reasons why we believe we have a credible foundation to pursue this opportunity. First, our domain expertise is deeply rooted in the economics of telcos.
For many years, we've collaborated with operators to pinpoint the places where value gets created, the places where it gets lost, and the ways it can be safeguarded. That experience has taught us something worth more than knowing transactions on their own. It's taught us how relationships, dependencies, behaviors, and exceptions work — the very things that decide if a transaction is economically legitimate. As AI systems start taking part in these transactions, that kind of domain intelligence matters more and more. The technology might be shifting, yet the core problems around trust, accountability, and economic integrity aren't going away. In fact, I'd say they're turning into much bigger stakes right now. Second, we see a chance to establish Subex as a platform-neutral layer for integrity and decisions. Telcos are putting significant money into their own AI platforms, data infrastructures, and automation capabilities. We don't think our future rests on swapping out those investments or owning each piece of the AI stack. What we can do is create capabilities that function across such environments, assist customers in governing critical decisions, safeguard economic outcomes, and set up accountability no matter what technology sits underneath. So consider how the makeup of network activity is itself shifting. In the past, networks were built mainly for activity begun by humans. A person placed a call, you send someone a message, you use data, and you kicked off a payback.
What we're seeing more and more is that machine bots and autonomous agents are going to start and carry out actions for both people and enterprises. That introduces a completely new issue around scale and accountability. Take, for example, answering the questions about who gave the authorization for the action, and whether it fell inside the intended boundaries — we know every agent has been given a boundary, but who is going to verify whether it stayed within those intended boundaries? Was the economic result legitimate? Who bears responsibility when something fails? And in that light, what we're witnessing is essentially the reinvention of the old maker-checker principle. Only this time, the maker could be an autonomous agent, and the checker itself could be intelligent, and the decision could be intelligent. So this — and those decisions may actually happen at a scale and a speed that traditional controls were never built for, since traditional controls were built to handle more human traffic, correct? This is traffic at a very different scale. So importantly, this is an opportunity that doesn't require us to bet on which AI platform ends up winning. That isn't the battle we're in. Our ambition is to make integrity relevant across platforms rather than being dependent on any single one.
Third, we see potential in this opportunity to generate a compounding edge. Our legacy operations have provided us with substantial exposure to intricate network and transaction settings. As we bring AI-driven capabilities into that setting, there's a chance for us to convert that experience into improved models, more robust decision frameworks, and more powerful controls. Each fresh deployment can sharpen our grasp of further patterns, risks, and relationships—subject, naturally, to customer permissions and data governance. Still, in my view, those learnings are capable of strengthening the intelligence we apply to later deployments. Thus, over time, we think this may produce a self-reinforcing loop linking domain expertise, product intelligence, and customer outcomes. And that is the edge we truly aim to build—one anchored not in technology, which is easy to copy, but in the accumulated intelligence of how complex economic systems really operate. So let me stress that this opportunity, in our view, is not limited to traditional mobile network operators. The telco sector is itself evolving. I'm certain most of you have heard by now about satellite and nonterrestrial networks, NTN, which are starting to challenge long-held assumptions regarding connectivity, network boundaries, and the economics of service delivery.
New entrants and technologies are going to produce networks that are more heterogeneous, fresh commercial relationships, and possibly additional kinds of revenue leakage, fraud, and settlement complexity. Meanwhile, mobile money, digital financial services, and enterprise platforms are growing more interconnected and increasingly automated. So in this environment, we see one shared challenge coming into view. As participants, transactions, and autonomous decisions multiply, establishing economic integrity matters more, not less. And we think the underlying capabilities we've built in telco can serve as a foundation for tackling some of these adjacent opportunities. We also want to be just as candid regarding what this means for Subex today. What I'm describing is a strategic direction. It isn't a claim that we've already established ourselves in this market. We plan to pursue it with the same discipline that guiding. Our first responsibility is still to strengthen and scale our core business. Business assurance, fraud management, and our managed service capabilities stay central to our near-term growth and profitability. Alongside that, we're advancing AI-led capabilities that tackle real customer problems, among them more intelligent detection investigation and increasingly autonomous workflows with suitable human oversight. So we plan to place these capabilities in customer environments, show measurable economic value, and build out from there.
Pursuing AI merely because it is AI holds no appeal for us, and neither does moving into neighboring markets just because they look appealing. What we will keep asking ourselves is whether our domain intelligence provides a real edge, and whether we can turn that edge into customer value we can measure and a sustainable business for Subex. The pace and direction of our investment will be set by that. So looking at the road ahead, I see 2 responsibilities that have to advance side by side. One is to keep building a business that is stronger, more predictable and more profitable. The other is making sure the business we are strengthening stays relevant to the industry problems of the next generation. Doing one at the cost of the other is not something we can afford. Restoring discipline, credibility and the ability to execute has been what the last 3 years were about. The phase that follows, though, has to take that foundation and use it to build a business with greater strategic relevance and the larger opportunity we see. Over time, our ambition is to get closer to the decisions that determine financial outcomes instead of staying solely within the systems that analyze those outcomes after the fact. So for where Subex began, I believe this is a meaningful evolution.
This also follows naturally from the 3 decades we've invested in learning how to do this. So our aim isn't to create an improved iteration of the Subex of yesterday. If there's a single thing I'd like you to take away, it's this: building a better version of the Subex of yesterday is not my objective. What we're doing is positioning Subex for where value is heading next, and that's toward the intersection of intelligence decisions and economic integrity. And we plan to earn that position through execution, customer trust and demonstrable results. So before I get into the quarterly financial performance, I want to highlight the appreciation and the value of our minority investments in FiberSapien and OnGrid. According to the latest valuations available to us, these investments have appreciated approximately 5x and 2x, respectively, relative to their original investment value. So these investments represent another dimension of the value we have sought to create through our strategic choices. And this appreciation further strengthens our consolidated network. With that, let me quickly walk through the financial performance for the quarter. All numbers are in INR. Revenue for the quarter stood at INR 85.9 crores against INR 79.4 crores for the previous quarter.
That reflects growth of 7.7%. For the quarter, normalized EBITDA came in at INR 17.77 crores, compared with INR 16.87 crores in the prior quarter. Normalized PAT for this quarter is INR 14.3 crores versus INR 16.9 crores in the previous quarter. PAT stands at INR 14.9 crores, against INR 14.22 crores from the quarter before. As of quarter end, cash and cash equivalents were INR 198 crores, compared with INR 184 crores at the close of the previous quarter. Thank you for your time, and I'll now be glad to take your questions and comments.
<<operator_script>> Our first question is from Mr. Parth Dalal, who is an individual investor.
My question is this: if AI-led operators handle fraud themselves and provide their own assurance, then what takes the place of your managed services revenue over the next two years?
Apologies, Parth — your line isn't coming through very clearly. Could you ask your question again?
Is this coming through clearly at your end now?
Yes.
Right, so my question is this — if AI enables operators to handle fraud and assurance on their own, then what takes the place of your managed services revenue over the next 2 years?
Actually, that's a good question. I don't believe it automatically takes the place of your managed service, correct? Because with managed services—there are 2 points to grasp here. To start with, plenty of OpCos rely on our revenue assurance and fraud management system, but they aren't the sole—you know, ours isn't the only system being used within a telco setting. A telco setting usually involves several systems in play, including, in many cases, their very own. So it isn't automatically the case that, once we're inside a telco, ours is the sole system in use. That reality holds true even now. And our managed services revolve around our product, correct? So we don't provide managed services for some competitor products or other products. Therefore our managed service revenue is largely tied to maintaining our products, upgrades, upsells, and so on. But I think there's fundamentally one thing worth understanding about why customers keep using us, and that's because a vendor such as ours operates across multiple telcos, correct? So we operate across hundreds of telcos, and what that gives you is the benefit of seeing cross-telco patterns and, you know, grasping how—say, a fraud is emerging from one geography and traveling to another.
Fraud is quite artificial, isn't it? That's exactly where managed services matter. And I'd say that's the reason people actually come to us. One of the points I raised in my opening as well — who ultimately signs off on economic integrity? Right now, in the system, that's us. So will managed services begin to be replaced simply because clients have their own systems? I don't believe that would happen, since that's already true today, in fact. That doesn't mean my managed services is shrinking. But what we're certainly doing on our side is making our own managed services much more efficient — which, in fact, we want our managed — so I don't think managed services as a concept disappears, but how managed services is delivered can certainly become far more efficient. And we ourselves are also exploring managed services, AI agents, and all of that. Agents already handle a lot of managed services for me today. So we're actually looking in that direction too. But just to clarify, the existence of one doesn't necessarily mean the other isn't needed, right?
Appreciate that, sir. Our following question is from Ajay Desai, who is an individual investor.
Great quarter on the numbers. One issue, though — we had quite a bit of trouble getting onto the call, so perhaps that's something to look at next time. And then, on this matter of building visibility with investors and, you know, having those investors meet each month, maybe. So if we're able to keep that going over the next few months, you know, perhaps by year-end, once we have a full year of numbers, our valuation will probably begin to reflect some better things too. So if you could share a plan — your plan — regarding investor meetings, what does that process look like?
The intent here—and certainly management's intent—is for us to keep up the engagement. Of course, during a silent period we tend to drop off the radar somewhat. NDRs aren't something we can do then. But now that our results are out and we're holding the earnings call today, we'll pick up the pace again on getting out and meeting investors. So continuing this is definitely our plan, and we're actually working with EY on a plan. So perhaps we'll publish a plan for investors, so you know which cities we're coming to, the dates and all of that. We'll work with them and probably publish a plan, but the intention is definitely to keep the momentum—if not to increase it. In fact, that's what we're going to do right now.
Our following question is from Mr. Haider Kachwala, who is an individual investor.
Ma'am, so... you gave qualitative guidance in the opening remarks, but could you call out something more quantitative in terms of numbers? That way at least we have something to look forward to. If you could just give some remarks on that as well.
I think this gets asked all the time. As for giving a lot of forward-looking guidance, we can't do that. But I'll say that, as I mentioned in my opening remarks, growth remains a priority for us. And when I speak of sustainable growth, to me it isn't just going Q-on-Q — it's also making sure that growth doesn't come at the expense of your bottom line. So we want to make sure our margins are durable, and that we're able to sustain margins within a certain range. It may not stay exactly the same every quarter — obviously that's understood — but we want to maintain some level of margin durability while we grow our business. So growth is a key focus because I think we have a good handle, and we have done the bottom line for — I think we've done many quarters now, maybe 9 quarters or 10 quarters. We've been demonstrating a very robust bottom line, or I'd say a fairly healthy bottom line. So right now the whole focus is on seeing how we can grow sustainably, and that remains the focus.
As for where we're headed, our aspiration is clearly, I'd say, to hit that high single-digit to double-digit range. That's the direction we're moving in. But in terms of trajectory, that's the goal we're aiming for, and the entire company is basically aligned around making sure we can seize that opportunity in the market. I'd say the same applies to EBITDA and PAT — they should keep moving along the same lines.
Ma'am, even though you say growth should be delivered on a sustainable basis — your growth, I'm seeing that you've grown by roughly 20%-25% this quarter 1 on a Y-o-Y basis. But when you look at constant currency terms, you're, you're just being — you're getting the benefit of INR depreciation. You're not actually—
Sir, how would you arrive at a 24% depreciation in the INR? I mean, that's roughly even, you know, across the last 12 months. So let me say this — I've been quite candid over the last 12 months. Okay, I don't have those figures at hand, but here's the broad guideline, right? Our— yes, do we gain from INR depreciation? We do, as do all other companies. But the benefit we get from the INR is roughly, you know, I'd say 2 or 3 percentage. It isn't— yes, and year over year it might be around, you know, 10% or perhaps...
Quarter-on-quarter?
On a quarter-over-quarter basis, it's actually been flat. So yes.
Ma'am, I agree — I'm not even talking about it on a quarter-on-quarter basis, but on a Y-o-Y basis, even if I came to...
On a Y-o-Y basis, that's probably — no, it'll come in at roughly 8% to 10%. Yes, you've got that right. Those figures, I think, are ones anybody can grasp.
Instead of posting 10%-12% growth year over year and then citing a higher, sustainable number, I just don’t think that lines up with what you’re telling us.
No, Haider, honestly every telco vendor out there today— alright, so I think we ought to frame this with a bit of context. To start with, the telco industry on its own expands at roughly, you know, 2%-3%. So, you know, telco growth sits at just 2%-3%. And we rank among the fastest-growing telco vendors, and I'll push back on that with you, and you're welcome to check the data. I do thorough data analysis. Right now we're one of the top telco growth vendors. Most vendors in our space, comparably, have in fact shrunk. We're currently expanding at 3x to 5x the industry's growth rate. So to me, that's genuinely quite good, and I know the baseline I'm starting from. We actually were, you know, flat and shrinking. So I think from that angle, expecting this to reach, you know, on a constant currency basis, expecting it to reach 25%-30% growth, that's, you know, that, that will take time. That doesn't happen within a year. I think within a year, giving you, delivering, you know, 14%-15% growth is, in my view, no small feat actually, considering what the state of the industry is and what the rest of the industry, you know, telco vendors are doing right now.
Alright, got it, ma'am. And then finally, regarding the cash on the books — it's certainly quite encouraging to see you holding roughly INR 200 crores in cash on your books, which comes to about 20%-25% of the market cap. So if you could share some guidance on how that cash would be utilized — whether you would be putting it into R&D, acquisitions, or even considering a buyback to create greater value for shareholders.
Regarding how we use our cash, we'll be examining opportunities for inorganic growth. That is, you know, the primary way we're looking to deploy cash. So we're actively, you know, working through this. We're actively, I'd say, running a pipeline at present to check whether solid opportunities exist for us. But mainly, deploying cash will go toward that. R&D, naturally, you know, R&D we do spend — you know, R&D expense rises, we do in fact raise the — or I'd say, you know, we raise R&D intensity each year, but that doesn't flow through the balance sheet. You know, our R&D isn't capitalized; it sits in the P&L. You know, so every R&D expense of ours runs through the P&L. So there, I don't believe that, you know, cash by itself directly comes into play. So much of how we deploy cash will be on inorganic growth. As for buyback, we can't carry out a buyback at this time because I think everyone listening on the call knows that our retained earnings are negative. So we cannot execute a buyback at present, in fact. So we'll try to see whether, you know, at some point that turns into a viable option, but right now that isn't an option for us.
Our following question is from Sanjyot Khare, who is an individual investor.
Thanks, and congratulations on a solid set of numbers. It's really encouraging to see revenue and margins improving quarter-over-quarter. And even last quarter, it was good to see that margins are still being maintained even after, I think, you gave an annual wage hike to the company employees. So that's really good to see, knowing that we are growing about 7%-8% for the last 2 quarters, and the second half is generally better for the company. So can we expect this growth to accelerate in the second half, H2? And hoping to see us crossing INR 100 crore revenue per quarter very soon. On that, are you seeing any challenges to the order one that's already there for execution, and to accelerating the growth?
Honestly, Sanjyot, there isn't anything significant at the moment. What I mean is that, in the sense that I believe we're currently on a solid growth trajectory. So the difficulties aren't really different from what I think everybody is already aware of, namely the Middle East. The Middle East, as you know, appears to be fluctuating between hot and cold, depending on which particular week you're referring to. At times it genuinely escalates, and at other times, well, it's simply subdued. So we don't know how the Middle East is behaving at this moment. So there is a certain impact, but that is somewhat already factored in, you know, to where we stand today. So I'd say the Middle East is one impact. And naturally, the US and elections are always a, I don't know, wild card. So how that unfolds remains to be seen. But yes, apart from the geopolitical issues that I think are visible to everyone, a portion of it has already been incorporated into our planning. I'm not anticipating anything, you know, anything different over the next 2 quarters unless a black swan event occurs or, you know, something major takes place.
And that's something we haven't anticipated at this point. Beyond that, I'd say, at this moment we're fine.
That’s great. For my second question, last quarter Subex announced that it would purchase shares in the market for what I believe is the employee fund — I think it’s the employee fund that sponsors.
Yes. ESOP.
So is that something you're looking to put into place during Q3?
Yeah, we’re definitely going to be examining that, and, you know, it might not happen all at once—we’ll spread it out a bit—but we will do it. Usually, when the trust purchases shares, I believe the sole limit we face is that the buyback can only occur when there’s no UPSI in hand. So that’s the one restriction we have, but yes, we will certainly be carrying that out, for sure.
Patrick Matthews from Retail Investor is up next with a question.
Well done on a solid set of results. You hit the guidance you gave last quarter, or at least the numbers you sounded confident about. I liked your opening comments — specifically on what opportunity AI offers, how it affects the telecom ecosystem, the new flows, the economic impact, and so on. Good vision and a good approach. I'd like to dig a bit deeper on whether there are particular areas we're going after. Will we take the point product route? Or build an AI platform that handles this? Could you give a little more specificity on the details? And how large is the market we'd be addressing by using AI as the lead technology to transform our business? As you said, we're no longer looking at incremental steps — we're looking at transforming Subex into, perhaps, Subex 3.0. So I'd like a bit more detail on that.
All right, absolutely. So to start with, regarding the TAM — I believe there's an investor presentation available publicly. We've identified what we'd call our defined TAM at $4.3 billion, right. So that essentially stands. That being said, what we're really after — again, our approach isn't to go with a platform strategy, because from what we observe, that platform strategy doesn't scale particularly well. Beyond that, many telcos have put money into building their own AI stacks, correct? So if you come in with your own platform, the discussion turns into whose platform is superior, whose models are superior. That's genuinely not where we wish to compete any longer, since it provides us no meaningful edge. What we're therefore attempting is to determine whether we can package up our domain expertise and the understanding we have, and build modular, independent stacks, right. And we actually have a roadmap — a technical roadmap — that we published at some stage as well. But fundamentally, what we're attempting is to progress in the direction of an architecture of autonomy.
We refer to it as AOA. What we're aiming for is ensuring that all our activities can be encoded so our agents are capable of performing them. And this happens within separate stacks, correct, where agents possess autonomy, can collaborate with one another, and can utilize protocols such as A2A, MCP, and similar ones. So that's really the direction we're taking. Thus, in response to your question, it's not really about a platform, but rather, I'd describe it as modular standalone point solutions, since fraud evolves so rapidly, correct? Or even assurance problems. So my view is that if you adopted a platform strategy, our platform might ultimately fail to address everything that could arise. That's extremely difficult to do, and I believe it's an investment not worth pursuing given how the world is changing nowadays. It's an investment that no longer makes sense for us. I think we're in a better position pursuing separate, interoperable stacks, you know, alongside the underlying telco stacks they might already be using. For example, they might be tied to a Snowflake or Databricks or DataRobot, whatever, you know, stacks or data lakes that a telco has committed to.
We ought to be able to coordinate our point solutions across both stacks, in a sense. So that's really the way we're going about this.
I completely agree on the product point — the way the products approach things, and considering our size plus our overall, you know, exposure and so on, that could well be the correct path. One suggestion, or rather a request: if this could be laid out in, you know, a public release — what our approach is and what we intend to do — that might give investors a clearer picture of where Subex is headed as we move into the future. Right now, the prevailing, prevailing view among most investors is that we have been permanently stuck in, you know, a sort of pattern of climbing out of many of our issues with no next step. I mean, perhaps we did incremental things previously. We might have adopted a, you know, ship's left approach 5 years ago, but that failed to pan out as anticipated. Now that our objective is clear when it comes to concentrating on our core strengths, and AI is presenting a new opportunity — that opportunity, our approach — AOA, as you mentioned, nice terminology. I think expand on — if you can expand on that and, you know, clearly articulate across a couple of slides what we are working on when it comes to transforming our company, that will — that might be well received by the investor community.
Fair enough, Patrick — I’ll take care of that. The truth is, we already have this laid out clearly, and we include it in our proposals and everything else whenever we’re working with telcos, but I suppose we just haven’t really presented it to investors externally. So we’ll keep at it. In fact, I’ll make a point of it — I’ll treat this as an action item, and the next time we’re out on the road, we’ll have it ready. We’ll make sure it’s all clearly laid out and properly fleshed out. We’ll get that done.
After the operator's instructions, we move to the next question, which comes from Abhishek Kale, who is an individual investor.
We’ve been bringing in new logos and expanding into new geographies for nearly 18 months now, right? If you can, could you tell us how much they’ve added to our results? I mean, what percentage? A rough estimate, if that’s possible and feasible—could you provide that? Those new logos we’ve brought on over this past 1.5 years.
So, in terms of — okay, so I think — it's tough to go deal by deal, but I will give guidance on, you know, broadly or in aggregate, how this plays out for us?
Yes, that's correct.
Right, so essentially the way it works—and the way to think about it is this: first, in any given year, whatever revenue figures we achieve, correct? We go into the year—ours is a business driven heavily by backlog, right? So we more or less sell during this year and then we consume that over the following few years. So it's a backlog-type business, you know. So in any given year, whatever revenue figures you're looking at, 70% of that is recurring or, you know, backlog. That means, you know, I start my accounts with a recurring base and then the remainder comes from new orders we secure during the year. Now whatever orders we secure in the year, the way it shows up is whenever you win an order, typically in year one of implementation, because that's the year when we implement the product, across the first 4 quarters you'll see that we take roughly whatever deal value we have announced. Typically, you know, again, there are some variations, but I'd say that, you know, broadly speaking, roughly 40% to 50% of the revenue will be recognized from that money, whatever, you know, the deal value is, within year one.
So, it can be 20% to 30% at times, and at other times 40% to 50%, depending on how the contract happens to be structured. But generally speaking, it falls within that range. And the remainder, once implementation wraps up — for us there's a milestone we call go-live. After our product goes live in a customer's production environment, from that point it turns into essentially a straight line. Beyond that, payments come in a straight line, which is typically an MS or an AMC that they pay us. So the rest arrives in later financial years. So if I put it that way, for example, to keep it very simple, if a deal is worth $1 million, then you can expect that over the next 4 quarters I would have booked perhaps $400K to $500K of that across the next 4 quarters. And after that, whatever remains of that $500K — if the deal ran 3 years, then it's essentially a straight line from there. Over 36 months, or if it's 5 years, then it gets spread across that time period. So that's typically how it works out for us.
Usually there’s a license value taken upfront, and that license gets recognized over the implementation period. Then comes the project implementation—that is, we’re implementing the product, which we refer to as professional services. That value typically depends on—you know—40% to 50% or whatever of the deal value coming in as we implement the product over the next 1 year. And once the product is live, it moves into maintenance mode and services mode and all that. And in services mode, we typically do a straight-line accrual. So that’s our model.
Okay, Nisha, a quick follow-up on that. I believe we've brought this up over the past 2 calls too — the time it takes, for want of a better way to put it, to make money from the point a deal is signed until the product is actually implemented. Last time, I think you mentioned a 45-day window, and you wanted to see that move closer to a quarter's worth of time savings.
You've got that right.
Has there been progress? In other words, did you manage to shrink that window even more over the course of this quarter? And then, likely, the AI — specifically the AI SDLC. That, I believe, is what our deck also pointed to, in connection with training for employees. So is any of that visible — is it having an effect when it comes to making that window shorter?
Okay, so the AI SDLC piece is mainly aimed at building things for the future. That's really for our core engineering side. It's not so much about implementation. In core engineering, what we've managed to accomplish is that we're leveraging AI SDLC, and we're kind of pushing it much harder into the market with our engineers. What we've achieved is delivering many more features within the same timeframe. Take this example: suppose in a quarter we were meant to do something like 10 features in a sprint. What we've observed is that we've been able to complete something like 14-15 features. So our capacity to do more has genuinely gotten significantly better thanks to AI SDLC. So I think that's one point. The other one is, if you've caught the announcements regarding FraudZap agents, even that agent deal we closed — all of it was actually delivered through AI SDLC. So in that respect, I think it's paying off for us. I'd still say that it isn't yet where we need it to be.
I mean, have we hit our benchmark numbers? No, I think there's still a lot more work to be done, but that's something we keep pushing forward with engineering. The second thing about the delivery timeline crashing comes more from, well, the library actually. So what we're doing is internally we're building a library of controls. Basically what that means is, you know, can you create reusable—reusability in your deployments? So prethink preconfigured controls, right? So can you have a library that then helps you kind of do that faster? So where we stand is that we've made significant progress on our HyperSense deals in the sense that, you know, the HyperSense library of controls is something I'd say is a lot more advanced. We're starting to work on our ROC, you know, library of controls, which is, I'd say, a little bit more early stages in that sense. So we're doing it product stack by product stack. So for any HyperSense, we're seeing a lot more time crash, which is almost—actually, I'd say that the last 2 or 3 implementations that were—that happened on HyperSense happened way before timelines that we'd given to the customer.
And once we actually measured it, there was nearly a 15% gain that we achieved in that area. So we're noticing that times are beginning to fall sharply for us, 15% to 20%. We're seeing right away that with HyperSense, we've been able to, you know, shorten the deployment timeline. ROC is one where we need to begin putting a similar effort into that stack. We're starting on that, but frankly it isn't as far along.
Alright, could I just ask one additional question, please?
Yes, absolutely.
Alright, are there any POCs from our side—from any of our other lines of business—where we're trying to determine whether our existing expertise can be put to use? Is there any of that work going on? Because I believe, on the last concall, or the concall before that, you shared that we're actively engaged with some companies and running some POCs. So, if possible, could you share more details on that?
On the POC side, here's what we've managed to accomplish — as I mentioned earlier to Patrick, we've been running proofs of concept around autonomy architectures, the AOAs. We've done agents, we've done MS agents, and also newer areas such as social engineering and the like. So we're running POCs with customers. A few of them have already been closed out — technically they concluded successfully — and now we're working through the commercial side: how do you take a POC and turn it into commercial value with the customer, and obviously looking at how we can negotiate some of that. But net-net, what happened over the last quarter is that 3 POCs closed, and I'd call them technically successful in the sense that — as you know, Abhishek, whenever a POC is run, there's an exit criteria, right, for that POC. What is it that has to be demonstrated? So we were able to complete 3 technical POCs and show that all the exit criteria were met successfully.
At this point, the sales team is, you know, turning it into a commercial arrangement and beginning to, you know, negotiate with customers and see whether we can convert those POCs into commercial value for us. So yes, actually 3 more have occurred in the last quarter.
Our following question is from Mahesh Kumar, who is an individual investor.
Well, first, congratulations on the strong growth after so many, many quarters. And a second congratulations for building up the cash reserve. When you took over, it stood at INR 100 crore. Today it's INR 197 crore. So congratulations on that. Now, my first question has to do with your opening remarks about what lies ahead for Subex. From what I heard from you, are you suggesting that you'll be building technology for AI agent orchestration and governance? That's the question tied to that future. If that's the case, then perhaps we'd need to change the company's name from Subex to something else. Okay. So you can answer that question. And the second question is, this squad of AI agents that we announced recently — is it the same thing we talked about six months ago on the investor call, when your chief strategy officer was telling me that they're trying to implement it on a laptop, because the customer doesn't want to invest in a GPU server? Is it the same customer?
No, this is a different customer. Actually, I want to openly acknowledge something here — and yes, it's a bit self-congratulatory, but I want to give a shout out to the team I have on this call. As you know, there are a lot of people on the call — sales, pre-sales, and our AI team. So I do want to give them a public shout out, because one of the great things that happened for us, Mahesh — and this was really good for us — was that the deal we announced involving AI agents literally closed in 90 days. From the first conversation we had with the customer, pitching him and saying, look, we want to do this for you, to actually inking the deal, it was a 90-day sprint. And we were able to sort of close it in 90 days. I think that is unheard of in telcos, and it was fantastic for us, actually. So in that sense, it's not the same customer we were talking about, but there are customers who still complain about the GPU and all that.
That said, this isn't that particular customer. Another point, though, is that our agents are now—well, to address that question, our agents are currently working on our Micropro. It's a multi-agent setup, and we can move from detection into, you know, investigation. As for my opening remarks, I think where this is heading, or where we see the opportunity, is whether it becomes an agent orchestration layer—we've already managed to establish that. What I'd like to push a bit further is to see if we can almost create—I think of it as a decision assurance and economic integrity layer. The reason is that even when an agent holds the proper authorization or follows SOPs, it still might not be the right economic decision, because of, you know, the decision it may end up making. So what we're moving toward is more, you know, building an economic integrity layer that ultimately serves as the verification layer for all the actions taken, you know, by humans and by agents.
So...
That is what I refer to as governance.
It's kind of a governance layer.
AI agent, orchestration and governance technology.
On governance — right, you've got a point there. So I tend to think about it more as, you know, ensuring that economic integrity sits, you know, at the heart of all our activities, and we want to be viewed as folks who can be trusted with that. And I want to push strongly in that direction to make sure we can validate everything that goes on in our ecosystem. So yes, that's the direction we're heading. Yes, you're correct about that.
If that’s what you’re planning for, then your framework is going to require zero trust security.
That's right. You've got that right. And we've actually already done that.
And can this group of agents scale up to 100 or 200 agents?
Yes, Mahesh. Yes, it scales. That's just how the architecture works. And these agents are going to carry out multiple types of activities. So, for instance, what we announced recently is mainly investigation, but other agents exist too. Again, since you grasp this from a technical standpoint, there are manager agents that hold a broader view. There are routers — you may be familiar with the decision models now emerging — that can route work-wise as well as cognition-wise to the appropriate agents, to do the work, to judge, and so on. So it's — and for us, another key criterion is doing this at a particular scale, because some of it gets audited. So it has to be highly repeatable. So it isn't merely that I can find 5 fraud investigations; I should be able to handle 50,000 and have it always come out right. So that's — so there's a great deal of engineering, or scaffolding, as we call it, surrounding the core agent as well.
All right. So, are you seeing a growing number of requests from other customers for this kind of solution?
We’ve begun pitching, and we’ve pitched to many—and frankly, we’re in commercial discussions with quite a few. My challenge to the team is this: you know, we’ve gotten off to a great start—a 90-day start. We need to see whether we can accelerate that further. But we are actually in active conversations with quite a few customers, because people have realized the productivity gain that something like this delivers.
Moving on to our next follow-up question, which comes from Sanjyot Khare, who is an individual investor.
Yeah, just now, I believe you brought up that there's been a solid response to the company's AI Agent Squad offering, and it's good to see that. There's another product, something like a partner ecosystem — partner management, I think — and in the last 6 months or roughly 9 months, we landed 2 big orders on that same platform. Are you seeing that this product is now getting more of a, kind of, are you seeing more response to it now, and will more such big deals be happening in the remainder of the financial year?
So that was our approach, Sanjyot, really. We — so PEM has always been about ecosystem management, or what we call PEM internally. And this is something that's been part of our offering for a long time, so it isn't a brand-new product as such. But what we did was we came back and said we need to modernize this stack, and we began reinvesting aggressively in that product. Because historically, what I've observed is that PEM has been a solid profitability generator for us. The bottom line on this product is actually pretty good. So we went and aggressively began, you know, putting investment back into the product. And we're modernizing it. We're adding newer features to it, and the market has responded quite well to it, actually. So our strategy is to push a lot more PEM into the market. And at the same time, while we're making these decisions and sort of modernizing and adding features, there's also a market dynamic that has played out, which is, you know, there are some players we used to compete with within the product, right, some other vendors. There's been quite a bit of consolidation in those spaces.
What that's accomplished is it has created an opening for us as well. So it's nearly, I view it as opportunity meeting preparation. We were getting ready to modernize it, add new features, make it — I'd say compatible with the next generation. And then the market opened up at the same moment. So it has worked out nicely for us up to now. So yes, I believe you're correct. Many of the deals we've announced lately have in fact been. So should this momentum keep going? I'd expect so, since even right now there are a good number of bids in progress around this product, in fact.
Sure. And secondly, another point from the presentation: there's a breakdown of revenue by region, and India's revenue is definitely quite small in scale — roughly about 3% of revenue is what we're seeing. And although it has still improved, I think it went up from 2% to 3%. Now, just a question: as the Indian telecom industry, even like new — I mean, Jio platform has really grown up and they've launched this IPO and everything. And plus even Vodafone is also now getting a lot of funds and all. So are we seeing that India traction will grow for the company, or is the company not really much keen to do much business in India?
So, I think if you look at it — and honestly, we'd love to expand our business in India — but I'm certain many investors joining this call also hold Jio shares. I'm sure that today you're benefiting from some of that. Yet if you go through the Jio prospectus, which we did, it states explicitly that they will develop their own stacks. They'll keep up their own stacks. They'll construct their own stacks. That's genuinely part of their perspective. So do I foresee a big uptick coming from Jio's, let's say, blockbuster listing? Honestly, for us I don't think so, because Jio's approach has been, you know, building their own. They're cooking in their own kitchen, right? So they're making their own kitchen. So that's not to say we don't make efforts to kind of, you know, work with them, but will you see a big uptake in the India business? To be very honest, I don't see it right now. So we already work with, you know, Vodafone, but we'll try and see if we can expand our business in Vodafone.
Beyond that, though, I don’t really expect that pie to grow in any meaningful way.
Thank you, sir. Because time is short, that was the final question for today. I'll now turn things over to Ms. Nisha for closing remarks.
Thank you very much. I believe a few questions did go unanswered. So to everyone we weren't able to reach today, please send them our way. We'll make certain we get back to you. Overall, though, thank you for the thoughtful engagement today. These conversations matter a great deal to us, since we learn just as much as you do from these calls. It also gives us a helpful perspective as we think about the business and the road ahead. So what I'd like to say is that there's plenty to look forward to, yet there's also a lot of work still to be done. Thank you for your continued interest and support. I look forward to our next conversation. Before I go, though, a very happy Diwali to all of you and your loved ones. I hope you all have a wonderful festive season, and we'll see you next quarter. Thank you very much.
Thank you, ma'am. That brings us to the end of today's conference, everyone. We appreciate you joining and for making use of Door Sabha's conference call service. You're welcome to disconnect your lines at this point. Thanks again, and enjoy the rest of your day.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Subex Limited transcript - plus 131,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Subex Limited earnings transcripts and 131,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.