← Throughline · holding view Deep analysis ↗ Q3 FY26
INFY Infosys Ltd · IT services Q3 FY26 · concall
Pattern: net ai opportunity after

Refused to commit on services revenue intensity per.

1 deflection · 5 weak · 10 clean — pushback across 6 of 16 Q&A turns

Focused evidence 6 of 16

Ankur Rudra · Unknownweak

So, the first question is, you highlighted in the morning that the net new opportunity from AI will be $300 bn to $400 bn. Could you also elaborate what do you think is the incremental opportunity is $300 bn to $400 bn on new services. Could you elaborate what would be the potential for the net opportunity adjusting for the compression that you might see in many of these services? A lot of examples through the day highlighted that everything can be done faster, fewer resources or from months to weeks. What does that mean from a compression perspective? And what are we looking at on a net basis?

So, what you are saying is we have talked about the growth expansion and what is the net number. So there, what we have shared in the morning is really on the six areas, what we see as the expansion from what we have seen externally, the $300 bn to $400 bn quantification from a couple of sources. On the compression, we have talked about, again, what we understand to be where the compression is coming from, whether it is on application development, whether it is on infrastructure. We have not quantified that number for any external use at this stage. We have said that the expansion number from what we see today looks larger than the compression number.

Vibhor Singhal · Unknownweak

Yes, hi, Salil. Just in response to the first question that you answered that we are kind of quantifying the compression factor. If I were to take a look at it, let us say, direction- wise, let us and compare this to the earlier digital cycle. There also, we saw an initial compression in the IMS and some parts of our business. And then, of course, the new opportunity took place. So we generally see a compression, then that followed by the opportunity that comes in. In terms of relative comparison, where are we in the cycle at this point of time? Have we seen the peak cannibalization of revenues? Do you think we can further go down? Or do you think from here, we are more closer to the inflection point where the net revenue from GenAI becomes positive? So in the entire adoption cycle of this technology, where would you place us at this point of time?

So there, if you step back a little bit, there are usually multiple dynamics that are at play. Typically, if there is nothing else changing, the macro is one important factor for Infosys for IT services revenue growth. And then there is a change of technol ogy, which is going on, which is another factor. So what we see today is that in many ways, for us, the macro is improving. If you look at what we see in the large markets we are operating in, for example, in the U.S., we see overall with the changes in regulations with tax reduction and maybe an impact on interest rates, we see some move on the macro. On the tech cycle, it is difficult to say where we are because each of the six areas are different in the way that will play out. So for example, some of the things on process can go very fast. For example, in customer service today, there is a huge move. And there, we will be going after areas that we are not today currently in our revenue base. So this is all incremental for us. But we have the technology, we have the partnership, we have the capability. In some of the other areas like modernization, we have a lot of good technology. It is something that will again be incremental now. Some of the others, maybe as the timeline flows out, we will see how they play out. Data is happening very much more quickly at this stage. In terms of the compression, those are things that, as we mentioned throughout the day, we see that visible, but it is not something which is large and it is not something that is insignificant. But we do not see an acceleration of that either at this stage. So my sense is if you look at our current business situation, next year, we definitely see in Financial Services a strong growth, we see in Energy and Utilities, a strong growth compared to this year. So we are already seeing signs of visibility. It is a function of AI leverage, meaning more AI being used. It i s a function of macro and it is a function of the compression. So these are a little bit more intertwined than would otherwise be.

Pankaj Murarka · Renaissanceweak

Hi. This is Pankaj Murarka from Renaissance. Salil, I have two questions. One, when you called out that we have 5.5% revenues coming from AI and in the context of, you know, that is still I think a small number, while we have seen a lot of use cases today, but in the context of, you know, the Fortune 2000 clients, I still think there are few and even the deal sizes seem to be small. So my assessment is that probably average deal size is about $4 mn to $5 mn or something like that. So how far are we where we start seeing $50 mn, $100 mn deals and where the adoption really becomes accelerated?.

So there, I think one of the reasons , what we shared today in terms of AI is to give a real depth on what we are doing on AI. Now as you rightly pointed out, we shared the number of 5.5%. We still have a lot of other work that we do within Infosys, which is making up the other parts of the business. What we see in this AI activity is it is going across many things in the areas that we described and becoming part of almost every discussion. And so our sense is it will now continue to grow. We will see how that growth is but starting at that sort of a level, there is a long runway because essentially over the next several years there will be a shift. If you sort of look back a few years, we started to call out our digital numbers when they were around 25% or 20%, something like that, and we had a shift over 3, 4, 5 years where it became 65% to 70%. And so that is the sort of a play that you have. We do not know if this is going to go in that sort of a range in 18 months or will it take 7 years, but we are well on our way with what we have created to start to play it as our clients are absorbing it. And even if it goes faster, we are ready, even if it goes at that pace, we are ready.

Aditya Chandrasekar · UBSweak

Hi. This is Aditya from UBS. Just a couple of questions. So you guys have spoken a lot about Topaz and it is good to see, I think it is moved beyond, you know, those pilot use cases to more enterprise-wide use cases. But if you could also touch a little bit on how we should think about the pricing models there? Because for example, in the walkthroughs we learned that some of the projects that needed huge team sizes have now, you know, been compressed to just use of the platf orm and maybe a very lean team. How should we think about how the pricing is evolving in or pricing model rather is evolving in those kind of projects and any rough or, you know, a framework at least to think about margins as well in those kind of projects? That is the first question. And secondly on headcount, I think you have given a plan on how you will be reskilling talent and hiring more specialized talent etc. But as of today you do have a wide kind of a fresher or bottom of the pyramid talent. How should we think about the utilization of that workforce now because incrementally we will be getting into more and more projects where we will have leaner team sizes and maybe just more specialized workforce rather than the kind of fresher coders, to put it simplistically? These are the two questions. Thank you.

So, I will start with pricing. So if you look at the pricing models today, the models are evolving, right? You have various examples of outcome -based pricing, you have examples of pricing which is combination of outcome-based pricing plus an agent pricing or a platform pricing. But I do not think there is going to be one model that is going to apply to every client. It will depend on the client context, what does the client want and how well we are able to justify the value that we are creating for the client. So it is always going to be the combination like it is always been in the past, you know, we did not have just one model. There were various models that worked and depending on the client's context you close on the pricing model and I think that is how it is going to remain on that.

Surendra Goyal · Citiweak

Yes hi. Surendra from Citi. So through the day across the presentations we heard a lot about value generation, savings, productivity. So my question is, is there a way for Infosys to capture that value better? And does AI kind of result in any change to that, either for good or for worse or it really does not matter? Because one of the key issues has been that like over the years we have seen a lot of presentations across the industry, talking about millions and hundreds of millions of dollars of value, but again it seems like a most of that goes back to the customer. So any way to kind of capture the value better? Thanks.

I think there, there was some discussion in some of the client examples on outcome based. Now if we managed to do some of those with some sharing, we might see some of the benefits of it. Today, if you -- in one of the earlier charts, we sort of showed the model is moving faster than the reality at the enterprise. So what tends to happen is the enterprise people at a more discussion level are expecting the model type of benefit in the pricing or the cost and we are not able to make that happen. So today it is not more equitable, but in the future when it becomes more aligned, the model and what is in the enterprise, then you could see that part of it could be there.

Kawaljeet Saluja · Kotakdeflection

Right. And Salil, for agents of third party, let us say what is the services intensity? So for example, let us say a dollar is spent on a third-party model, and you basically customize it and configure it for the client environment, then what is the services revenue that you get versus a dollar which is captured by let us say a frontier model company or an external third-party agent provider?

So there, we do not have very deep exact stats on that that we can share. What we see is different, because if you look at a software development lifecycle part of it as you know is the cost of building the software. So, in this case the agent, but there is a significant part beyond that which is integrating it into the environment, making sure that it is working in that environment and the performance attributes and then the security attributes. So we have got some examples of what we have done as we shared today in these situations, but we do not have a statistic that we can share that one equals X because of the ratio. As we go through the next few quarters, we will definitely internally build up a larger data set on that, and then we will see, if that becomes something that we can share publicly here.

▸ Other Q&A (10)
Kumar Rakesh · Unknown

Hi. Thank you for hosting us and taking us through in detail through your verticals and service wise capabilities. What I have understood is that you spoke about one of the key advantages Infosys has is strong understanding of clients' data estate and the context of the client. Now that would be something which many of your peers would also claim to be having. The second part is the partner ecosystem, which you are developing and building strong relationships. Again, that is something which here or there, put together your competitors also would potentially say that they have. So what exactly in terms of delta or incremental capability that you see that Infosys has that helps you to have a better right to win wh en going for enterprise implementation of AI tools?

So there, the focus in addition to the points you mentioned is one on the platform and Topaz Fabric capability that we build, where we built our own agents and we can integrate other agents. The other is the way we have identified clearly what the six areas of growth look like and how we are executing on that, working with our clients to make sure that we are positioned in that by reskilling our people, by building investments in those capabilities and making sure that each client, we are building that capability to grow into those areas. That is where the real difference, the execution is where the real difference will come.

Aditi Patel · Unknown

Has our pipeline to TCV conversion timelines improved, because now leveraging AI, we are able to build prototypes or working models much faster and clients are able to view the ROIs much faster?

So, if you look at the large deals that we have signed in the last few years and last few quarters, we have not really seen the timeline of those large deals shrinking at this point in time. We have seen the deal timelines remaining similar based on what we have signed till now. Of course, the future is yet to be seen as we sign. But at this point in time, we have majorly, we have not seen the deal timeline shrinking significantly.

Sudheer Guntupalli · Unknown

Hi, Salil. Yes, thanks for the insightful presentation. So recently, one of your competitors had made a comment that many ERP migration programs are seeing a significant compression in time from years earlier to weeks. So do you see that practically happening? Or you think it is probably a corner use case in a specific context and that might not be extrapolatable to the entire ERP implementations or migrations as such? Thank you.

On ERP migration, we have not seen that, that specific point is valid. So what is valid is what I think some of our leaders talked about, which is when you look at modernization, which is not only an ERP migration, but it is an overall modernization approach from old legacy landscape to a more current landscape. There, we see that the speed or the timeline is much more compressed. And of course, the cost is much more reduced.

Jonathan Lee · Guggenheim

Hi. This is Jonathan Lee from Guggenheim. Thanks for hosting us today. A lot to be excited about here around the AI opportunity. Can you help us understand the level of investment necessary there to reskill and hire laterally to meet the opportunity in hand? And how should we think about the impact to margins there ex potential currency benefit?

So there, I will start off, Jayesh may have some things to add. One of the things we have been very clear about is we want to make sure we invest in this AI capability build- out, AI Topaz Fabric and platform and other tools that we have built to invest in scaling that up, scaling up the go- to-market and scaling up what we need to do in terms of training. We have also put in place some years ago, a very strong program to support our margin and make sure our cost is more and more efficient. So, our view for the period in the future is we will maintain our margin guidance, and we will take all of that, that we save, which is quite substantial from our margin program and invest that into scaling up AI even faster. So we are ready with that from the operating, from the income statement point of view. From the balance sheet, we are also ready to make as required, appropriate acquisitions, which will fit our overall value framework, which is what we have done in the past. With that sort of a mindset, we will continue using the balance sheet as well.

Pankaj Murarka · Renaissance

Sure. One of the most important things that we learned during the discussion today is the context. So in the context of, you know, what you laid out, the new opportunities of $300-odd bn over the next 5-6 years, for a long-term stakeholder or an investor in Infosys, how should one think, what are the three or four things that will change from a financial matrices perspective probably 5 years out as we navigate this journey from where we are today?

So, if you look at, you know, every tech cycle, and this is no different from a cycle perspective, of course the metrics can be different. But the way we have always articulated it is, if you are riding the tech wave and ahead of the curve, it should reflect into better growth, better RPP and therefore better margins. We said that in the digital era, we are saying the same thing today. If you look at our RPP for last 2 years, we have delivered, you know, superior RPP. Our RPP has grown 3% both FY '25 and FY '26. If you look at margins, it is showing resilience. We grew margins by 50 basis points last year. We are stable margins this year despite all the investments that we talked about. So I think that is exactly what is going to boil down to, if you are looking at, you know, any tech cycle in my mind, from a services perspective.

Gaurav Rateria · Morgan Stanley

Hi. Gaurav from Morgan Stanley. Nandan made a comment in his presentation about build versus buy. Do you think the lines between software and services are now getting blurred, and what does it mean from addressable market perspective for service providers? And second is a question related to the evolution of pricing model that you talked about. If it creates room for non-linearity, what is the headroom that you get from investment point of view to accelerate your journey in AI? Thank you.

On the first one, I think my understanding of that is essentially it expands massively the amount of work that we can do , if we can look at some of those. And I t hink there it will probably be some of the things on the edge, maybe not the core sort of systems of record and so on. But you could imagine some of the things on the edge which could be more easily built, as Nandan was saying, in the build as opposed to the buy. And then if that is readily doable, and it is effective for the client, then there will be multiple, it is not going to be one thing for every client. And so there is going to be different builds for different clients which my sense is will be a larger opportunity for Infosys.

Kunal Tayal · Bank of America

Kunal from Bank of America. Given all your investments into small language models as well as proprietary solutions, I was wondering if there is a bigger opportunity that you foresee in either of mid-market customers or then emerging market kind of customer base? Mid- market because , I wonder if there can be larger turnkey programs that can come your way and emerging markets because if it is not going to be a labour -intensive model, can the profitability of these projects now meet thresholds better than earlier?

So there, first the small language model, I think for us, is a very good indication of the depth we have, and it is very useful on a limited data set in a large client already today. So we are seeing some benefit of that whether it is more for alignment or more even for code development. You heard I think, in one of the sessions the discussion on our own model for code development into that. We have not looked right now at the mid-market because the cost of sales is very different. So we have to figure out if that will work in that market. And emerging markets, we have not looked beyond, I mean, I would not call them emerging markets, and there are growth markets that we have looked at. For example, we think, markets in the Middle East are very strong. But it is not like an emerging market but it is a growth market , where we will look for some of these models but right now the adoption, it also depends on the adoption in that environment, in that geography, before we can go and adoption in the mid- market as a client base also.

Sandeep Shah · Unknown

Yes. Once the enterprise clients enter the post AI adoption where most of the application modernize, data layer has been created, cloud migration has happened. In that phase the investor worries the terminal growth rates could be much lower because of the applic ations, data could have been modernized.

So there, I think in terms of what happens in the post -AI world, of course that seems a little bit far away today but what we think is, I was with one of the people who was building the foundation model a couple of weeks ago, and they gave a good sort of example. They said that the amount of software demand that is there for writing software is becoming 100x in terms of the size. So even if you go into the post-AI world, there is productivity impact, we still see a huge amount even if you assume a 10x productivity benefit to a developer of what is available in terms of what is to be developed, what is to be built. So the post-AI world in my mind is not a static world where everything is done, it is a world where there are large enterprises starting to use agents in many areas, having very strong platforms, using for example Topaz Fabric as a capability but also building new functions, new features, managing things. It is much like what we have today where there are new things happening even on platforms, older platforms, which let us call it are somewhat stable today in terms of they are implemented. But there is always new work that people are looking for. But that is the nature of the economic growth where there is always new features, functions that tech is driving and more tech across the enterprise which can give you different areas to work on. So my sense is in the post -AI, of course the transition is very exciting with all the work we will do, but there is even more work, because there will be more and more things that will be going on in that post-AI era.

Kawaljeet Saluja · Kotak

Hey hi. It is Kawaljeet from Kotak. Thanks a lot for the presentation which was quite insightful, Salil and Jayesh. I have a few specific questions. The first question is that for AI agents that you are deploying in a client environment, are those homegrown or are those of frontier model companies?

So on agents there are multiple people who are building agents. So first we are building agents. Second, some of the foundation model people are building they are a little bit more broad -based. Then there are third- party companies that are also building agents. And then of course the public cloud players are building agents of their own, and other third-party agents that they are providing. So there will be a host of those agents from which some selections will be done. And of course clients are also building agents.

Shaan Chugh · Capital Group

Hey Salil and Jayesh. Thanks for organizing today. This is Shaan from Capital Group. I guess just a quick one, you know, for the last many quarters our sort of net headcount has been pretty flattish as we have muddled through sort of the macro challenges and so on. I mean with the AI services picking up, are you internally gearing up for sort of net headcount to start ticking up again? I know we are hiring sort of a lot of freshers and we also have that counterbalance with some internal efficiencies and of course just natural attrition. So when you factor all of that in on a net basis, is that something you are gearing up for this year or should we expect that to still take some time given the AI-centric efficiencies that you are factoring in?

For the year, we have added ~13,000 net headcount for the first three quarters. My sense is, we will continue to add headcount as we go through. And it sort of comes back a little bit to an earlier discussion, we were having, which is there is a macro element and there is an AI element. And we will see, my sense is the macro will potentially be better, and of course we have a very good sense of the AI opportunity set. So when you put both of those together, if in the last three quarters we have done ~13,000 net headcount increase, my sense is we will continue with the headcount increase in the coming quarters as well.

▸ Prepared remarks (5 blocks)
Thank you, and great to have you all here in these tumultuous times. Today, I will talk about tech transitions. I have had the fortune or misfortune of being in this industry for more than 40 years and have seen a lot of transitions. So I thought I will talk less about that and more about why this time it is different and what are the implications of this transition. This time, the AI transition has been much faster than earlier transitions. If you look at the number of years it took to reach <strong>1 billion</strong> users, Internet took more than 10 years, smartphones took 5 years, AI is taking a couple of years. Now what has happened this time, is that this is a much more fundamental change to the way businesses will operate. So this is not a layer of technology. This is a fundamental change in the way we do things. Obviously, there is a technology dimension and it is all about having AI native architecture. But there is a whole business dimension to this. We cannot run business the old way, and businesses have to change, the customer journeys have to change, all those things have to change. Now one clear learning we have is modernization of legacy systems cannot be deferred anymore.
If you really want a firm to take advantage of AI, you have to fundamentally clean this up. So this is a massive, massive cleanup job, which everybody is dealing with. Many large companies are spending 60% to 80% of their IT spend on maintaining systems. There is no business value out of that. They want to go from 60% or 70% maintenance and 30% new systems to 30% or 40% maintenance and 60%, 70% new systems. Now the other thing which is there is, as AI becomes a bigger part of the spend, the balance of advantage is moving towards 'build' rather than 'buy'. But the main thing is that the technology is far ahead of its deployment. Because of this race and spending billions and some AGI and all that, the technology is moving faster than the ability of enterprises to deploy it. So my view is there is no opportunity gap. If anything, the opportunity is bigger than ever before. So do not get distracted by that. So it is not an opportunity risk, it is an execution risk.
So today, we are doing AI work for 90% of our large 200 clients. So this is not something which is just here and there in pilots. We essentially see six large areas where there is growth opportunity for AI services. So what I am showing you in the Hexagon in the AI value framework is not theoretical. These are things that are actually happening on the ground with Infosys. This is what we are executing. And this for us today represents 5.5% of our revenue in Q3, and it is growing at a robust pace. We understand that the opportunity is between $300 bn and $400 bn in the year getting to 2030, so over that time frame at that time in 2030. At the same time, we have several entities, have made estimates. We have our own views. The AI productivity leads to compression in IT services. However, today, we have a clear view that the opportunity is massive and large and that will become the driving force of what we will grow and drive through in the next coming years.
This for us today represents <strong>5.5%</strong> of our revenue in Q3, and it is growing at a robust pace. We understand that the opportunity is between $300 bn and $400 bn in the year getting to 2030, so over that time frame at that time in 2030.
if you look at this year, 9 months into the year, we have been able to maintain our margins stable. That was on the back of FY '25 where we expanded our margins by <strong>50 basis points</strong>. All of this is after absorbing all the investments that we saw through the day today, whether it was partnerships, whether it was tech investments, whether it was training investments, the sales and mark eting investments that you see in our P&L already which has impacted, you know, 50 basis points. So we have absorbed all of that and delivered on our margins and maintained the margins stability. If you look at our RPP for last 2 years, we have delivered, you know, superior RPP. Our RPP has grown 3% both FY '25 and FY '26. If you look at margins, it is showing resilience.
Watch next