Hello and welcome to On the Record with me, Shereen Bhan. We're at the Infosys headquarters here in Bengaluru, and the reason we're here is because Salil Parekh has decided that he is going to move on and put a successor in place. I was here in 2018 within the very first month that Parekh had taken over, and today, as he gets ready to exit Infosys, he is the longest-serving CEO, a non-founder CEO of the company. Just to give you some numbers of what the growth has looked like under Salil Parekh's tenure, sales growth of 11%, profit growth of 8%, the stock has returned 7% and ROE of 28%. That captures the tenure performance for Salil Parekh here at Infosys. As he gets ready to move on from Infosys, but perhaps also start up all over again, joining me now is the man himself, Salil. Thanks very much for joining us. It's great to have you. As I said, life comes a full circle. I was here in 2018 in January when you had started your tenure, and I'm here as you have decided to transition.
But, you know, Salil, they could have moved a special resolution. They could have given you more time. Did you want more time? Was that even a conversation?
So first, Shereen, it's great to have you back here. I think, as you just mentioned, it was wonderful when I started, you were here, and I think it was a bit of a war, at least for me, given everything that was going on in the company. And today, in a very different place, I feel at least the company is. So thanks for being here, and thanks for doing this. I think, as you shared, the last several years have gone, I think, quite well for the company. We've grown very well, 10 billion to 20 billion. The whole focus on digital has helped us. Now, AI is really well established. We had 8% revenue last quarter, and we can see, with the market evolution, where that's going. So in all of that, what I had the opportunity here, you know, leading an iconic institution, was just an incredible blessing. And given all of that and the way Infosys is positioned, it looked like the right time. We have a phenomenal team within the company. We have leaders in the market across delivery, across all of our functions. So I feel really good that this is a great time to have that sort of completion of my second term and have a new person running the show now.
But, you know what they say, third time is a charm, Salil. So you didn't want to take that chance.
No, absolutely. I think, you know, all of those things are good things, but equally, Infosys is in a very good place. And we've had, I think, an incredible run going through many different sorts of global environments. And given where we are, given where we've positioned ourselves on AI, it's a good time to make sure that we focus on all of the things that have happened in the company that have supported us for the good.
You know, I'll get to AI in just a second, Salil, but let's walk through your wonderful campus here. Let's talk a little bit about Ashish Dash, who, of course, comes in to take over. October is when the transition, in a sense, the hand-holding will start. He will move from LA to Bengaluru. Were there only internal candidates that the board considered, or were there external candidates as well that the board looked at?
So, the first guy is a phenomenal leader. I've known him for many years. We've worked together. He's a good friend and done very well with employees, with clients, and also with the portfolio that he's looked at, you know, both in terms of making sure there's growth, making sure there's economic returns. So very, very strong performance. The board went through, I think, a very good process in the way they looked at the situation and made all of their decisions keeping in mind what we were looking at as a company into the future.
And of course, we have an excellent set of leaders within Infosys, and we've had this. I mean, the company itself has been here over 30 years, but many of the senior leaders have been here for 20, 25, 30 years. So fortunately, there's a good set of people in the leadership domain within Infosys itself.
You know, Salil, as we look forward, I also want to look back, and I also want to ask you what you believe we should now calibrate as far as growth expectations for the company as well as for the industry is concerned. And I was just looking at the numbers to really get an arc of what you've kind of, you know, overseen. FY19 9% and of course we saw that big bump up in FY22 of almost 20% and then about 16% in FY23 and then now back down to about 3% FY26. You know, would it be fair to assume that we're going to now, for the foreseeable future, be stuck in this 3 to 5% kind of range?
So first, I think what you described, if you look at the organic growth we've had, we've had across that time, from when I started, the fastest among the large companies. What we see, and you've seen this industry and Infosys for years, there are cyclical aspects of the tech spend. So, for example, when global macro GDP growth is more managed and doing well, the growth typically on tech is higher. What we see now is really the start of the AI expansion. If you look at our own AI growth, we shared publicly, it's double-digit quarter-on-quarter growth over the last several quarters. So it's a very strong foundation. If you look at it as a standalone at over 8%, it's already a very large part of what we are doing. A lot of work on agents, a lot of work on modernization, a lot of work with the new coding tools. When the other pieces of the global environment, the macro, the GDP, all start to come together, there'll be more tech spend overall as well, because part of what we've seen recently is much more attention on cost. And so my sense is as that cycle comes off, we will start to see good growth again. Now, of course, there's a difference. When I started, it was a $10 billion entity; today we are a $20 billion entity. So that has its own changes in how the growth comes, but I see over the medium longer-term horizon, growth more a function of the macro and how we execute on AI. So good outlook on growth.
You know, Nandan has given a number of between 300 to 400 billion is what he expects as the addressable market, the AI opportunity by 2030. Now, if I were to ask you how much of that value do you believe you can realistically capture over the next few years?
So there, really, it's up to us how we execute. So today what we are seeing is for our client base, our industries, when clients are looking at making AI work, let's say a bank wants to build a new KYC agent, they are calling Infosys to work with them with all of the foundation model companies to build it for them. And in any of those processes, any of the functions, clients—the telco company wants to modernize from an older technology to a new one using AI—
Infosys is the one they call. So we can capture a huge amount of that. That's really an execution play now for us in that $300 billion. We've defined the six areas, you may recall, of what are the specific places where the growth is. And now we've started the execution, and it looks pretty good. If we continue the execution well, which every indication is we will, that market share could be quite massive.
You know, let me take you back to the days of the digital transition, right? I mean, companies like yours started reporting digital revenue to the point that you didn't need to make that separation anymore and everything became digital revenue. When do you believe that's likely to happen as far as AI is concerned? You've just started reporting AI revenue, and it's 8% today as far as the company's concerned. When do you believe you will no longer need a separate line?
So there, again, that description is exactly what we are looking at. As you remember, digital we started reporting when it was about 20%, I think, and we took it over three or four years maybe to over 60%, when it became part of everything we were doing. So I don't know if it's exactly the same because there are indications that AI could go faster.
But it also depends a little bit on the macro, because today in AI already there's a lot of attention on token cost.
So this is moving very quickly. Meaning, even if you look a few months ago, there was only discussion on can we use the best models for everything. And today there's already discussion on can we use the appropriate model for any task we're doing. So within Infosys we have built something called Topaz Fabric, where we work with 15 models. And within each of the models, we work with the older version of that model as well. So if you look at token cost, for the most recent models, if it's 100, for the most efficient it's almost five or ten in terms of ratio. So all of that has to play out. When we do that, my sense is today we're only looking at companies doing X; can we do it better with agents? Tomorrow we're going to look at, instead of X, can we do 10X in a company? And so those sorts of changes, we can expand it even more massively. So I don't know what that horizon is, but it's a transformation, it's an execution, it's a pivot. So the faster we go, it could take, you know, fewer quarters, it could take a few longer, but the path appears fairly well laid out.
You know, I think at the start of the year, the nightmare for Indian IT and the big question that was being asked is, will you be cloudified? And I think from a market perspective we saw the kind of value erosion, the wealth erosion that took place. The recovery in the last few weeks is still underway, of course, still significantly lower than the kind of peaks that companies like yours as well as the sector have enjoyed. But is there now an acknowledgement or an appreciation that it's not all over for the Indian IT services industry, because you know that the obituaries were being written at the start of the year.
So my sense of it, we've always had the view that the work we do for large companies, which is our client base, requires that as they bring in new foundation models, given that we know the context, what's going on in the landscape in a company, that we can help. And that seems to be getting demonstrated now. So we even internally had that view. There were different views externally, but we seem to be seeing that coming back. But the real test in my mind is as all of that comes to bear over the next few years, but we can see the path to it for sure. If you look at it even like recruitment, last year we recruited 20,000 college graduates. If we didn't have that belief, we wouldn't have done that. This year we already have the same similar plan, another 20,000. In the first quarter we've already done 4,000 college hires. So we see, you know, the individuals coming in from colleges, absolutely AI-native people. We are training them on our own existing approach and then making sure that they learn all the new tools that we have. We have about 80,000 people already in the company today working on coding tools for clients and some internal projects. So to me, at least, I see this as a great opportunity for the next phase of the transformation. Infosys, my sense, we will continue to scale and grow, and I think the sector itself, as long as we're executing to go into AI, should be okay.
Yeah, you know, that's the future. Let's talk a little bit about the near term. And I think what the market was concerned about at the end of your earnings, and of course I don't want to go too much into the earnings because you've done extensive conversations around that, but the very sharp cut in guidance, especially on revenue, that has worried people. What is it that you're seeing in terms of the significant deterioration in the last three months itself that's forced your hand to announce such a sharp cut?
So we changed the guidance. One of the reasons, and we shared all of this, was we had a one-time event in terms of client change of plans, which we described. That was the bulk of it. And then we saw, the macro being a little bit uneven, some of the changes in volume. When something like this happens in the first quarter, it has a flow-through through the whole year. But outside of that, if you look at our large deals, very strong. If you look at our margin, actually higher than what most people were expecting. And we see a stable sort of an outlook for margin. If you look at free cash flow, very strong at 950 million. So all of the business parameters are looking in good shape. And then we had AI, which is looking strong as well. So my sense is overall the business is in excellent shape, making sure that we go through all of the changes, whether it's with AI, which is a transformation, and with the changes in the global environment, which is just making sure we are navigating through that.
Right. Salil, you know, I wanted to get your comments on what we heard from Arvind Krishna at the end of IBM's earnings. And he talked about the fact that because there is such a priority on infrastructure spend at this point in time, that perhaps people are holding back on traditional IT services spends. Are you sensing that from your clients also?
So we have seen, basically, our clients much more focused on cost activities. But we have not seen a change otherwise in the spend. I think a lot of clients are becoming more sensitive on how they spend, what we discussed earlier on the token usage. So that is an efficiency that they are driving through. We continue to see, for example, on AI, a lot of spend, which is where we're getting this sort of growth. And the other piece is a little bit more attention on cost, which has been going on more in light of the macro, really for the last few quarters, year or so. So that is how we've seen it, without any changes.
Yeah, you know, competition of course has been intense in the sector, but is it particularly intense now? I mean, many of your peers are calling out the fact that competition is off the charts. Somebody's breakfast is your lunch, somebody else's dinner is your breakfast. I mean, how much pressure is that putting on margins, for instance, and do you expect that to only worsen as we go forward?
So there again, we've taken a view to be extremely clear that we want to make sure that anything we do fits what our economic profile is. So you've seen our margin has remained resilient all through this, and actually last year it improved a little bit on the prior year. And we see that being part of what we do. The competitive intensity is quite heavy, which you described.
Worse than what you've seen in the last 10 years? More intense in the last 10 years?
I've always felt it's equally intense. So I don't see that it's increased, but maybe there are different companies at different stages of their life cycle. So some people maybe get a little bit more intense when they're at different stages. We've taken a view to make sure that we've done over $3.6 billion of large deals. So we are winning huge deals consistently. Equally, we've been very careful on one or two instances where the economic outlook is definitely not fitting what we're looking at.
How many did you walk away from because economically it was not making sense for you?
There, you know, it's hardly... There was a discussion that we had at the time of the results where we talked about one specific instance in a European client. But it's literally nothing which is in larger numbers. Very small. One or two in that sense, because we are extremely competitive, but we are also really clear that there are some things that we want to make sure that we manage what our economic sort of profile of the company is. And these things sometimes in the short run may be sort of unusual for other people. Other companies might see some benefit, but we are very clear that we want to do this in a way that makes sense for the medium and long term.
You know, what I also want to understand from you, Salil, and I think this is something that the industry is grappling with: as you move more towards an agentic economy, as you move more towards agents within an organization, what is that going to mean structurally as far as margins are concerned? You know, what is the impact going to be on the way that you price?
So there are different models that are already coming into place. So first, we have in small instances things which look like agents plus our own software engineers sort of working on joint teams like that. Second, we are looking at things where there are business outcomes which also become part of it. It's not all of it, but that'll become part of it. But these things, at least what we've seen in our industry and in Infosys, is it takes time to make those changes. For the large part, it's still similar the way the pricing works. And there's a lot of pricing productivity benefit which clients themselves are seeking when they see all the benefits that we can deliver. And so all that will also go on. However, we are building these new things which are quite incredible. If you look at some of the agents, the work that's been going on, some of the work we're doing with coding tools. So there are places where the pricing is, let's say, quite good for the company. There are benefits that the clients are receiving which are quite massive. So both of those things are happening now.
So the benefits that the clients are receiving on account of productivity, do you get to leverage on that or, you know, how does it work?
I mean, usually there's some sharing. I think the bulk of it still goes to the clients, but we do get some of it because in many instances when we bring those ideas to clients, they immediately see the benefits. So, for example, we're doing a project on modernizing a technology environment. So it will have a one-time cost which the client will bear and pay us, but it will also have a huge benefit for their running their infrastructure, their environment ongoing, which will be much reduced. So the client will be extremely happy in this instance, but the overall mix, they will benefit, and we will receive some benefit.
You know, how agentic will Infosys look in the next five years? And what is that going to mean as far as your workforce is concerned going forward? You're hiring today, as you pointed out, last year 20,000 fresh graduates, but as we move more into this agentic future, what will it realistically mean both in terms of hiring as well as in terms of workflow?
So the work is, you know, it's like today's foundation models that exist are like a very great efficiency tool that an engineer can use, a software engineer can use. So for them, they're doing different things by using these tools. It's like an advanced toolkit that you have with you. So my sense, and this is evolving, so it's not like we've reached an asymptote or anything in the foundation model technology yet. So my sense is that evolution will keep enhancing. So the people who learn, which is our reskilling, or the new graduates who learn these technologies well, will be more the beneficiaries.
They'll be really native AI people, as you know, as years go on in the next year, two years, five years. And that will be the type of person that works. We have a plan to sort of have 6,000 frontier engineers. It'll be over the next few years, but over five, seven years we'll have more and more of that sort of capability within the company that can work with clients. So my sense is they'll be absolutely AI-native type of people who are working there.
So, these frontier engineers, would this be the same as FDEs, or is this a different league altogether?
