Your vision statement of becoming the leading AI-led tech services firm, how should investors evaluate that over time? If you can just highlight which metrics, what drivers, what should we see to see the success over the next three to five years?
So, we define those five pillars. Each of these five pillars also have sub parameters. Internally, the way we are looking at is, how do we drive these five pillars? For instance, we talked about what is the business value? How do we reimagine the business value chain as the fourth pillar? We talked about how we redefine our services through leveraging AI. So, each of these have sub pillars. The way internally we are going to be measuring is on how we are progressing on each of these individual metrics. See, obviously, overall metric could be on how would be the AI driven revenue that comes to the organization.
My first question is around the new-age services revenue, which you called out at about US$11 billion. That means there is still a large legacy revenue base, almost two-thirds of it. How do you see managing that in the coming years? More importantly, this new-age services revenue of $11 billion is growing at about mid-single digits, so the growth is not phenomenally great in that piece as well. The new-age services growth is not significantly strong, and the legacy revenue will most likely continue to get cannibalized as we expand AI implementation across our service lines. How are you going to manage your growth in this overall context?
Kumar, as pointed out, our overall new-age services revenue is growing faster. First of all, you'll understand that, there is moderation and subdued performance due to overall market sentiment; that's the overhang we have. But within that, new-age services are growing faster than traditional services. With increased investments under a stronger strategy around new partnerships, new investments, and initiatives like data centres, and our new strategy, we believe new-age services will grow faster to offset any deceleration or drag in traditional services.
Got it. My second question goes back to something we discussed earlier: how do you define being the world's largest AI-led services company? Are there any specific metrics?
Kumar, we will try to publish. See, at this time, we took it as a vision and identified five pillars we are working on. Like I was telling Ankur, we look at the sub-pillars where we are putting our energy and effort. The obvious metric is AI-driven revenue; we have published where we stand and we will continue to publish it as we go forward, and that's the external metric. But internally, we are going to be driving all the parameters. To us, those parameters are equally or more important because there could be a lag in the external revenue we report, but it will keep us honest. We will focus on internal parameters and individual line items Aarthi described.
Secondly, the US$1 billion spent that we highlighted, can you tell us how historically it has trended? Is the intensity of investment on rise as we are going through these technology changes, or have we been consistent at whatever level you want to define in making those investments?
We have been making early investments across, and if we take it as a percentage, as I said, our focus would be to make the right investments. Some part of the investments will also be repurposed into things where we want to focus on. And we would, if required, to the other question previously, we will elevate our investment need also. 'Growth with profitability' would be the mantra. It need not be that we will shrink our investments in the near term to let go long-term growth.
My second question first is to Aarthi, and I'll let you think about it, which is, how do you foresee your organization structure to look like by 2030, because you will have your agents as co-workers with your engineers.
Arup, your question on how the organization structures would look like. You know, in a world where humans and agents work alongside, I think this is an area which is evolving. The more we move to Level 4, Level 5 autonomy is where you'll have agents which can do a task that a human does. Today, if you really see Level 2 and Level 3, that is where the human is primary and AI is in a supportive role, in an assistant role, if you will. So, this is something which is evolving, but where we are seeing this play out currently is more in the autonomous GBS. In business process, we're seeing that there are certain tasks in the business process value chain which are being completely done by agents. In BPS, we have got this model where human plus AI is working together, and there are tasks which are autonomous. How do you orchestrate work? That's where we are building orchestration layer, where the orchestration layer really knows what task is being done by the human plus what tasks are being done by the AI, so that oversight is coming in that layer. But early days.
You did say that you want to maintain the 26% to 28% band. However, if you look at the last 7-8 years, it's been tough to sort of stay there. Given a lot of the investments, you will be doing going forward, doesn't it make sense to maybe change the band to around where we are right now, giving you a lot more operational flexibility and be more competitive in the market?
See, overall, what we believe is, with our cost structures, we should be able to operate in the 26% to 28% band. We believe that the early investments which we have been making has been the source of how we are able to maintain sustained industry-leading margin band. And we are taking the challenge that considering all the investments, we will incrementally make, we would want to shift towards 26% - 28%. To the point in terms of the last couple of years, you would also see that there would be various points of times where we would have come closer to 26%, and then we would have made some investments. So, it won't be that we will be stuck on not making the required investments at the cost of profitability. Growth with profitability will remain our mantra, and we will not be shy of investments, at the same time we will be driving margins towards the 26% - 28%.
First question is, advisory/consulting focus. That's an area where, let's say, TCS may have not prioritized this, at least explicitly, in the past. Now, as you drive down the advisory path, both through organically as well as through acquisitions, what are the challenges in execution that you see? And more importantly, what has changed for you to prioritize the advisory/consulting part so much?
So, if you see the overall tech and advisory space, I think most of the incumbent firms are also going, or rather experiencing a lot of change, both in terms of what customers expect them to deliver, as well as the way they need to deliver it. I feel this is the right window for us to look at a more updated model of advisory or consulting, which will essentially be more driven by AI and analytics, and coming from our place of strength, which is technology. Given some of the stack conversations we have had, I don't think success can be just about providing advisory, it is also about going through all the other layers of the stack to actually help the customer deliver it. But we'll have to build that initial muscle of advisory as well, because that's where the impact, or the conversations get pegged at the right level for us to drive that impact across the layers. So, we see that as an opportunity, right opportunity for us to consider building that capability.
The second question that I had is for Samir. How do you define that US$ 1 billion of investment? Is it annual? Is it to the P&L? Is it OpEx, CapEx? If we can just get some more color on it.
This is one billion dollars in annual spend, completely on OpEx, completely on the P&L. The key components are on the learning and development part, learning and talent-related, industry-specific or new services-specific R&I, and specialized infrastructure.
Two questions. First, in any technology transformation cycle, we have seen first-mover advantage along with scale. As TCS looks at data centres, and TCS has always enjoyed the scale advantage, do you think scale will be essential, along with being a first mover?
Scale definitely plays a role. One of TCS's advantages, apart from delivery execution and client trust - multi-decadal trust; our scale and breadth and depth will play a positive advantage.
So, we will put capital to use to achieve that scale advantage as well?
When we saw the full stack and what we talked about, the investments, we will be leveraging and making investments where we see focused returns coming in. Yes, absolutely, we will be using our capital. But we will also be partnering. As we saw, if you take the data centre case, the investments would be prudent and it would be structured. So, the balance sheet does play a role.
My first question is perhaps to Samir or Mangesh, which is about your India AI DC venture, which is going to be completely opposite to the asset-light DNA that TCS has maintained over the past 20-25 years. Are you not thinking about coming up with a special purpose vehicle to do it, or is it going to be done under the TCS fold? But the key question over here is where is the synergy? That's what I'm getting at. Where is the synergy with TCS's overall services business in the global scheme of things?
So, to answer your first question, there is a special purpose vehicle. We'll be making the investments with locking in an anchor customer. And as I talked about the structured investment, we will be leveraging the balance sheet. We have also announced an equity partner i.e TPG. We have announced creation of a subsidiary HyperVault which is formed as a special purpose vehicle to invest in AI DC. In terms of synergy, we benefit from both front-end synergies. As I said, the anchor customer typically would be a hyperscaler or an AI-led company, and it is not just about the business. It's a 360-degree relationship which we will get from them. Also, the back-end synergies, where we can leverage, because the data centre is all about cooling, power, and connectivity. With the #OneTata advantage, we can also look at getting optimized value from them or any other players which would be available.
First one is, last year I think for the first time you actually showcased COBOL to Java application modernization and things like that. It's been a year now. There should have been a lot of learnings on nuances. Just trying to understand, that's a very large market, obviously, and how much adoption are you seeing on this, and how rapid is it? Do you think you'll see a lot of those next year? And, just to get a feel of these things, you highlighted a project with 50 million lines of code being converted. Typically, as an example, how large are these projects, and how long does it take?
Yeah, I'll take the first question. On the modernization front, there is immense demand. In the past, there has not been a proper business case for technology modernization. Particularly when you want to move from a technology like mainframe into any new modern technology, there was no financial business case. Even from a risk perspective, not many people knew what was the business logic that was used in those old systems, and to reliably translate them into a modern technology has always been a challenge. But what Generative AI does is, gives you the ability to understand, and a human can validate the understanding, so you can encapsulate the knowledge that's residing. It also provides a framework through which you can modernize to a new technology. The example that you talked about is progressing quite well. Again, these are also very complex programs. It's not that you turn on a switch and 50 million lines of code convert themselves into Java. It requires an amount of human intervention and validation. But of course, that's one large example. What you see more and more common are what Aarthi calls rapid builds, we showed it as CodePlus as a platform here, where we have the ability to move from X to Y, like from Tableau to Microsoft platform, or from a TIBCO code to Java. There are multiple opportunities that exist, and these are all much simpler because the codes are more recently written within the last 10 to 12 years, and not very large or complex like the 50 million lines of code. Such programs are very time-bound. Outcome could be measured, and results and value can be achieved in a short time. Mostly, we try to do it in a three-month period.
The second one is, earlier today you showed us how we could rapidly create a use case, and how it's very interesting for organizations. I understand that very well, because CEOs would love that because it creates momentum in the organization in terms of GenAI creation. But what I'd like to understand is, there's a lot of talk around creation of an AI fabric or a foundation layer. What does that look like? My understanding is something, and I'd like you to ratify whether right or wrong, and what it means. My understanding is that you have all the APIs, all the knowledge, all the context in a layer. Does it mean that if I'm able to create that use case rapid prototype with just the front end, you can agentically stitch together something if there is maturity on that fabric layer? If that is the case, then from a TCS perspective, where we have always been master systems integrators, what happens to business in general, and how long does this transition take?
If I may just add, if you look at the technology debt across the organization, it's in core systems, it's in integration, it's in data, it's in reporting; so across layers. The CodePlus platform that Krithi talked about, and some of it you'll see when you go to our Executive Briefing Center. So, what it does is, it actually accelerates X to Y migration, as I like to call it, across these layers, like Krithi gave some examples. So, I think what we're also seeing is, the same modernization programs that we did last year, this year we are able to do with much more productivity. One, technology capability has advanced. Plus, our own understanding of work with AI and what the human in the loop needs to do, has also advanced. In terms of projects, we are seeing the mainframe, the 50 million lines of code, those are all big projects. But there are also a number of short-cycle projects that we are seeing. Today, we are seeing a lot of that in BFSI, and other industry verticals are now starting to catch up. But I would say that we're seeing a lot more in the BFSI space. Now, coming to your second question, I would like to answer it more simply from an SI (systems integrator) perspective; the role that we play. I think I sort of briefly covered it. If you really look at it in any enterprise, while we can do the rapid builds, build the AI apps, agents, right, fairly quickly, where the work goes is in the plumbing; connecting it back to the data, connecting it back to the systems. And more importantly, if you look at the Infrastructure-to-Intelligence (I2I) layer, across each layer there is a significant amount of work to be done to integrate. And today when we move... the way we're building apps today, we're connecting it to the model and to the data. But going forward, to Krithi's point on building learning systems, the more we start creating these learning systems, creating the decision infrastructure, that is where a lot of work across all layers of the I2I framework we showed you. When it comes to agents, not just individual agent building, but agent-to-agent interaction, agent interaction which is the fabric layer that you alluded to, while you can build many things quickly, but stitching things together and bringing the contextual knowledge of the customer is where we come in.