Hello and welcome to the CNBC TV8 special. We're coming to you live from Davos 2026. I'm Shireen Bhan and it is my great pleasure to welcome the IT minister Ashwini Vaishnaw who is leading the delegation here in Davos to join us for the CNBC TV18 Davos dialogue where we intend to put the spotlight and the focus on India's AI ambition, its AI aspiration, which will be the layer that India will focus on. What will be the strategy going forward? This is also going to be a curtain raiser to the big India AI impact summit that's scheduled to take place in New Delhi in February. To talk to us about all of that is of course the minister, but we have an eminent set of panelists here with us representing the best of Indian tech, and we of course have a room full of business leaders global and Indian to help chart out India's AI agenda. This is going to be a brainstorm, a big brainstorm at that, a participative consultation here live in Davos to understand the next steps that India must take as it charts out its AI journey. Without further ado, Minister Vaishnaw, let me start by asking you: the India AI mission has an allocation of a little over 10,000 crore rupees. Separately, you of course have the semiconductor mission. Both of them running in parallel. As you look at the AI tier cake today, where do you believe India is best poised, best positioned, and what will be the next steps in this journey?
The AI stack basically has five layers: the application layer, which is where the AI would be used; second is the models layer; third is the semiconductor, the chip layer; fourth is the infrastructure, data centers; and fifth is the energy layer. We have to play in all the five layers. Given the size of our economy, given the way our country and our people are so tech savvy, the way our people have adopted technology in the past, the way our IT services and all the three big leaders, top leaders from the IT services are present here. They have been providing IT services to the world. So I think we are at this point of time working on all the five layers. Of course, our very big focus will be on making sure that our IT services companies keep that momentum and they provide the AI-enabled services to the world, and we will also be having a bouquet of sovereign models which, as the world today fully understands, a 50 billion parameter model is good enough for doing 95% of the work in the entire economy. So we will be having a bouquet of models, about 12 of them, which are very focused, which are in the range of 50 to 120 billion parameters, which can run on a very small GPU cluster, maybe two GPUs or three GPU kind of cluster, so that we can provide the services and AI applications at a very low cost to a very large number of our people.
You know, Mr. Vaishnaw, let's address the issue of the AI infrastructure buildout. What we have seen at this point in time, it's largely been pushed by the US and China. Billions, hundreds of billions of dollars invested in building out AI infrastructure. As you see this opportunity opening up in India with data centers and so on and so forth, the role that the government will play, but more importantly the catalyst that the private sector can play. We have now seen large companies, multinationals like Google talk about $15 billion investments in India to set up data centers. You've got Kiti here, TCS and TPG tying up to do data centers as well. What is the plan as far as the AI infrastructure buildout is concerned?
$70 billion investment is already confirmed. It's actually getting rolled out, and we think that in the next 12 months there should be another minimum $50 billion investment, and most likely another anywhere in the range of 50 to $80 billion is likely to be added to the 70. So we should be looking at anywhere between 120 to $200 billion total overall investment in AI infrastructure.
As the infrastructure side is concerned, you talked about sovereign models and work is on there as well. Companies, private companies like Saram and of course the government is also aiding those developments. What's the update on that front?
The models have come out very well. Tested on so many different parameters, so many different metrics. The models have really come out very well and we have tested them in real cases, real life situations in the last few months, and very soon we should be in a position to launch the entire series of models. First and second, many of our startups are really innovating on the way the models will be created in future. In our understanding, and that is what most of the tech world leaders say, this is just the beginning of a multi-decade innovation journey, and the first few models, let's say even if you look at the top 10 models of the world today, they are all just a first set of models which have come out. I think one of the tech leaders said that the brain uses just three watts of power and a typical AI chat GPT kind of a large data center would be using thousands of megawatts of power. So that means there is a whole lot of innovation still to happen in the future. So many of our startups who are focusing on innovation on better engineering, how to create that next level of model, where is that breakthrough in the next model? That is what many of our startups are focusing on, which will bring more opportunities for the country.
So what can we expect by way of announcements before I get everybody else to chime in, minister? What can we expect by way of announcements at the India AI impact summit? This is a curtain raiser, so give us a sneak peek into what we can expect.
The AI impact summit has three major objectives. Objective number one is impact: how do we make sure that AI technologies get used? There is ROI, there is increase in efficiency, there is increase in productivity, there is multiplier to the economy using AI. That is one. Number two is how do we make sure that technology becomes accessible to India as well as to the entire global south. And we have created a model of common compute: 34,000 GPUs which are available as a common compute facility to the entire country. Can we have similar models for the world? The third is about how to keep AI safe. We have seen the harmful impacts of specially generative AI and how do we keep our societies safe from those impacts. So these are the three major outcomes expected from the AI impact summit. Of course, there are many global leaders who are joining us, there are tech leaders who are joining us. We will be seeing a lot of investment announcements also there, and we will also be seeing the rolling out of our models.
The rolling out of the models will happen at the impact summit.
Okay. The budget is around the corner. Higher allocation likely expected for the AI mission?
These are multi-phase programs. For example, semiconductor. This is the first version of semicon getting almost on the verge of completion. Now we definitely have to look at a 20-year journey where we build this industry in our country to a level which is our real position in the world. Similarly, AI is going to be a multi-phase journey where there will be a series of initiatives that we take in AI.
Okay, let me open it up now and Kiti, let me start by addressing the issue with you from the lens of what TCS is doing and you know you aspire to be the world's largest services AI services tech company, but from your own lens as well as the imperatives for India. What should the next chapter in our AI journey as a country look like?
See, if you look at from the traditional lens, what most of the Indian companies are good at, the traditional services companies, I agree with what the minister said. He talked about the five layer approach and in thesis we call it the infrastructure to intelligence. There are multiple layers from infrastructure, there is a model layer, there is application or the agent building layer, and the experience layer on top. And while many of the IT companies in India have not built their own models, we should not lose fact that the value actually comes in understanding the business context of our customers, ensuring the data for our customers works for them. And I think that's our core capability, and most of us are focused on that, and you see us delivering value from that.
So there's going to be no change? You're not going to go the LLM way?
I'm not saying there won't be change, but don't underestimate what we are doing. We somehow seem to underestimate or press aside saying 'you guys are not doing models, your legacy'. No, what we actually do is making a real for our customers. Model is one component, but making the models work and work for the customer is the most, that's where the value is.
Well, you know, speaking of value and to Kiti's point, there was a lot of gloom and doom around the fate of the Indian IT services sector and what AI could potentially do. But each one of you in your conversations with me have said you see this as a tailwind, but you also see this as a reset. Sal, talk me through what this reset is likely to mean for companies like yours, but more importantly for the sector and for India at large.
So, I think AI is genuinely something that is changing both our company but the industry and India. What we see are huge, large value creating pools of opportunity. There are six of them that we've identified. For example, there's a whole set of work on AI engineering. There's a whole set of work on software development and modernization. And those combined have a large value that all of us, all the industry for the benefit of India will enjoy. And those are growth vectors that we see coming in the coming years. Yes, there are other places where we see some shrinkage, but this growth we see over the next several years compensating for it. So there are practical examples as Kiti was saying, as the minister was describing, in terms of where we can help global companies. For example, if you're doing something on modernizing the infrastructure, those things were not affordable for companies in the past. Today with agents, they will take less time, less money, and those become much more affordable, and we are starting to work on those areas now.
Rishad, just to take that forward, because I think the question that's being asked today is what is it going to do not just for global growth, but what is it going to do for growth in India? What kind of tailwinds does this provide? And more importantly, the ROI question as well, because significant investments are being made. What is the realistic return on investment period that we're talking about?
So first of all, Shireen, I think the way you should think about the services industry: any technological change I believe is good for the industry because to make that technology relevant inside of large organizations that are complex and messy, that have legacy, that have data that doesn't speak to each other, that's not organized, that are buying companies that are spread out across the world, you need partners to make that happen, right. So the opportunity for both India as well as for companies in India is to really take this and make large organizations adopt it. To Kiti's point, to make it be adoptable, you have to move from these general purpose, highly plateauing LLMs to be highly contextual, to understand the context, to understand the domain, to understand the workflow, to understand the data, to bring it to life on scale. And I think that's a huge journey and an opportunity for companies like us, all the way from thinking about the AI advisory story to all the way to making it real, with a very strong DNA of having safety and trust and strong governance, fairness, lack of bias, etc., and change management very implicitly a part of the DNA of the process as opposed to an after-the-act element. I think India separately also has a huge opportunity as a country to demonstrate scaled adoption with speed, showing it in many many specific areas, right, areas of education, healthcare, governance. So there's a huge opportunity both to take the story, to take all the infrastructure investments that have happened, and to make it be consumed both with India as a story as a market, but also us taking it to our global customers. And so I'm quite excited as to how this journey unfolds over the next three to five years. It's also a question of how bumpy it can be, and maybe it can be a little bit bumpy, but there's a huge opportunity. The last thing I will mention is just the element of talent, right? We've been a talent destination to the world for the longest time, and I think there's a huge opportunity for us to continue being that talent destination with AI, and our investments both as companies but as a country will be incredibly important to keep that differentiation because there is no progress on AI without the right, relevant, scaled skills around AI.
Absolutely a very important point that you make there, but minister, let me address that issue that Rishad just spoke of. You know, as we saw with digital public infrastructure, with the buildout of Aadhaar and so on and so forth, the government didn't just create the enabling infrastructure, it also in a way created the demand. In AI as well, to Rishad's point about deployment at scale, can the government, will the government play the demand generator?
Actually it is doing, it's doing. And the thought processes like this, there will be certain use cases and applications of AI which will come from the enterprise, from the corporate sector, from the businesses. There of course there is a very clear business commercial model which the IT services companies will play. There will be a very large use case scenario where the individual, let's say the MSMEs, let's say the farmers, let's say the students, let's say a large section of the population which cannot see a clear commercial use case. That is where the government will have to step in, and that's what we are thinking. That's what we are working on right now. We'll be creating a set of about a series of use cases where the government will be funding the development of applications and solutions using the bouquet of models that we are creating, and in the back end we'll be creating a data center, a large infrastructure which basically will be serving those use cases. So that way it will be one large diffusion, large scale diffusion of AI technologies. To make sure that talent pipeline keeps coming, because once you start using AI, then only that talent becomes more relevant. Just studying it in a college is different from actually using and applying it, right. And third, create that large demand flywheel for AI.
So where, what will be the initial use cases? Have you identified where you intend to make that first start?
Hundreds of them. We are already working on them in weather forecasting, in case of agriculture, in case of healthcare. Healthcare is going to be a very, very big user of AI, and India has that opportunity to be the leader in predictive and preventive healthcare because most of the legacy systems, as Rishad said, the developed countries and the rich world have healthcare systems which have serious legacy problems which are very difficult to overcome. So we can be the ones who can provide, using our technology base, AI base and the semiconductor program and the large IT services, we can actually be the provider of a large preventive and predictive healthcare program to the world.
You know sir, we usually have a situation where industry asks the government what the government can do for industry. Let me flip that around and ask you: what is your ask? What is your expectation of industry as we make this journey?
So, we have been working very closely. In fact, I'm in one-on-one conversation with most of the industry leaders already, and that has been the practice in the past also. For example, our entire semiconductor program was basically designed by the industry partners. Same is the case of AI program. AI mission is all designed by the industry partners. The one ask which I've been requesting all the industry friends is: help us develop the course curriculum for AI world, this fifth industrial revolution, so that our students who come out of the colleges are well prepared for it. Like in semiconductor, like in 5G, we developed course curriculum, went to AICTE, used the industry knowledge for creating those courses, and went to AICTE and got those courses implemented. Same thing we would like to do in AI, and all the industry participants are working on it.
Well, you know we are in Davos, the theme of Davos is the spirit of dialogue, so in the spirit of dialogue I'm going to throw this open and get recommendations, suggestions from each one of our business leaders here on what India must focus on, the India imperatives to meet our AI ambition and our aspiration. I'm going to start by asking you. You've of course been a long believer in the India story, Cisco's been a committed investor in India. You know, what do you believe the next steps need to be? What do we play to ensure that we maximize our strengths?
Well, so firstly, we are just honored to make sure that we continue to partner with India. We've had a fantastic relationship over the course of the past few decades, and that continues to keep accelerating. I think the biggest reason for the success is the public private partnership that we have with the Indian government. You almost feel like you're working with a startup, and the agility with which the government works is truly remarkable. So thank you again minister for all the support that you and the others in the government have provided us. I would say that look to Cisco for providing technologies for the infrastructure buildout of the data centers: low latency, high performance, energy efficient networking that goes across your data centers, safety and security capabilities, and then also the data side of the house where we've actually trained these models largely based on publicly available data, but as agents start to come about, there will be machine data that's actually going to be the fastest amount of growth. 55% of the growth is going to be happening with machine data. So, I'm hoping that machine data getting correlated with the human generated data that we've trained the models on provides different insights for resilience as well. So, we're excited to work with everyone. I think all the gentlemen on stage here we have close partnerships with, and I'm pretty sure that it's only going to get closer and closer over time.
