- 9 hours ago
At the India Today AI Summit 2026, Microsoft India and South Asia President outlined how self-reliant AI can transform life, work and governance across the country.
Category
🗞
NewsTranscript
00:00Welcome once again to the India Today AI Summit and we are delighted to be joined by someone
00:08who's part in and part of this global shift that's taking place, Puneet Chando, President
00:14of Microsoft India and South Asia. Welcome. Thank you. And we've framed today's debate
00:22in a way or discussion as life, work, home, creating an Atmanirbhar AI, a self-reliant
00:32AI. Is that an oxymoron? Given the fact that AI in itself means you're completely dependent
00:40on your chat GPT. How do you create an Atmanirbhar world, Puneet, when AI by itself is effectively
00:48reducing us almost to slaves of chat GPT or any AI innovation. So I want to start with
00:55that. Can you really create a self-reliant AI? First of all, Rajee, thank you for having
01:01me. I don't think it's an oxymoron and I don't think it's turning us into slaves. I think
01:06it's amplifying us and hopefully it'll get us to do a lot more value-adding, interesting
01:10work in life. I think back to Atmanirbhar India, Atmanirbhar AI, I think that does not mean
01:16that we should build AI in an independent siloed way by ignoring the world. I think
01:19it means two or three things. One, invest in infrastructure, which India is doing. And
01:23as Microsoft, we're pulling in $17.5 billion. Second is making sure we've got sovereign AI
01:28solutions for India, be it public cloud, be it private cloud, be it co-pilots, but just
01:32making sure that we're meeting the demands of India where India needs us to. And then
01:37AI eventually is a five-layered cake, right? You've got energy at the bottom, you've got tokens,
01:41you've got models, you've got applications, you've got chips. I think India should take
01:45a very, very thoughtful stance on out of the five-layered cake, what are the areas where
01:49we should really build ourselves and what are the areas where we should get the best
01:53from the world? This is the time to be truly frontier. Every week something else is dropping
01:57and this is the time to both build and get the best from the world.
02:00Okay, so let me take off from what you said. The Prime Minister ahead of the AI summit seemed
02:07to suggest that there were three pillars to India's AI push. Innovation, inclusivity and
02:16sovereignty. Let's break each of those down. Innovation. The truth of the matter is the United
02:22States and China are way ahead of the rest of the world when it's come to innovation. They built the
02:27foundation models. Some statistics suggest that 80% of the investment in AI infrastructure comes just
02:34from these two countries at the moment. Are we therefore playing catch up when it comes to
02:39innovation or do we in a way simply become the back room for what China and the United States
02:47are doing and tag along with them?
02:50So I think the Prime Minister obviously said this very well, innovation, inclusion and
02:53sovereignty. I would add a couple of things to this. One is diffusion. And Jeffrey Ding had this
02:59fascinating book and he looked at countries that are innovating and countries that are diffusing
03:02technology and he said that countries that are diffusing technology are getting more value out
03:06of it. So while the world is innovating, India is starting to innovate. There are a lot of Indic LLMs
03:11and a lot of launches happening today and this week. I think India should equally focus on diffusing
03:15AI from classroom to boardrooms, from agriculture to healthcare to enterprises to public sector
03:20and that's happening. So I think that's one thing we should push on. The second is also skilling and
03:25going back to innovation, inclusion, sovereignty. If you want a billion Indians to rise with AI,
03:30a billion Indians need to know how to use AI. And I think that's where we should focus a lot
03:34more. So agree with those three. I would just add diffusion and skilling on top of that.
03:39But surely you need to invest far more in innovation than we are doing. Most Indian companies
03:45traditionally haven't invested enough in R&D and thereby have perhaps lost out at times
03:52in a world that's rapidly changing. Do you believe that needs to change as a first brick in a way
03:59to
04:00build an Atmanirbhar AI?
04:01No, you're right. It does need to change. As public, as private sector, as enterprises in India,
04:06as multinationals, I think we have to add value in the country and none of what we have has any
04:12franchise value unless we're not adding value to the country, right? So we have to invest in R&D,
04:16we have to invest in infrastructure, we have to invest in skilling. That's why we're skilling 20
04:20million people in India by 2030. So you're right, all of that needs to happen. But again,
04:24I'll just say, listen, while we innovate and push on innovation, don't forget diffusion.
04:28Because the more we can diffuse this technology, get it into the hands of every Indian, every company,
04:33every public sector institution, I think there's a lot more value there that we should not miss.
04:38Which comes to the second eye, which is inclusivity. How do you, in a country that is so deeply
04:44divided, with such sharp inequalities, make AI truly inclusive? You said AI for a billion Indians.
04:51How do you really make it? It sounds great. How do you actually make sure that a little kid in
04:58Dhar in Madhya Pradesh is getting the same AI technology benefits that, let's say, a child in
05:05South Delhi or South Mumbai is getting? How do you bridge that divide? Or will we have an AI divide
05:10like we've had a digital divide for a very long time?
05:13So Rajiv, I genuinely believe that AI is different from any other technology we've seen.
05:17Unlike the Industrial Revolution, where we waited for many, many years to get that
05:21technology, AI is here on day one, right? So we've got, like anybody sitting in Boston today,
05:26or Bangalore, or in a small town in India, he or she has access to the same models, same
05:30applications. So I think that's one thing, which is we don't have to wait many, many years.
05:33And then to your point around inclusion, it's already starting to happen. I'll give you two
05:36examples. With the Ministry of Labour, we're building something called ISHRAM. There are
05:40300 million unorganized workers in India who don't know how to create a resume, who don't
05:44know how to take an interview, who don't know how to apply for a job. We're creating an AI
05:48platform for those people, these unorganized workers, to create a CV, do a mock interview,
05:53get ready for a job, and then get them employment. Just a terrific example of embedding AI
05:57into systems. And again, AI, unlike other technologies, does not require you to be a
06:03PhD. Like I'm a rusty coder. I used to code 30 years ago with AI. I'm coding again. English
06:07is becoming the coding language now. So I think the ability to get AI into the hands of Indians
06:12is much easier than previous technologies. We do have to get our act together. I'm just
06:16giving you one of examples. This needs to happen at a scale for a billion Indians to get
06:20there, but it's possible.
06:21Which brings me to sovereignty. It's a word which is being used by many. We need to ensure
06:27AI sovereignty or certain digital sovereignty. How do you actually achieve that? Because
06:34increasingly you have to work with big tech global companies. AI by its very nature cuts
06:40across national boundaries. How do I control what an engineer sitting in San Francisco is
06:46doing, the information he or she is feeding me, or the AI models that they are creating.
06:51Do you believe it's possible to have AI sovereignty by simply creating large language,
06:58homegrown models? Or is there another way out? How do you create genuine sovereignty over data?
07:05So, Rajiv, there are two ways to frame sovereignty. You can use it as a code word to slow down,
07:10or you can use it as a code word to speed up. And I would obviously say, listen, we got
07:14to speed
07:14up on sovereignty. And there are two things we should do, right? One, we should build in India,
07:18and we're already building in India. So, for example, our co-pilots, which is, I was talking
07:22to Mr. Puri about it, the AI agent, all the processing is happening in India. So, data
07:26privacy, sovereignty, everything that happens in India stays in India. The model stays in
07:30India. So, one is build in India, meeting the sovereignty requirements, but also get
07:35the best from the world. Going back to, like, my biggest challenge is every day something
07:40is changing on the models. And if you're not keeping up with the frontier that's happening
07:44in the world, we will fall behind. So, I would say build for India, and there's a lot of
07:47build happening, but also get the frontier from the world. This is the worst time to
07:52slow down and use our sovereignty as a way to slow down. In fact, use sovereignty to speed
07:55up.
07:56No. So, are you saying you build partnerships with the rest of the world? A Microsoft,
08:00for example, is a good example. You're transnational. Do you sort of, does a Sarvam AI, which is
08:06seen as the government's, in a way, answer to creating a foundational model for India? Do you
08:13collaborate with a Microsoft, which is a global corporation, and at the same time, believe
08:18the data is Indian?
08:20The answer is yes. And by the way, I'd like to believe we're an Indian company. We've
08:23been here for 35 years. We're as Indian as anybody else. Sarvam is a deep partner of ours,
08:27by the way. And the other thing on Rajdeep on this one, right, if the technology is changing
08:31every day, none of us have all the answers, be it Microsoft, be it Sarvam, be it any other
08:35multinational, be it an Indian company, be it the Indian government. So, the idea is how
08:39do we all come together and build on what others are building versus trying to create
08:43walls, right? So, this is the time to bring everybody together.
08:45The reason I'm asking this, if you saw in the union budget 2026-27, a 21-year-old tax
08:51holiday was offered to foreign companies, which are going to use cloud services, come
08:56to India, provided they partner with Indian-owned data centers in India. Now, are we simply,
09:03therefore, going to drive this sovereignty by creating large data centers, which the Adanis
09:09and the JOs are going to put a lot of money in, billions of dollars? Is that going to build
09:13data sovereignty in the truest sense of the word? That foreign company is coming here,
09:18it's getting a tax holiday. Is it going to retain the data or will that data be shared
09:22with the people of India? Do you believe that that's the way forward, those kind of partnerships,
09:29Indian-owned data centers working with foreign global tech companies?
09:33So, the budget announcement on data centers is a terrific word of confidence for building
09:37infrastructure in India. And as I said, we're bringing in 17.5 billion dollars over the next
09:41five years. The question is, can we even do that faster, bring more capital into it, build
09:45more infrastructure, partner with Indian companies to your point? And I think the point here is
09:50not just saying, listen, there are some companies who will build this with us. The idea is to
09:54build AI infrastructure for every Indian, right? Going back to saying, listen, can we get every
09:58Indian access to AI on their mobile phones? That requires infrastructure and capacity,
10:02which is what we're building. And then can we serve the world out of India, right? So,
10:06obviously, Indian data will stay in India, but can we serve Middle East countries? Can we
10:09serve Africa? Can we serve some of the South Asian countries out of India, given the infrastructure
10:13that we're building? And that's a massive opportunity.
10:16But to build an Atmanirbhar Bharat as it is, the key lies with small businesses, MSMEs, micro
10:24industries. The question is, how do they benefit? How does a small business benefit from AI
10:31without the high costs involved? Do you believe AI will make it actually easier or more affordable
10:38in a way for smaller businesses?
10:41I think there's some work to be done. But if you think about the thesis of AI, Rajdeep,
10:45what are we trying to do, right? For the first time in the history of mankind, we are saying
10:48we have the ability to create an intelligence, right? Can we create, can we use compute and compute
10:53to become cognition? Can we create intelligence out of machines, right? That's what we're trying
10:57to do. And by the way, every business today is a bundle of intelligence, bundle of expertise,
11:01be it a small or medium business or be it a large business. And the idea is, can we get
11:05intelligence on tap? Today, if you're a small and medium business in India, you want to solve
11:09a problem, what do you do? You hire people, you bring a team together. We're not talking
11:13about a possibility where a small or medium business in India can get intelligence and expertise
11:17on tap, right? Can they really get that intelligence to solve the problems or build the business
11:22they want to? So that's the thesis of AI. Now, the question is, how do we make sure we
11:25get it to the hands of SMBs at a price point that they can afford in a simple way? They
11:30don't have technology, they don't have people, they don't have the expertise. How do we simplify
11:34this and get this to work for them, which is where Rajdeep, a lot more work needs to be
11:38done, but we're committed to doing it.
11:40Is there an example that you can give, a successful example where, for example, you manage to work
11:45with small businesses or a sector to make it more effective, affordable, and successful?
11:52Lots of these examples, and I don't know if it's a small or medium business, there's a
11:55company called Creative Arts Network. They're a small company, they're building content for
12:00Bollywood and TV. We're working with them to use AI to build content and, again, reduce
12:05the cost of building by 100x, right? And now for an SMB, which is a small business, if you
12:10can potentially reduce the cost of building content by 100x, imagine the value that can create
12:15and the more value that they can create, the more content they can get out and the more business
12:19they can generate. So there's just one example, but across industries, healthcare, we're building
12:23co-pilots for doctors. Can we get doctors, like my most interesting story is you go to
12:28any hospital in India, the senior most doctors are out of time, the junior most doctors are
12:32out of work. The junior most doctors are writing notes, right? Can we create a co-pilot so the
12:37junior doctor can become as proficient as a senior doctor, right? Can we really give them
12:41the expertise, judgment, intelligence, so that they can treat patients and patients don't
12:45have to wait. Lawyers, same story. Can we create a co-pilot for
12:49lawyers? So I could go on, but it's across the board, Rajdeepin.
12:53Because one of the challenges, and this is a question I'm going to ask all my guests on
12:57this, is we have a huge demographic dividend, but we also have a large population which is
13:03unskilled workers. And the fear remains that as AI spreads, unemployment of those at the bottom
13:11of the pyramid will be real. Jobs will go. Is that fear mongering? Or is that a reality
13:19that we've got to be mindful while trying to create an Atmanirbhar AI architecture?
13:24So a lot of this is exaggeration, but I think you also have to then kind of just be clear
13:28-eyed
13:28about this, right? You can't live in a fantasy world. The clear-eyed answer, Rajdeep, is the
13:34color of jobs will change. Your job and my job both will change. And ultimately, your
13:39job is a bundle of tasks. For some reason, we've equated work with jobs and titles, but
13:44the atomic unit of work is not jobs and titles. It's actually the tasks that you do. And what
13:49AI will do, Rajdeep, will unbundle everything you do and unbundle everything I do. And a
13:53lot of things that you and I think are work, for example, me sending 50 emails or attending
13:57seven meetings or updating three trackers, feels like work, but it's not really work.
14:00We know it at the end of the day. I think AI will take that away, which means you and
14:05I will have more time to do value-added work, but we also have to really bundle ourselves
14:09much better, right? Which is we have to upscale ourselves so that we can really do stuff that
14:13is truly value-added. I think the implication for India and for all of us is that we're
14:17all living with this industrial age era template, right, where you learn once and you earn for
14:22the rest of your life. My grandfather lived like that. My father lived like that. I'm
14:25almost living like that. That will break. We are the last generation to have long-term
14:30careers. My kids will have a portfolio of things they'll have to do and they'll have
14:34to focus on skills, not jobs and titles. And I can imagine India in the next few years
14:38where there's a certain set of tasks that need to be done. People have skills and there's
14:42intelligent matching of those two happening. That's the world we have to prepare for.
14:45That's fascinating. So what you're saying, the real challenge in a way is skilling. Rather
14:50than talk about jobs, jobs will come provided you get the right skills for those jobs. Am
14:56I broadly correct?
14:57You're broadly correct.
14:58Okay. You know, since you just mentioned the possibility that we may lose our jobs or
15:03our jobs will be redefined in the coming years, we've now got AI anchors. So I want you to,
15:09I have a question from you from my AI anchor. Listen in.
15:14You do. Fantastic.
15:17Hello, Mr. Chandok. My question to you is, what are the big advantages of AI and what is the one
15:23big
15:23concern?
15:24By the way, that's so fascinating. I should get my AI to talk to your AI. Hopefully in the next
15:28few
15:28months we'll get there. I think the biggest advantage of AI, and that goes back to the thesis,
15:32right? It's a remarkable technology. For the first time, we have the ability to create intelligence.
15:36And intelligence is the most valuable commodity on the planet, right? Everything of value that
15:40we have created as a human species is driven by intelligence. And I'll tell you one quick story,
15:44right? The latest models that we have are equating to 150 IQ points, 150 IQ points, right? And let's
15:50assume the IQ of this room is roughly 100. Imagine a world where there's 50 IQ points available for you.
15:55There's a plug in the wall. You can plug in and get 50 IQ points. That's the power of this
15:59technology,
16:00right? Now the question is, how do we get this into the hands of every SMB, as you said, every
16:03company,
16:04every public sector, every Indian? So that's the advantage. The thing that I worry most about,
16:10Rajdeep, you spoke about this, which is not so much jobs going away, but our ability to scale at
16:14scale in India. We have to scale a billion Indians. As Microsoft, we are committed to scaling 20 million
16:20people over the next few years. I don't think that's enough. We have to do more. We all have to
16:24do more.
16:24But for a billion Indians to rise with AI, we need to scale them. They need to know how to
16:29use AI.
16:29AI. People will not lose jobs to AI. People will lose jobs to people who know AI. And you better
16:34be on the right side of that equation. So that's one thing we all have to solve for.
16:38Because we are already seeing tech companies who are re-looking at their models. There have
16:44been layoffs in some of India's major tech companies as well. Is that again something we
16:48are going to have to learn to live with for a while, that there will be disruption? AI is the
16:52big disruptor in our lives. Going ahead, do you think eventually companies will recognize that
17:00you can take full advantage of AI without necessarily shedding a workforce?
17:10The answer is yes, if you can scale your workforce, right? And the one thing I tell people around me
17:17is saying, if you stop clinging on to your job description, doing what you were hired to do
17:23is no longer enough. If your only value add is I can do my work faster, I can do more
17:28volume,
17:28I can do more efficiency, AI can compete with it. So stop clinging on to the job description.
17:33You really have to, and again, when you think about AI, skilling is misunderstood. I would say
17:37three things. One, you have to learn to think with AI. Every time you're doing something, you have
17:41to get AI to work with you as a thought partner. Second, you have to build with AI. And I
17:46said,
17:46I'm a rusty coder, but I'm coding every day again, because English is a new coding language.
17:50Like I say, right, you can never get fit by watching others go to the gym. You have to go
17:53to the gym yourself. With AI, you have to go to the gym yourself every day, every day. And then
17:58finally, get AI to work with you like a teammate. The real unlock for us is when you think,
18:02when AI moves from a tool on your phone to a true teammate, right? Deep, somebody who's sitting with you
18:06and helping you do your work better versus just a thing on your device. So if you do those three
18:11things, think with AI, build with AI every day, and then get AI to work with you as a digital
18:15colleague, I think we will solve this. Because, you know, I asked you at the outset,
18:19the fear that you will eventually become a slave to AI. You're saying you don't have to become a
18:25slave to AI. You've got to harness the power of AI to simply become better at your job.
18:31AI is a cognitive amplifier, right? Which is, imagine the power of this, right? All of you sitting in the
18:35room, on your phone today, have a PhD level researcher, a PhD level, like somebody who's got
18:41medical knowledge and legal knowledge and science and everything, right? That's sitting in your
18:45pocket today. The question is, how are you using it, right? Some of us are using this flying car to
18:50buy groceries. Literally, you have a flying car in your phone, the way these models are developing,
18:54right? So I think if you choose, and it's a big mindset shift for all of us, right? If you
18:59choose
18:59today, you can be more intelligent, you can be more efficient, more productive than anybody else on
19:05earth, if you really use this technology to your benefit, right? So get it to you. Again, if you're
19:09still doing what you were hired to do, in whatever role you're in, you're missing the bus. This is
19:14guerrilla warfare against irrelevance every day. If you're not learning and building with AI every day,
19:19you will be irrelevant, right? So that's my one ask of all of you.
19:22One final question. In a country as vast as ours, with so many different languages,
19:26more and more AI models are trying to adapt to this diversity, build language models. Do you
19:33believe that's the way forward, given that AI is the ultimate globalization weapon? Or do you
19:39believe you can indigenize AI to fit the unique diversity of this country through these large
19:45language models?
19:46The answer is the days of a company or an individual betting their farm on one model
19:51are over, right? If you're following this world every week, something else is dropping and the
19:55models change. I think the answer here is model diversity, be it enterprises, be it public sector,
20:00be it individuals like you and I, we will all want a series of models to work with us.
20:03And that does a few things, right? One, it allows horses for courses, right? Which is,
20:07I don't need a trillion dollar parameter model to do a simple use case in my organization,
20:12I can use something cheaper, I can use something which is more made for purpose for that one,
20:15right? So that's one. Second, AI still is jagged intelligence, right? Sometimes it's super
20:20brilliant, sometimes it's not that great. But if you get multiple models to work with you,
20:23which is how I work with AI, I've got three models debating each other and getting me to
20:27the right answer, we will solve this jagged intelligence problem as well.
20:30One final question in one sentence, what is the most competitive advantage that India has
20:36in this entire AI revolution, as we're calling it? What is the one thing that we have that others
20:42don't? I think India has got the talent. We've got 22 million developers today on GitHub,
20:46the largest developer population in the world. We've got digital public infrastructure that we
20:50build and Mr. Kant is here, right? He's obviously much smarter on this than I am. Like, if you look
20:54at the digital public infrastructure in India, if you can embed AI into all of that, be it Adhaar,
20:58be it UPI and get our developers, we've got the massive, the most amount of developers on the
21:03planet to really build AI public infrastructure in this country. I think that's something nobody else in
21:08the world has. Well, more power to that, more power in a sense to this large demographic
21:14dividend that we have, which hopefully will last at least the next 20 years and harnessing that,
21:20skilling that on scale will be the big challenge in a way to build a Atman Nirbhar AI, as we
21:27are calling
21:28it. Please give a big hand to Puneet Chandogh. I think that was a fantastic exposition of where we
21:34stand when it comes to our AI opportunity. Thank you very much, Puneet.
Comments