People are building in India, for India. The job creation opportunity, the ability to both upskill and do (ambitious) things is great (here). At the same time, the United States will make its decisions in the interest of American competitivenes, (which) has always had an element of talent coming in, and that will continue. It will also have certain reforms in some policies and visa regimes.But at the end of the day, America will decide what is in its best interests and its competitiveness. India has a great market internally. It will also participate in the global market because of the talent (it) has. There is no denying that anyone who does not tap into India’s human capital is making a choice (to) not be competitive, in some sense. So therefore, all of this will have to be done one country at a time, where they make their decisions that are in their interest.
Today’s AI revolution is said to be bigger than the industrial revolution. How do you place in context the changes that you are seeing?
The right frame is to view it as the next platform shift. There is a fantastic piece of work done by an economist called Diego Comin, where he compared the diffusion of, for example, spindles. It took something like 120 years before they diffused in the Global South, as we call it today. Whereas here, for the first time, you have the ability to take something that may be like steam, like electricity, but diffuse it fast across the entire globe. And so that is something that has never happened in 5,000 years of human history. I have an equation where I call tokens per dollar per watt-whichever community, country, company gets to exercise that well is going to really benefit. It is not just talking about it or celebrating technology for technology’s sake. It is a broad skill that is required in order to be able to use it in every sector of the economy-healthcare, education, manufacturing, in frontline work, in knowledge work. And that, I think, is the key difference.
On the geopolitical front, competition in AI is rising, China is lowering the cost (for large language models). Is American supremacy (in AI) being challenged?
Discover the stories of your interest
The United States cannot take anything for granted. We need to compete. We need to innovate, to compete. For example, I think about what we were able to do even with some of the AI models, and the point about making them cost competitive -we have reduced the prices of GPT models by about two hundred times. This is hyper Moore’s Law in action. It is fine for us to see countries like China or, quite frankly, even in India, there will be people who will develop models, and that is healthy. We have to continue to push both on the capability side and on the cost side, so that we can meet the diverse needs of technology around the world and do so with safety and trust.
So, would it be fair to say that we will see a lot of disruption in the Gen AI sector this year? Also, in certain quarters, there is talk that Microsoft is overly weighted towards OpenAI, and now there is this conflict with (Elon) Musk challenging (OpenAI’s) change to a for-profit structure.
Disruption in our industry is a given. The question is, how do you innovate? I always say, what happened last year is what happened last year, what happens this year is what we stay focused on. Then, you look forward to making sure you are in a position to innovate next year. For us, we are very thrilled about our partnership with OpenAI. We are very thrilled about the work we are doing on our own, and the choices customers have, partners have when it comes to Azure in terms of open-source models, great models from Open AI, models that we have built, like Phi. We will continue to innovate on all layers. A year from now, I do not think the conversation will be about just models. I think it will move to all the other considerations. It will be about what is the UI (user interface) layer with Copilot, about which we are excited. What is the layer around data? In an agentic world, they are all stateful AI applications which require data, they require a real AI app server. So, between foundry, fabric, and Copilot, we feel very well positioned.
Recently, the government cautioned that India needs to be careful about how it adopts AI given unemployment is a huge problem here. What is your perspective?
Absolutely. At the end of the day, one has to think about the return to labour in terms of total wages. You really want to make sure, frankly, that there is more opportunity. And also, regulate what jobs get what wages in a world where there is abundance of some new technology. These are all things that have to be carefully thought through by policy makers and in democracies. That said, sometimes you overstate the displacement too. Let us take call centre employees. In the beginning of the 2000s, there were around four million call centre employees, and now there are seventeen million, and this is after having introduced so many new technologies-IVRs, multiple generations of them, and then we even introduced self-service bots. Because, with abundance comes more usage, new opportunities. Who would have thought that by putting typewriters in front of a billion people, we would have this explosion of jobs called knowledge work. That is what the PC revolution did. When I see something like GitHub Spark, I kind of look at what is going to happen in software engineering and say, wow, we may have a billion software developers, not one hundred million.
Microsoft is investing $3 billion over the next two years in India, a small fraction of the $80 billion being spent globally.
This is just the next tranche of our investment and will continue with the density of usage of AI. Our investment is not (made) in isolation. If I think about it, our investment in our data centres is the best stimulus for India’s energy sector. We are making an upstream investment, it will stimulate downstream usage, and then the economic benefits (will be) much more broadly spread.
Last year, you had said that AI will play a huge role in India moving faster towards its goal of a $5 trillion economy. One year has passed, a lot has changed very quickly in the AI world, in the tech world. What are your thoughts?
That is correct. There is a virtuous cycle between some of the programmes and the (government projects) yojanas, the India stack, the digital public goods that are unique in India. But it is that juxtaposition of the programmes and the tech stack. What has happened with UPI-it is just miraculous. And then you couple that with the entrepreneurial energy in the country, plus the demographics. That virtuous cycle is what I think can drive broad-spread economic growth that is inclusive as well, where healthcare is better, education is better, industry is able to benefit-small and large business.
