At a moment when India’s AI ambitions are under global scrutiny, Vivek Raghavan, co-founder of Sarvam, is out to prove that world-class, hard technology can be built from India.
If We’re Not Ambitious, We’ll Only Build Small Things: Sarvam's Vivek Raghavan
Sarvam cofounder on made-in-India AI glasses, deep-tech risk-taking, and why India must prove it can build world-class models
Pratyush Kumar (left) and Vivek Raghavan
Advertisement
Speaking to Outlook Business on the sidelines of the India AI Summit, where the AI startup launched a frontier large language model with 105 billion parameters, Raghavan reflects on returning from Silicon Valley to build deep tech at home, and explains why success, for him, will only be real when every Indian can access services through simple AI-powered interfaces.
Edited Excepts:
What is the significance of the launches you have done at the AI Summit for Sarvam and for India’s AI ecosystem?
We want to be able to show that we can build world-class models from India. That is the goal. We want to inspire not just people at Sarvam, but the whole ecosystem that tomorrow many developers and many startups can also do things that show we can build from India. Most of the people involved in training the model are very young. So I think this is also a success for the youth of India. In the end, we have to have a belief that we can do anything. I hope this is a small contribution towards that.
Advertisement
Can you tell us about the glasses that PM Modi wore? What is the plan and when can people expect to use them?
We are planning to launch it by May this year. The goal of these glasses is that they have the capability of capturing images, video and audio. The intent is that they are fully made in India and will be completely connected to the Sarvam AI stack.
You are an entrepreneur who came back from Silicon Valley after building and selling startups there. Do you think the kind of work you are doing will attract more people to return to India and build?
That is the intent of what you do. Not just that I built after coming back from the Valley — for 15 years I actually worked in what we call today India’s DPI as a volunteer.
This time we decided that a company is the right way entrepreneurship is the right way to achieve the goal. In the end, achieving the goal is what is more important. We want to build a deep-tech company out of India.
How do you build a deep-tech company out of India? You have to show success. Based on that success, people will say that it can be done here too. That’s the point.
Advertisement
Both of you at Sarvam are PhDs. In India, we don’t see many deep-tech founders with PhDs, unlike Silicon Valley. Do you think that matters?
I don’t think there is anything special about a PhD, notwithstanding that I have one. Nowadays, we see that a lot of expertise is even there with high school kids. You just have to find the right people.
There are many people with PhDs whom I wouldn’t choose for anything. It’s the right person that matters. You have to find people who have that spark and that ambition to do big things.
If you are not ambitious as a country, then we will just do small things. Silicon Valley certainly changed my thinking that risk is something you should take. Technology risk is different from other kinds of risk.
This is just the beginning. We have not proven anything yet. But I hope we eventually reach a place where people routinely think that building hard technologies is par for the course in India. That is my hope and vision for the country.
Advertisement
You say you haven’t proven anything yet, but there is a lot of excitement around Sarvam. When will you personally feel satisfied? What is the milestone?
My milestone is very simple. It has been my milestone from the beginning. It is not a financial milestone.
When we had UPI in India, it became a system better than similar systems in many other parts of the world. But most other things in India are still hard to do.
If every citizen to the last citizen of India can get access to any service or product easily, through a simple interface, and actually get that service delivered, then I would say Sarvam is successful. If we can do that as a country, then Sarvam is successful.
When do you think AGI is coming?
Not relevant. It doesn’t matter if it’s five years, three years or ten years. It doesn’t matter.
Elon Musk recently said that beyond the Turing Test, we should think about a “Galileo test” where AI can say it is right even if everyone says it is wrong. Do you think AI could reach something like that in the next 10 or 20 years?
Advertisement
We are at the very beginning of this revolution. We haven’t seen anything yet. From a global perspective, big things are going to happen. Change will come at a pace that we don’t even understand right now. It is exciting and scary at the same time.