In India, from the investment point of view, there is more focus on AI applications compared to the foundational models. Why is this the case?
There's a realisation across the world that it takes an immense amount of resources, compute, talent and time to create foundational models. Given the way the scale of AI development has happened, there's a choice to either concentrate efforts on the foundational models or to actually start counting on the application.
First, the history of the evolution of technology has always been the application of technology to solve real-world issues, rather than just showing that one has the swankiest piece of tech. I think the application is where the cash flows will end up actually coming from.
Second, India has a whole host of issues. The UPI platform and Aadhaar have shown that the moment you solve these civilisational issues, that becomes the nucleus around which a lot more technology and value ends up getting created.
Third, it plays into India's services strength, which has primarily been about not creating apps, but on application, allowing people to integrate these technologies into what they have.
There's a trend of Indian AI founders moving to the US for access to better capital, funding and talent. What’s your view?
Let me put it this way, a dollar from India and a dollar from the US is the same. So, a lot of people move there because there are a number of accelerators there (in the US) as of now, where they'll cut a $1.5-2 million cheque without asking too many questions. So, access to easy capital is more a noise than a signal. In India, the quality of AI founders and the density of AI talent is still actually large.
How many companies do you think are moving to the US?
I think roughly about a 1,000-odd companies end up actually moving there, but India gives birth to 5,000-6,000 startups a year. So, a 1,000-odd going there and many of them ending up dying is not the worst part to happen. In fact, we've seen examples of where people start off here (India), move there (US), then come back here (India). There's a sort of flip-flop that happens and many of them do this under very bad advice.
What is most important is where are your customers and where is your exit, and the capital will actually follow through. So, for companies whose customers are in India, it's better to build in India.
You are part of the India Deep-Tech Alliance (IDTA). What’s the goal of the alliance?
IDTA is an alliance that's been created by various Indian investors. Our firm, 3one4 Capital, is one of the founding members, and we have other members like Gaja Capital, Celesta Capital, Chiratae Ventures, and others.
There's a collective consciousness that given all of us have been doing deep tech investing, let's have a strict, rigorous definition and make a commitment to the future. The alliance is a testament to the sentiment we all share. We have been doing deep tech investments and we'll continue to do so and may allocate a higher portion of our capital. We counted that number and that’s how we reached an investment goal of $2.5 billion.
What would be the process of making these investments in deep-tech and AI?
It’s not a fund we are creating, but we will commit a certain amount from our existing funds. We are managing a capital of around $570 million, and close to 15-20 per cent of it is going towards deep tech. When we do our next fund, I'll have a higher deep tech allocation. It would be similar for other firms, as they may have dedicated deep tech vehicles. So, this is a combination of dry powder — a combination of powder that's getting collected over the next 12 months.
How many companies is the alliance looking to invest in and what would be the average ticket size?
Five years is the average portfolio construction size but the money may be fully-exhausted over a period of seven-eight years. A minimum of 100 companies will end up getting funded and it can go up to 250. But depending on the market, it can go as low as 50 companies.
The average ticket size depends on which stages we come in. So, Gopal (Gaja Capital) invests in late-stages, and we come in earlier stages. We do an average of $2 million, and Gaja can end up doing an average of $5 million.
What role do you think the RDI fund is going to play to catalyse funding for deep-tech firms?
RDI is a nucleus. It’s building the bridge between research to actual development and commercialisation. Previously, a large amount of intellectual research would actually be done over here, but developed and researched out in the West. The research part has now become a lot stronger because the funds like the RDI have allowed these researchers to start commercialisation of their structures and to actually give them the support they need.
 "Siddarth Pai, founding partner at 3one4 Capital")