"India missed out on chips, models but real opportunity is in applications"

Nitin Sharma, partner at venture capital firm Antler India, discusses where Indian AI startups can compete globally and why founder quality matters more than age

India may have missed the early layers of the artificial-intelligence technology stack — chips, infrastructure, and large language models (LLMs) — but venture capitalists are betting it can compete in AI-powered applications and services. Nitin Sharma, a partner at Antler India, says his firm has become one of the world's most active AI investors by backing founders who can adapt quickly as models and markets evolve. In a video interview with Peerzada Abrar, Sharma explains why Antler is compressing startup creation into weeks rather than months, helping Indian founders relocate to Silicon Valley, and why founder quality now matters more than domain expertise or age. Edited excerpts:

Q: Antler has been named the world's most active AI investor. What's driving this conviction?

We are most active globally and we're also now most active in India. We have 51 AI-native or AI-first investments. These are AI native or AI-first companies that could not have existed without LLMs or before 2022. It's quite clear this is the biggest technology shift of our lifetime. When you're investing so early, you don't look for traction or data. You look for just clarity of thinking from the founder and you look for where the size of outcomes can be the biggest.

Q: Is this a spray-and-pray strategy, or is there a specific thesis behind this volume?

Over 80 per cent of our investments happen through our residency models, which means we're focused on company creation. We spend many weeks and sometimes months with founders as they are ideating and validating. This is very different from a spray-and-pray approach, where a deal comes in, you make a few calls, and decide to invest.

Q: Where is India genuinely competitive versus where are we still playing catch-up?

India clearly missed the bus on several foundational layers—infra, energy, chips, models. At the same time, we are still seeing some exceptional founders building tooling companies and developer platforms. Where the bulk of the opportunity lies is in applications and services.

 

Every workflow across enterprises and businesses can now be reimagined in an AI-first, AI-native way. The opportunity set here is massive.

 

That's why we focus far more on the founder than the idea. We back founders who are fundamentally exceptional — people who can adapt, move fast, and stay nimble. They are not rigidly attached to a single idea or business model; they evolve as the landscape evolves. We still haven't seen truly AI-first, category-defining products in areas like edtech, healthcare, or agriculture. That shift is coming over the next couple of years.

 

In periods of high uncertainty and rapid change, the right approach is to back outstanding founders—not fixed business plans or static products.

Q: Are there specific AI application areas where Indian founders have structural advantages?

Seventy per cent of what we are investing in is essentially Applied AI. (It) Could be companies which are trying to do something like Shopify in an AI-first way, could be companies which are trying to do hardware plus AI, things which were not possible earlier. We have funded an AI-native sleep assistant, an AI-native cooking assistant. We have also funded founders who are trying to apply supercomputing and AI to other problems like gene editing.

 

The services part is also very interesting because we are known for services. We will have many new types of services companies because services are not going away for a long time, these models will be very fluid. India's services companies will have to adapt. Some will die but in the next few years we will have some new types of services companies because Indian founders are very uniquely positioned to run a services company.

Q: What's the evidence your AI Residency model works?

We got about 2,500 applications. We selected 1 per cent of them, 26 teams. We invested in six companies. So it was very selective - 6 out of 2500. Globally we've had unicorns come out of residencies. We have a company called Airalo which has scaled from hundreds to millions of dollars of revenue, with high profitability.

Q: What kind of investments do you make through this programme?

Within four to six weeks, startups from the time when they apply get a decision and $500,000 (or about Rs 4 crore) of upfront investment right away plus many multiples of that reserved for follow on, plus about $1 million of AI-focused perks and $3 million of general perks.

 

For the ones that are focused on the US, we run Antler Embark where we are taking these companies to the US. Within a few months they are up and running as a funded, US-focused startup which otherwise would take two or three years.

Q: What AI categories are you most bullish on for 2026? And what's overhyped?

We're not avoiding ideas as much as we're avoiding certain founder profiles. We're cautious of founders who are simply slapping 'AI' onto a company name without a deep understanding of the problem they're trying to solve.

 

Our median founder age has dropped from around 35, two years ago to about 29 today. We're absolutely open to backing very young founders, even those who haven't worked extensively in a particular domain, as long as they can figure it out quickly. What we're far less excited about are older founders who are approaching this space with a very traditional, SaaS-first mindset.

 

A good example is a former intern we backed who became a founder. He was 23 or 24 and decided to go to the US through the Embark program, focusing on SAP maintenance. Six months later, he had signed a Fortune 100 customer and raised a round from a tier-one global VC.

Q: How much does infrastructure hold back the AI companies you back?

For founders building in an AI-native way, you need more capital, faster decisions, and much more support. That's why we've compressed our process into a few weeks, doubled our cheque size, added perks worth $1 million-$3 million, and increased follow-on reserves. While compute costs may rise, manpower costs reduce meaningfully because teams can build faster with AI tooling. We believe $500,000 to $1 million is sufficient for most applied AI ideas to get started and run early experiments.

Q: Is the trend of founders moving to the US — some call it brain drain — rapidly increasing?

We call it the epicentre effect. This is not unique to the AI wave. It happens every time there's a major technology shift — whether it was the PC/internet wave in the mid-1990s, mobile and cloud around 2010, or even parts of Web3 between 2015 and 2020. Anytime there is a big new technology shift many founders gravitate to the epicentre from all over the world and that happens to be Silicon Valley.

 

One thing that has happened is that younger founders are moving. I think this is a phase, it's not permanent. Maybe another year or so. I suspect that in one or two years you will see the slowdown because more will need to be done on the application and services side and for founders who are doing those things they will build those companies in India.

Q: How do unit economics look different for AI companies versus SaaS or marketplace startups?

Think of edtech for example. You had to spend a lot of money in marketing. One of the reasons the products themselves were not delivering good outcomes.

 

ChatGPT is the example that all of us use as a great product. AI products, because they are actually fairly intelligent, should be able to engage users more, should be able to provide value which is better, should be able to keep users for longer. Fundamentally these products should deliver more value to the user whether it's a consumer or a business and hence, if you have a product that is really solving a user's value your unit economics will never be bad. ChatGPT's retention curves are the only product ever in the world where with time the economics are getting better, not worse.