AI has come a long way and is no longer just a Silicon Valley buzzword. Everywhere you see these days, be it startups, enterprise software, smartphones or even healthcare apps, artificial intelligence is quickly becoming part of everyday life. But while the global AI race continues to be led by companies like OpenAI, Anthropic, and Google, where does India really stand in all of this? Can India build meaningful AI companies of its own, or are we simply becoming users of technology built elsewhere? These are the key points that need to be addressed.
In an exclusive conversation with Digit, I spoke with Rahul Agarwalla, Managing Partner at SenseAI Ventures, about India’s growing AI opportunity, the hype surrounding the industry, the current funding landscape, and why he believes India’s biggest strength may not lie in competing directly in the frontier model race, but in building AI-first products that solve real-world problems. Here are the key excerpts from the conversation.
Rahul Agarwalla: Both are true. Real value is absolutely being created. If you look at the revenues, there are already AI startups globally doing over $100 million (roughly Rs 959 crore) in revenue, even outside the foundational model companies like OpenAI or Anthropic. These are real businesses solving real problems.
Even in our portfolio, we’ve seen startups go from zero to a million dollars in revenue in an extremely short period. One of them did it in about a year, and another did it within a quarter. Revenue is the biggest proof because customers only pay if they’re getting actual value.
At the same time, there’s definitely hype as well. Some valuations are probably stretched and there are companies getting funded simply because AI is the buzzword right now. But every major technology wave has hype attached to it. That hype also pushes adoption because companies suddenly realise they need to start investing in AI solutions.
Rahul Agarwalla: I don’t think it’ll jump 5x. My expectation is probably closer to around 2x growth, somewhere between $4 billion and $5 billion. The reason is that application-focused AI companies don’t need massive amounts of capital compared to infrastructure-heavy businesses like data centres. The economics are very different.
India also has a huge advantage in engineering costs. The cost difference between Indian and US engineering talent can easily be over 10x. So, given the large talent pool we have we are able to curtail cost from that and in building AI fundamentally, especially at the early stage age. The only real cost is people. You can get credits from all the AI labs, which are the open APIs of the world, or from hyperscalers like Amazon, Google, and Microsoft, who will all give you credits to build products. It’s only when you become large that they want to make money off you. So in the early stages, the real cost is people, and our people costs are much lower than what you see in the West. That means our ability and leverage on the same amount of capital is much higher.
Rahul Agarwalla: I’m not saying we should stop building models entirely, but India’s biggest opportunity is clearly in applications. Around 75 per cent of the startups we saw last year were application-focused, and I think that’s the right direction because that’s where we have a real competitive advantage.
India has decades of experience building software and solving business problems through technology. That capability translates very naturally into AI applications. The biggest value creation in AI will eventually come from solving real-world problems, not just from building foundational infrastructure.
That said, we still need sovereign AI models for Indian languages, culture and local needs. Foreign companies may never deeply prioritise those areas. But the goal shouldn’t be ‘beating OpenAI.’ The goal should be building systems that work best for Indian users and Indian use cases.
Rahul Agarwalla: We use something called the VDAT framework to evaluate AI businesses. VDAT stands for Variety, Data, Architecture, and Team. These four things help us identify whether a company is building real AI technology or just wrapping an API around an existing model.
Team is important because if the founders don’t deeply understand AI, it becomes difficult to build meaningful products. Architecture matters because using the ‘latest AI model’ isn’t enough. The technology choices need to fit the actual problem being solved.
Then comes data. Without strong proprietary data, it’s very hard to build defensibility. If your product is built only on public data, large foundational models already have access to that information too.
Finally, there’s variety, and how complex the problem really is. AI becomes valuable when there are multiple inputs, multiple outputs and highly contextual decisions involved. That’s where genuine AI-first companies stand apart from simple wrappers or demo products.
Rahul Agarwalla: I think India is already building very meaningful AI companies. The companies we’re funding are generating real revenue and creating actual impact for customers.
One example is a dermatology startup in our portfolio. Users upload a photo of their face/skin, the AI analyses it, asks questions like a dermatologist would, and then generates a personalised treatment plan reviewed by doctors before medicines are shipped.
What’s important is that this isn’t just about convenience for metro cities. It’s solving a genuine access problem in tier 2 and tier 3 India, where specialist healthcare availability is very low.
So when people ask whether AI is real in India, the answer is yes. These companies are improving productivity, solving accessibility problems and building scalable businesses at the same time.
Rahul Agarwalla: Double down on applications and data, very clearly. That’s where India has the strongest right to win globally. We already have the engineering talent, software experience and startup ecosystem to build world-class AI application companies.
I’m not saying we should completely ignore foundational models. We absolutely need sovereign AI systems for Indian languages, communities and cultural contexts. But trying to directly compete with companies spending billions on frontier models is probably not the smartest allocation of resources right now.
The better strategy is to build applications that solve meaningful problems, while also creating AI systems that specifically serve Indian needs. That combination gives us the biggest opportunity globally.