India’s AI Opportunity & The Future of Startups | Fireside Chat with Sanjeev Bikhchandani

本站添加: 2026-05-30

[00:00:00]Sanjeev, you are widely recognized as a visionary leader and a pioneer of India's internet landscape. We want to broadly deal in three three areas. One is, of course, the whole AI related challenges and opportunities. The second subject is the evolution of newest tech companies. Third, time permitting, you know, investing into these companies, especially when they transition into public market. So, coming to the first one, Sanjeev, let's begin with a critical observation regarding India's position in the global digital landscape. India stands as world's most enthusiastic consumer of global digital platforms with staggering user numbers.

[00:00:37]We have largest Facebook audience, over 400 million users.

[00:00:41]Instagram, 472 plus million users, surpassing US. WhatsApp, you know, is over 4 530 million active users.

[00:00:49]The key and we are already a second largest user of Anthropic, you know, just behind US. The key concern that arises is India has primarily been a consumer rather than a builder of these foundational platforms. This leads to the core anxiety. When global investors assesses India through an AI lens, they often perceive it as world's largest captive market for tool builds elsewhere, rather than a nation posting the next Nvidia or OpenAI.

[00:01:13]Having closely observed the rapidly evolving landscape of global technology and innovation within Indian context, Sanjeev, we are keen to understand your perspective. Where do you see India digital tech stacking up globally? And do you believe India can emerge as a meaningful innovator of newest tech in coming years? If so, what will it take us?

[00:01:30]Better.

[00:01:40]So, it's not as if India will make a lot of LLMs and make them fast and, you know, and so on.

[00:01:46]What we have going for us is data. We've got >> [music] >> possibly the largest amount of data after China.

[00:01:54]And China is a walled garden, so you don't get Chinese data.

[00:01:58]So, the question really is what can we do with our data?

[00:02:01]Uh But our GPUs will, at least for the foreseeable future, have to come from outside.

[00:02:09]The LLMs we use will be global LLMs, uh right?

[00:02:13]Of course, there's enough open source out there.

[00:02:16]Okay, so you can do some stuff.

[00:02:18]But our IP will be built around our data.

[00:02:21]Now, can we do stuff? Answer is yes.

[00:02:25]Uh it will necessarily have to be collaborative with uh mostly the US companies.

[00:02:33]Uh maybe a little bit of China.

[00:02:36]Uh and it's not as if we can't uh succeed with that strategy.

[00:02:41]We can.

[00:02:42]Would I love to have a private Indian LLM that's global? Answer is yes.

[00:02:46]Do I see it happening next 5 years?

[00:02:48]Answer is no.

[00:02:50]And we have to live with it.

[00:02:52]>> But but Sanjeev, uh you know, uh in the context of you know, our consumer tech companies, you know, uh including yours, Naukri or some of the ones that you uh of course invested early, uh you know, like Zomato or uh or Policybazaar. You know, they deal with you know, millions of data points like what you said, which is also a which is also a treasure in some extent. And of course, you know, uh they deal with millions of users, billions of data points. Uh you said 5 years, but uh I mean, Prasanna in his opening remark mentioned that you know, US took 30 years to build all of that.

[00:03:22]>> Listen, let's just understand how much capital it takes to build an LLM.

[00:03:26]Right? So, let's take a you know, Open AI ChatGPT.

[00:03:31]Let's take a you know, Anthropic. Let's take any other LLM.

[00:03:35]I mean, tens of billions of dollars, right? Is what it took. Now, and it continues to suck in a lot of money. And then there are data centers, right?

[00:03:45]Uh So, in India, who has that kind of capital?

[00:03:51]Uh it's not venture capital.

[00:03:54]And the venture capitalists did not create the, you know, the the LLMs in the US.

[00:04:00]Cuz venture VC funds are like 8 years, 10 years, 12 years.

[00:04:03]This is a 20-30 year game.

[00:04:06]Right?

[00:04:06]Uh, you're here from Excel, right? You think you think Excel in India can fund an LLM for 30 years? No. The funds are not structured like that. They don't last long.

[00:04:15]Okay? Who can do it? Government can do it. Large public sector companies with big balance sheets can do it. Very large private sector, you know, companies like Reliance, Adani, Tata, Mahindra, they can do it.

[00:04:26]Right? You need to deploy that kind of capital.

[00:04:30]And very often, uh, there is no light at the end of the tunnel. You don't You don't know if these LLMs will they be profitable?

[00:04:36]Don't Even now in the US, you don't know.

[00:04:38]So, I don't know, you know, whether we are positioned to do it. I mean, you know, we can sit over here, you know, but we're not putting money in.

[00:04:45]We're telling them, "You put money in."

[00:04:47]Nobody's going to do it.

[00:04:49]Right? Uh, so, so, I think in India, we have to do We have to cut our coat according to our cloth.

[00:04:55]We do not have the kind of capital that US and China can deploy.

[00:04:59]And therefore, we are There there will be fewer concerted efforts, possibly government-supported, which which will perhaps make some headway in this area.

[00:05:09]>> So, uh, so, Sanjeev, in some sense, you know, recalling your early days in Naukri, uh, when your competitor jobosahead.com, you know, uh, ahead of times raised a lot of money and, you know, posed a potential disruption. Do you see a moment like that? Uh, do you think you were lucky, of course, but do you think, you know, our ecosystem will be lucky?

[00:05:27]>> See, the ecosystem may may or may not be lucky. Companies have to be lucky.

[00:05:31]Now, will there be enough lucky companies and good companies? Answer is yes.

[00:05:35]Okay.

[00:05:36]Uh, we ourselves, for example, uh, we began to invest in AI pretty early.

[00:05:41]And before GenAI, I mean, around 2008-9, we began to do stuff which today is called AI. We didn't know it then. You know, we were doing it.

[00:05:49]But now if I look back, that's AI.

[00:05:51]Right? That time we didn't call it AI.

[00:05:54]Right now we've got about 130 AI engineers. We've got another 30, uh, you know, ML engineers. 150, 160 inside our company just working on Naukri.

[00:06:04]Uh, to using AI to improve what we are doing for our clients. So, uh, Sanjeev, we'll uh, shift gear to the new age tech companies in India. When you embarked in your journey for naukri.com, you know, internet penetration was of course very nascent. I mean, you acknowledge uh, that, you know, when you uh, hosted Naukri on internet, India had barely 14,000 registered users. Yeah, but fast forward today, the digital economy in India is booming. It's a vibrant startup ecosystem has emerged already. The startup ecosystem has undergone a remarkable transformation. And here what it is. You see, the focus is now shifting from the fundamental question that does the problem even exist uh, to the target market uh, you know, sizing uh, to the validation of unit economics and then to next is profitability.

[00:06:50]Uh, where do you see, you know, this going? What is the next phase of this evolution? Where do you see the trajectory?

[00:06:59]So, when we started, uh, there was no venture capital in India and there were 14,000 Indian accounts.

[00:07:04]And we had no money. We were broke. Uh, and we said we have to break even in this. If you don't break even, you can't survive. So, we had a sales target.

[00:07:11]>> [music] >> Our sales target was simply this, that if we can get 500 companies uh, in India to give us 1,000 rupees every month to list one job on the website every month.

[00:07:24]Right? Well, that's 5 lakh rupees a month. Uh, that's 60 lakh rupees a year.

[00:07:28]Up. Uh, if you can reach that in 3 years' time, uh, you know, we can be profitable. I can take a salary.

[00:07:34]It was a simple sales target.

[00:07:37]You know, don't talk about the vision vision. There was no vision, yeah. There was just a simple sales target. So, companies that have a sales target break even and make a profit.

[00:07:45]Companies that have a vision make a loss.

[00:07:47]Right? Let's be clear about that. So, better have a sales target and keep it going.

[00:07:52]Achieve it, make a profit, then you can evolve a vision. So, I think I think Then people vision you know you know you know, all of that is good. But better have a profit now.

[00:08:09]Get to break even.

[00:08:10]>> So, so Sanjeev, on the related note, right? You know, you coined a term called repeatable profitable unit.

[00:08:17]Up, you know, what I learned is that you look at almost 1,000 startups every every quarter as a as a firm, of course.

[00:08:23]Did you in the recent time, you know, come across, you know, something, you know, which embodies this philosophy?

[00:08:27]>> You know, we we used to sell by direct mail. Okay, I'm two companies. Uh we had a mailing list.

[00:08:34]And they would send us checks in advance along with the job information, which we would upload on the site.

[00:08:40]So, we had reached three lakhs a month of sales.

[00:08:42]Okay, when we raised venture capital from ICICI Venture.

[00:08:45]And then one day Hitesh had a brainwave. He says, "Yeah, let's hire some three four sales people in Delhi market and see what happens."

[00:08:53]We had got the money in the bank.

[00:08:58]So, he said, "Okay."

[00:09:01]Hitesh hired four sales people and they went out in the field to try and sell.

[00:09:05]And you know, we discovered a few things.

[00:09:10]The The total cost of an average sales person uh was bikes that were my neighbor.

[00:09:17]We basically 10,000 rupees salary.

[00:09:19]I think sales incentive, uh you know, [music] conveyance.

[00:09:24]Uh you know you you load on the office rent, the air conditioning, the you know, shared computer that cost all that.

[00:09:31]Uh so And we learned that within four months' time, the average sales guy was delivering revenue of There is no variable cost in putting up an extra job or not really.

[00:09:47]Except data entry and that's very little.

[00:09:50]So, essentially, the sales person was making a 28,000 rupee profit on his own cost.

[00:09:59]And this number was headed north because 50,000 was growing.

[00:10:04]So, we said, "Boss, this is a profitable unit. A sales person is a profitable unit and it's repeatable. Just keep on adding more sales people across the country until the last sales guy is not making a profit on his own cost."

[00:10:21]So, we That's what we did. Within 3 years, we had got 11 offices around the country and 240 sales people and we had grown 35 times in top line and we made a profit.

[00:10:35]So, what we always say is for a company to scale profitably, you need to discover your repeatable profitable unit and then double down on it. Step on the gas.

[00:10:48]Now, your repeatable profitable unit could be anything. In our case, it was sales guy.

[00:10:53]>> [music] >> Uh in the case of a QSR chain, it could be an outlet. It could be a hamburger.

[00:10:58]Could be whatever.

[00:11:00]Uh but, find your repeatable profitable unit and just double down on it.

[00:11:04]>> Interesting. Interesting. Thank you, sir. I'm mindful of time. Of course, I would love to ask more and more questions and keep learning, but my last question, you know, and a very quick uh response from you on what is the mantra for surviving difficult times as an investor like we are currently on.

[00:11:17]>> I I think stay focused behind and getting behind frugal entrepreneurs.

[00:11:23]Okay. And guide them through a tough time.

[00:11:28]And yes, sometimes you may have to give uh more money than you thought to guide to of allow them to come through and you have to demonstrate the conviction.

[00:11:36]Uh and and and and play the long game. You know, we have been uh we invested in 2008 in PolicyBazaar. We are still there. 18 years have gone by.

[00:11:47]Uh We invested in 2010 in Zomato. 16 years have gone by. We are still there.

[00:11:53]So, in India, if you have to invest in a startup, you have to invest in it. You have to play the long game.

[00:11:58]>> Uh thank you. Thank you so much, Sanjeev. I'm going It was uh really really candid and insightful and uh you know, thank you so much for your time.

[00:12:05]It's absolutely a pleasure and all the best to you. Thank you, everyone. Thank you.

[00:12:12]>> Investments in securities market are subject to market risks. Read all related documents carefully before investing.