Sajith Pai on Startups and Venture Capital

Shruti Rajagopalan and Sajith Pai discuss investing in the Indian economy and the different approaches that are needed for different markets.

In this episode, Shruti speaks with Sajith Pai about the various sectors of the Indian market, unique payment systems, public versus private investment, the role of artificial intelligence and much more. Pai is a long-time media executive turned venture capitalist. After working in media and entertainment, much of it in The Times of India Group, across strategy and corporate development roles, he moved to Blume Ventures, one of India’s leading early-stage venture firms, in 2018. At Blume, he actively focuses on and support its investments in the domestech space, including consumertech, smb saas, b2b marketplaces and more. He also writes on tech, business, culture and their intersections.

SHRUTI RAJAGOPALAN: Welcome to Ideas of India, where we examine the academic ideas that can propel India forward. My name is Shruti Rajagopalan, and I am a senior research fellow at the Mercatus Center at George Mason University.

Today my guest is Sajith Pai, who is a partner at Blume Ventures. Sajith is a long-time media executive turned VC. At Blume, Sajith supports investments in media, ed tech and e-commerce, while simultaneously helping Blume building a research and knowledge platform.

Sajith has an MBA from IIM Ahmedabad and a B.A. in economics from Chowgule College, Goa. He is also a writer, and you can find his writings on startups, e-commerce, venture capital, culture, political economy and education on his website,

We spoke about the 2023 Indus Valley Annual Report written by Sajith and his co-author, Amal Vats, at Blume Ventures; the many countries that make up the country of India; artificial intelligence; the venture capital community in India; manufacturing growth and more.

For a full transcript of this conversation, including helpful links of all the references mentioned, click the link in the show notes or visit

Hi, Sajith. Welcome to the show. It is such a pleasure to have you here. I’ve followed you and your work for a long time, and I’m very excited to have this chat.

SAJITH PAI: Hey, thanks, Shruti. Thanks for having me over. Likewise, big fan of your writing and the podcast. It’s probably my favorite podcast.

RAJAGOPALAN: I’m so thrilled to hear that.

PAI: Don’t tell Amit Varma that.

RAJAGOPALAN: I know, I won’t. I’m not giving Amit Varma any free publicity if I can help it. I think he has more than enough. I want to kick off a conversation with the latest Indus Valley annual report. This, of course, is all written by you and Amal Vats, who’s both your co-author and also your colleague at Blume. This is also a continuation of the same effort that you have made last year. The lovely thing is, you’ve made it almost a pedagogical exercise that goes viral in the Indus Valley, which is the Indian startup pun on Silicon Valley.

The Many Countries That Make Up India

RAJAGOPALAN: There’s just so much to learn, I feel, in terms of the broader narrative of what’s happening in India from this report, not just the startup scene. I want to start with this lovely classification that you and Amal have developed, which is India 1, India 2, India 3, where India 1 is the very elite, about 120 million in size and per capita of Mexico roughly, and perhaps more elite in education than what is signaled by Mexico as a country. Most of the startups operate here. This is the base of the most sophisticated consumers, the most mature users who spend a big chunk of their consumption expenditure in the space.

Then you talk about India 2, which is about another 100 million people, and then the rest of the country is India 3, which is really sub-Saharan Africa level when it comes to GDP per capita and consumption expenditure and so on. My sense from this when I see the overall startup market is that perhaps it’s a little too hot. What I mean by that is, you know especially e-commerce and fintech, they are growing much faster than India 1 and India 2 are growing in terms of GDP per capita.

Is that a fair way to think about it, in the sense that it is the GDP per capita of $2,200, which is the absolute concrete ceiling for the growth of many of these startups, or am I missing something else? There is some unexploited consumer base, which is not that tightly linked with income, and all the money flowing into these startups is not really just hot money ballooning into something else, but actually genuine growth?

PAI: Thank you for capturing that so well. You’re absolutely right, in a way, that there is a—I want to choose my words very carefully because I’m both an inhabitant of that world, yet I aspire to be an observer and critic. I do feel there is a disconnect, if we can say, between the funding that India receives and the fact that the economic engine, and pretty much sometimes the sole market for many products, is really India 1. I daresay there is perhaps a smaller market tucked away within India 1, which we refer to as India 1 Alpha, which is the world of iPhone and Starbucks and Netflix and all of these DTC, direct-to-consumer brands that pop up.

That is like a Taiwan, you could say, 25 million people, 8 million households, $35,000 per capita income. That’s tucked away within India 1. I think a fair way to look at this disconnect is to say that the Indian startup ecosystems solve a certain problem for the West, which is it solves the capital allocation problem for a tiny set of managers sitting in the U.S. and Europe, worried about the fact that they have to pay those pension bills which are due and looking at all of their investments.

These are folks sitting in the large endowment, the university endowments, sometimes in organizations such as CalPERS or Ontario Teachers’ Pension Plan. What they do is, they look at this large Excel sheet and say, “We’re not getting terrific returns on the market. We need to get even better returns. Let’s look at alternative assets.” Within the alternative assets, venture has been the hero category. Really within that, we’re looking at India as a market. Clearly, what I think India does—in some ways, we are solving a certain capital allocation problem, to put it bluntly, for a certain set of managers.

I do feel that if you look at actually the Indian venture market, which this last year, 2022 calendar year, was about $24 billion—$24 billion consisted of broadly. There is a very early stage, which is where the fund that I work out of, Blume, operates—barely 10% of that 24, not even 10% of that 24. The bulk of that is growth money, which comes from large hedge funds, crossover funds. They’re investing in startups because of a very unique problem or unique situation that is arising.

Historically, venture funds are used within startups, and those startups go to IPO [initial public offering] in three, four or five years. Amazon, four years; Google, five years or six years. As tech started giving great returns, more money started going into tech, the elongation of the journey from the first funds received till IPO, it really got elongated. What happened is a lot of these larger funds saw the pre-IPO funding or growth pre-funding as a viable proposition because it said, let’s suck out more value out of this instead of leaving it for retail investors to grab.

I think a lot of that is happening in India 2. For example, boAt, which is one of the most successful direct-to-consumer brands in the consumer electronics space, got money from Warburg Pincus saying, “Do not go IPO. Here’s the money; take it and build more.” Similarly, Lenskart was able to stretch out its IPO. I’m probably bringing in too many items and probably stacking arguments a bit much, but fundamentally, you’re right, it is hot. There is a disconnect between the actual market as well as what is coming into the venture market. A lot of that is coming from U.S. larger investors seeking to solve a particular problem, and that might be creating certain problems of indigestion, if I may put it like that.

Elite Imitation: Venture Capital Edition

RAJAGOPALAN: If I may add, I think you’ve explained the financial side very well, which is the demand for this kind of growth, which is being fueled by U.S. endowments, private equity funds and, of course, the big VC [venture capital] network and world. The other part of it, I feel, is also perhaps a misunderstanding of India 1 Alpha, which is Taiwan, versus India 1, which is Mexico, versus India 2, in the sense that there are very few goods and services like the big e-commerce platforms like Flipkart, Amazon or maybe ed tech, which is an aspirational product for India 2 and India 3.

They are willing to spend on ed tech to hope that the children and the next generation will make it to India 1. Other than some things like in these sectors, it seems like fundamentally the India 1 [Alpha] subset and India 1 and India 2 are fundamentally different markets because it’s not easy to monetize India 2, India 3. You can see on YouTube, there are lots of people willing to give their eyeballs on YouTube. It’s a relatively poor country, low opportunity cost of giving the eyeballs for lots of content.

Jio has made content very cheap, but none of them can ever enter the subscriber model willingly, other than maybe gamification or some small pockets and niches here and there. It’s very difficult for Netflix to capture the 800 million market, which is on very, very high-quality broadband, in the same way that it is for YouTube to capture that market. Is some of this growth also being fueled by this fundamental misunderstanding of what is the fundamental difference between these two sectors?

One I can think of is just the ability to pay, both in terms of cash flow and income. The other is the user experience. You’ve written about this on your blog separately on how India has native-level English speakers. They’re a tiny, tiny group, they’re basically Taiwan, and then the English-comfortable. English-fluent, that’s a slightly larger group. If we really want to capture 800 million people who are online to come onto the services, the user interface also has to fundamentally be different. It shouldn’t be in English. The transaction shouldn’t be entirely reliant on English fluency. Is that the supply-side misunderstanding when people are trying to fuel this growth?

PAI: Yes, it was a great question. Let’s unpack that. You’re absolutely right in the sense that there are two broad markets, India 1 and India 2, and both behave very differently. Both transact slightly differently, and both interact with products, software products as well as some hardware products, differently. India 1, and whether it’s India 1A, but India 1 broadly is very familiar with—it’s very similar to how, for example, their peers or counterparts would be in the U.S. They are happy to look at dense, information-packed flat-screen monitors or glass rectangles in their hands, click on colored rectangles using their mouse and pay with their credit cards.

India 2, many of them don’t have credit cards because, A, we have only 35 million credit card holders, all of whom live in India 1. India 2 struggles to get credit. B, they are mystified by the checkout experience. Many of them actually found out that the checkout page, and including the page which takes them to the OTP [one-time password]—which is a very unique Indian thing, that every transaction has to be consummated by feeding in a one-time password which comes to your mobile phone—is only in English. That’s strange, right? That page mystified a lot of people, apparently. Even the format of shopping, looking at 10, 12 products, choosing between them, there is a tyranny of choice for them too.

They’re used to going in and asking for humko daal de do. The shopkeeper gives them daal. I would say in all of these things, there is a disconnect with what is the native UI [user interface], the native transaction model, the native business model, which is well suited for India 1. When you go to India 2, a lot of these things fail. What you need to do then—and some of the successful Indian startups have evolved a digital vocabulary that is well suited to the needs of India 2.

Sometimes it can be as dream11’s Harsh Jain said, that if you are charging 300 rupees in India 1, do not charge 300 rupees. Instead, charge 10 rupees every day, and you will have customers. Sachetization of that. Second, for example, COD [cash on delivery]. COD was invented by Flipkart because they saw that they were struggling to penetrate that. COD is cash on delivery, where you actually pay when the goods come to you, and solves for a lot of challenges.

BNPL—till now, RBI has stymied it—which is buy now, pay later, emerged as a way to overcome the lack of enough credit cards. Then, for example, voice UIs have emerged. Google says 28% of the search is voice related in India. It’s very high. Many of these models such as Meesho, DealShare have all emerged which are very native to India 2, which are built around resellers, which are built around maybe group interaction and buying.

Even though they’ve not reached China-level numbers, many of these attempts—or ShareChat, for example, trying to popularize native UI—they’re all impressive efforts. I don’t think anyone has really hit the breakaway scale in India 2, other than YouTube, WhatsApp, Facebook.

RAJAGOPALAN: They’re all free.

PAI: Maybe you can—they’re all free.

RAJAGOPALAN: With ad revenue.

PAI: With ad revenue. That said, to be very honest, Shruti, India 2 ad revenue—I don’t think India 2 is easy to sell to advertisers. The bulk of the ad revenue that India sees, $7 billion, et cetera—about 80% comes to Google and Meta, Facebook Meta, which includes Instagram and YouTube as well. About 20%, 25% comes from the others, of which about a billion apparently now comes from retail media, which is Amazon, Flipkart as well. All of this is strictly speaking India 1.

India 2, there isn’t enough monetization yet, but what seems to be happening is there are people willing to pay small sums of money. If you can get into that model, don’t charge them—what Harsh Jain said—don’t charge 400 rupees at one time. Instead, charge micro sachets. Netflix, for example, they’ve actually reduced the price cost, which is 149 rupees in India. They’re just throwing the kitchen sink at the market. Perhaps it’ll get them some traction.

I think the better way for them would have been to do sachets, the way mobile phone data is sold or historically used to be sold. There are a few—ShareChat is one. Stage, which does Haryanvi content, and they call them themselves Netflix for Bharat. Kuku FM, Pocket FM, they’ve all come up with very tiny, inexpensive monthly models or weekly models by which people can pay. We’re beginning to see early signs of consumer paid, but we’ll have to see how big this gets. You’re absolutely right.

India 2 is a country of, I would say, viewers, not really consumers. I don’t think they’ve gotten to become consumers taking out the credit card or wallet or what have you, UPI [unified payments interface]. Very early signs are there, and some of those companies are beginning to double down and keep improving their models. That is, yes, but India 2 is still not a paying market.

Sachetization and Micropayments

RAJAGOPALAN: No, and it’s really interesting because the sachetization is something that Unilever and P&G—and we had Surf laundry detergent, which usually, at least if you live in the United States, comes in these horrible two-, three-gallon sizes. In India, it used to be a tiny sachet because there was a storage problem, there was a cash flow problem and, of course, there was an income gap problem. You can’t spend your weekly or monthly income on a giant tub of laundry detergent.

They had that for everything from hair dye to shampoo to soap. In fact, during COVID, we saw hand sanitizers being sold in tiny sachets. This is something that is very well known in the FMCG [fast-moving consumer goods] consumer goods market. Even cigarettes, if you go to a paan shop, are sold in singles. They’ll open a box of cigarettes, and they will sell you one at a time. It’s funny, as economists, we have absorbed this, but I guess some of the startup community is still figuring this out.

On that, my question is, is that learning a little bit delayed because of the nature of the startup founders, that (at least until two, three years ago) most of them were from urban areas, they belong to India 1A, coming from elite families, speaking English, making DoorDash for India and then converting that into then a slightly more Indianized experience? Is it what Alex Tabarrok and I call the typical elite imitation—if you inhabit the Netflix world, so that’s the world they live in?

Or is it because the way the venture capital funnel and situation works, it is not accustomed to these kinds of small and micropayments? They like to see big numbers, big growth. They want to capture a particular type of consumer, and they’re not that thrilled about capturing these consumers. Where do you think this is coming from?

PAI: Partly what you’ve said is true, in the sense that these are India 1A. Typically to start up, you have to be privileged to start up. No one who’s really, I think, trying to take care of their family is going to come into startups. They want to join Infosys or a TCS or even a McKinsey, for example. You want a stable job. All startup founders are super privileged, even though they will all say they are middle class. Perhaps that typical Indian tendency to say you’re middle class is all there, but they’re all privileged.

Shruti, certainly the fact that they are not used to some of these transaction models is a factor. I think an important factor is also that most Indian founders start by looking at the West, at Silicon Valley. You learn from, for example, YC’s Startup School. You listen to the podcasts, for example, whether it’s 20 VC or the Colossus’ “Invest Like The Best” podcast.

You hear of those models, right? Then you will go to Y Combinator, or you’ll read the First Round Review. They read Paul Graham. When you read the canon, when you read all of that—and then none of this is explained there. Paul Graham doesn’t write about sachetization, right? 20 VC’s Uber interviews don’t talk about cash. They say credit card, and just, there’s no OTP there. Uber had to introduce cash in India.

I think there is a tendency for the startup world to think of everything from themselves, derive everything, first principles. I would say that, because I have done 18 years’ experience in a traditional Indian company where a lot of FMCG professionals came into, and because a lot of my batch mates from B-School went to FMCG companies, I’ve read a lot of work in that. I know for a fact from first principles that there is a job to be done, positioning, segmentation. There are enough models which exist, clear quality, minimum viable product. There you have the test market, for example, and you have frameworks, for example.

For various reasons, I think Indian startup founders like to look at the West, try and apply that directly, and then when that doesn’t work, then they look at China. China is also very different. Not all of China is like India. China is in many ways an affluent country. Shanghai and Beijing and all that, $20,000 plus per capita income. Then, finally, they have to sit and derive. Startup founders learn fast. The moment they see something working in India, they double down on it. The speed of learning is incredible. I would say there is a certain, I would say, community of founders who come together to teach what happens to India.

For example, the SaaSBOOMi, there is a direct-to-consumer DTC insider community coming together to say that, “Hey, you can’t apply what happened in the West. This is what you need to do.” Enough writing is beginning to happen and enough podcasts happening—a certain body of knowledge is getting developed as to how do you crack India 2. I think the primary reason for that, or I think one of the two reasons for that, is we look to the West and then try and apply that here, and B, it’s a very alien world for us. Both of those factors.

Cash on Delivery

RAJAGOPALAN: Yes. The cash-on-delivery statistic is fantastic. I think this was last year’s Indus Valley report, where you had this figure in there. It said that 50% of e-commerce transactions, which are mainly tier-one transactions—these are the sophisticated users, they’re the consumers—50% are actually cash on delivery. I found that staggering for a couple of reasons. Is this happening because there are only 35 million credit card holders? Or is it happening because they are a little bit nervous about credit cards getting hacked?

They’re still new electronics users. This is my parents’ generation. They’re a little bit nervous about just giving their information away. They’re worried about hacks and so on. Or is it more fundamentally about trust in the Indian market, in the sense that we know contracts are not enforced? They’re willing to do UPI when they’re already in the auto-rickshaw and it’s not asynchronous. It’s exactly at the time of buying that you’re paying. Or when they go to the vegetable market or the grocery store, they’re happy to do that.

But when they buy now and pay later, when there is a gap, which necessarily requires some kind of contract enforcement. In case it doesn’t happen, they want cash on delivery. They are bypassing the problem of weak state capacity and so on. What is the propelling reason for such a high number of transactions which are cash on delivery?

PAI: Yes. Trust is certainly one factor. Second would be, for example, convenience as a factor, which is, for example, if you are buying apparel—and apparel is a very sizable proportion of Indian e-commerce, close to 30%—sizing is a big issue. India is a country famous for its lack of standardization. Because, historically, e-commerce used to be mobile e-commerce, in the sense, not that things are purchased on mobile—they were—but more than 50% of e-commerce was mobile phones at one time. Because that was how narrow and shallow the market was.

Now, of course, e-commerce has mobile—it’s only about 30% or so of the market. Apparel is now very close to mobile phones in terms of 27%, 28%. Other electronics around 10%, 15%. Because apparel is in a very sizable portion, almost 70%-80% of apparel is purchased on cash on delivery, even in slightly affluent households. Because of the sizing issue, they buy two, eight, 10 items extra.

Third is checkout friction. Sometimes you feel that you’ve historically not got things on time, or things take time. Sometimes you pay and that item is not there; to get the money back is awkward.

I would say these are three factors: lack of trust, the convenience of the checkout and sizing or those convenience factors. I think they all come in. We did check with a very senior person in one of the logistics companies which serves e-commerce. He said that no, it has not changed in the last three years. The minor blips during COVID when people started paying—and demonetization was another one. But otherwise, other than these two times, it’s historically been the same number; it’s not changed too much. Yes, there you go.

Public vs. Private Investment

RAJAGOPALAN: Again, this is what I really love about the report that you’ve put together because it tells us something, the deeper cracks and fault lines in India. At least what I took away from this is, the digital public goods infrastructure that we have, which is your India Stack, which is Aadhaar, UPI, the credit system and so on—we’ve just leapfrogged into the next century. We had nothing, and then we just leapfrogged.

The physical infrastructure is lagging, but it is slowly catching up. We do have electrification. We do have clean water, piped clean water, which is almost nearing 100%. We have now metros in tier-two and tier-three cities. All of this is picking up. But there is a lag compared to the digital infrastructure.

Then finally, when it comes to human capital, which is investment in education and health, that is lagging behind both of these, which is why there is an issue with something as simple as staffing district magistrate courts. We can’t fill in all the vacancies, which obviously has implications for contract enforcement.

People, even though they have gone to an English-medium school, are not able to transact in English beyond a fifth-grade level, so they have trouble with the UX or UI interface. It feels like this is another layer which is underneath everything you are telling me about the startup and the venture world, which is booming, which is, this stuff has not really kept pace. Is that a fair assessment of how we’ve invested in the country?

PAI: Fair. Can’t argue with that. I would say that if I look at, for example, stats (and I’ve been part of the education industry, so to say, before), the number of people—there’s this talk about, hey, India graduates 1.5 million engineers, and we all take pride in that, the second largest after China or whatever. But when you actually dig deeper, actually it’s not 1.5 million; 700,000 or so graduate. Within that, I’ve come to find out that the quality engineers—the kind of engineers, for example, who could probably write clean code or write at least well-functioning code, who are on par with the U.S. engineer—would be around 70,000, would be about 10% of that.

When you add up all the IITs and NITs, even tier-one colleges, 70,000 to 80,000 is that number. When you actually dig deeper and when you just push the numbers, even though it sounds sizable, the quality numbers are a tenth really of everything. Even if you do that with hospitals, if you say that of hospitals which have well-trained EHPs [essential health professionals] or which have these facilities, again, it’ll come down.

Some of it is definitely lack of, I would say, investment—certainly, lack of public investment. I think a lot of it is—I’m not an expert here—is also to do with the way the state has allocated resources and into, I would say, tertiary prestige education of symbols. I think Lant Pritchett said it very well when he contrasted India with Indonesia, and he looked at how India’s education investments by the state has gone into tertiary elite symbols, whereas Indonesia has gone into a little more fundamental one. Not surprisingly, Indonesia has 2X India’s per capita income. India exports engineers to Indonesia and minerals to Indonesia.

Again, I’m not an expert on this thing. I think there are also challenges, and probably it’s something that I want to answer in the next edition. I find it very surprising that Indian cities, large Indian cities, can’t raise money. There is no municipal bond market like in the U.S. or China. I’m not an expert. Your podcast with Alain Bertaud was very illuminating on this account. I do think, because a lot of investment is centrally driven and so on, we also don’t have enough investment going into where it’s really needed, I feel, which is probably primary education, primary healthcare, et cetera. Really, I’m not the expert here on those. I need to be cautious about overstretching my limit.

RAJAGOPALAN: No, I think you’re an expert on those on a different margin, because when you look at how Uber is driving (no pun intended) the transportation gaps that it’s filling in urban sectors, you’re able to see a different part of the puzzle on the missing public infrastructure and urban sprawl than, say, someone like Alain and I are able to see. Because we’re looking at a different picture through municipal bonds or fiscal federalism or something like that.

It’s also a weirder thing where this startup market is so consumer-focused, whereas the government is so things-focused. We want to build things. We don’t necessarily want to invest in people and then let people go where they want. Of course, that is about both urbanization, but it’s also about, do they want to be a linguist or do they want to be an engineer or do they want to be a doctor?

So we are only going to establish the absolute pioneering institutes of linguistics or engineering. We’re not actually going to invest in people and get them up to a point of, they can make any choice that they want. Is this lack of investment in human capital, both health expenditure, education expenditure at the grassroots primary level—is that going to again create a very natural ceiling for the startup world when it comes to going from . . .

One way of saying it is that the startup people go from catering to India 1A to India 1 to India 2. Another way of thinking about it is, the people who were traditionally in India 2 move up into India 1. Both of these things, does the lack of investment in human capital, do you see that as a natural barrier, or will India bypass it in some other way? Because India’s growth story is this bizarre outlier story. It just bypasses all these traditional problems. How do you see that playing out?

PAI: India is a country of leapfrogging. Our story is a story of leapfrogging, and Indian ingenuity will find a way. To your question, I have to say that if the Indian startup ecosystem is a much sizable proportion of India, then it would be in an equivalent country, like it’s even in China, for example. If you look at it, for example, all the money that it’s gotten is about $140, $150 billion over the last 12, 13 years.

Roughly, I would say all of the value that is created (it’s all notional to a certain extent; some of it is listed) is around $650 billion. That is about a fifth of the Indian market capitalization. It’s sizable. If you take just unicorns, unicorns are about 10% of India’s market cap, which is two—double of all other countries. If you look at jobs, for example, startups account for 8% of the organized, formal market types of jobs in India. Startups punch well above their weight.

Increasingly, what’s happened is—and one of my favorite expressions is, supply creates its own demand up in India. What you’re seeing is, all of the money that has come into the Indian startup ecosystem is beginning to seek return. When it seeks return, it expands its limit. If in the U.S., for example, they said that, “Hey, we’re going to solve SaaS, we’re going to solve some consumer things, and now the big money is going to go into space and into biotech, deep tech.”

In India, a lot of startup founders and venture money is actually beginning to go into sectors typically you wouldn’t see necessarily in the U.S. For example, we’ve invested in a startup which is training AHPs, allied healthcare professionals, because there’s a gap in India. We are beginning to look at sectors where there’s an absolute choke point. For example, there are only X number of new veterinarians coming out or X number of audiologists coming out, and they’re very critical.

Startups are beginning to look at trying to work with existing partners and amplify their reach or their ability to serve. That is one. B, startups are beginning to drive—for example, if you take Zetwerk or Nexprt or Groyyo, these are all Indian startups in the manufacturing space. What they’ve done is pioneer a new way of manufacturing, what I call parallel manufacturing, where they work with existing MSMEs [micro, small and medium enterprises], most of whom are not super efficient.

Really what they do is, they get an order from the West and spread it across states and try to use technology, diversify risk, so create cloud factories. I think it is very hard to set up a factory from scratch like China, with 10,000, 30,000 workers. It’s not easy. Just to buy land is not easy. To create this cloud factory is something that Indian startups have pioneered. I would say both these examples tell you how, for example, startups are actually saying, “Look, we need to solve some of these real problems.”

Fortunately, I think there is enough patient capital there coming in. All of these experiments—I don’t know how many of them will work out, but we are beginning to see more braver experiments, more grassroot experiments. Agritech has a bunch of them. Health has a bunch of them, and so on.

Artificial Intelligence and the Labor Force

RAJAGOPALAN: I completely agree with you. This is especially—I think some of this is driven by founders who are not coming from very large cities and very elite education centers or have gone to university in the United States or the India A category. They’re just much more in tune with the real-world problems that people want to solve and are willing to pay money for. On the question of veterinary advice or allied health workers’ advice, do you think India will again leapfrog into AI directly?

Because we can see that GPT-4 is able to pass the bar exam in the U.S. It is able to give pretty competent medical advice on basic questions. It can’t cure cancer, but it can help you through a sinus infection. India has strong regulations on prescriptions and things like that. It’s just never enforced and never followed. Do you see India leapfrogging in that direction?

Therefore, a lot of people are interested in getting India 2 and India 3 just as viewers because they can collect data, they can fine-tune the algorithm. It will help large language models become better. They’ll get better at people speaking into large language models and have a personal assistant or a personal medical adviser, a personal veterinarian, as opposed to traditionally the way other countries have done it, which is build up a human workforce to solve these problems. I see that as the next frontier of how India leapfrogs this problem. I don’t know what that will do to India’s labor force potential, but everyone can’t be a Dunzo delivery person, right? There’s got to be something. Do you see AI solving that?

PAI: Very interesting. I really don’t know. It’s a truth. At this point, I’m looking at all the debate that’s happening around GPT-4. Beginning to see that code writing and certain defined white-collar tasks—like, for example, there’s a friend who’s using it to write marketing copy—can be sped up. In minutes he’s able to get what would take, for example, a freelance writer, like, half a day to write.

There are aspects which are actually where, for example, you can actually see that you may not need as many engineers, but you may need engineers to test that out, for example. Or until GPT-4 can buttress that, so it can mimic testing. Or GPT-5, that will be. I do see that wherever there is a need for a human element to be bundled in—and if GPT-4, for example, you’re fundamentally giving someone the caregiver and ability to kind of not be fully regulated. You don’t have to pass the bar exam equivalent or a license sheet exam equivalent, but you have, for example, a certain kind of iPad with the ability to kind of find out answers. Certainly, you could see variants of that coming.

You may have, for example, the market spreading into, “Hey, I want Harvard- or Yale-trained medic who will give me complete human care.” B might be that, “Hey, I’m okay with human plus medical whatever, and I’ll pay X dollars for that.” And third could be that, “Hey, I’m happy to query a bot and at my own risk.” Would models like this come? Potentially there will be experiments.

Where they could come from—they could be vets. There’s an absolute shortage of vets, for example. And to have, for example, someone like that who’s able to query and provide the care component and with an expert component coming through, for example, like a GPT-4, GPT-5 backend. That could be a possibility.

Certain jobs would get created. Certain jobs—I would say, a lot of the translation, freelance writing—all of those jobs may disappear. Early days. I would love to see how this evolves, but I would say that new opportunity that could open up, like I think you alluded to, so I must speak, is really people plus expertise and sort of an expert light if that could come. That could certainly be—but yes, but really, I don’t know, Shruti. It’s really interesting to—

RAJAGOPALAN: No, I really appreciate your answer. My instinct is also that it’ll be human plus expertise. You know, we have such a shortage of vets, which means GPT-4 and -5 will do the entire chatbot questionnaire, and then the vet only needs to actually physically see you for 10 minutes as opposed to an hour. Instead of creating a new army of veterinarians, we use the existing veterinarians in a much more efficient way and supplement them with a bot as opposed to a medical assistant.

I also very much see it going that way. I’m not hugely worried about, oh, it’s going to kill all the jobs. India already has such a tiny fraction in the formal workforce, and that group can get retrained. It’s a fairly elite group. I’m not panicked about that as much as the leapfrogging part. India just bypassed household broadband, which you talk about in your report.

PAI: Yes.

RAJAGOPALAN: I wrote a Substack on how India just bypassed household phones because for the longest time, there was this huge shortage of phone connections because it was completely state-owned, state-controlled, quota-ed. And then we went straight to smartphones. The actual number of landlines never quite took off in India compared to its GDP per capita. You know, this leapfrogging thing is something we’ve seen happen over and over again, which is quite interesting.

Manufacturing Growth by Stealth

RAJAGOPALAN: You were talking about manufacturing growth a moment ago, and there are a couple of interesting things. In the report, the phrase that you used that I found lovely was “manufacturing growth by stealth,” which is not the typical, a big factory is going to be built and deployed in a particular area. It’s going to employ lots of local workers, thousands in textile manufacturing or shoe manufacturing or something like that. Then we go from there, and that’s very much the South Korean, Taiwanese, Malaysian, Chinese and now Bangladesh story. Everything from shoelaces to iPhones got built that way.

In India, for various mainly regulatory reasons in India, regulation has always punished manufacturing scale, mainly through factor markets, both your labor law, the way the land markets function. Scale is very difficult to achieve in India without having government blessing and government intervention. Now we see scale in infrastructure and things like that, which clearly is pushed by the government, but not so much in manufacturing.

What are the natural limits of manufacturing growth by stealth? To a very large extent, this manufacturing by cloud can reduce transaction costs of manufacturing across a hundred different small factories. Can India actually become a manufacturing giant despite bypassing the scale problem? Or once again, we’re going to hit a natural limit after we exploit the capacity in these?

PAI: I’m going to say that it is going to be hard to cloud factorize your way, if that’s a legitimate term. And you’re right, there will be finite limits to which you can do this because after a certain point in time—it’s this entire Ronald Coase’s, I forgot the term now, that wonderful essay which he wrote about coordinating costs. Tech can take care of certain coordination costs and can push out the boundaries or bring in the boundaries. I’ll keep them out, and I have the technology to coordinate.

I think there will be finite limits soon because you will find that for complex orders, high-value orders, it’s going to be hard to rely on. Then you want to say that, “Hey, I’m going to bring in certain people in house, and I’m going to work with them for my really high-value orders,” and all of that stuff will happen. I feel that to really hit scale, to really hit some of the numbers that Vietnam is hitting—and actually look at the numbers, we didn’t put it out there because it wasn’t the kind of data that—and again, I only had one slide, but that slide fundamentally said, and came from an A.T. Kearney report—which said that the bulk of the production which had left China had actually gone to Vietnam.

A lot of this also is Chinese majors setting up in Vietnam, but fundamentally it’s gone to Vietnam, not so much to India. I think again, if you look at some of the recent numbers there has been manufacturing, for example, exports hasn’t grown the way it’ll have. I would say that there has to be a strong state-supported policy and be willing to see some of that in the PLI policy. Then again, it’s a long process; people have to see it over many years. Without a strong, state-supported, muscular policy, incentives—even state-supported, doesn’t have to be federal, but even state-supported, it’s going to take off.

I think startup engineering will get to a certain point because—and what it’ll do is it’ll create the short-term problem. It’ll solve scale up to a certain level, but the moment you go beyond $200, $300, $400, $500 million, it’ll be challenging for you to work with clearly small factories over a point. I think that’s the story of startups. They’ll solve to a certain point, and after that, they’ll have to go full-stack. Which is another learning that I’ve had, that many Indian startups start—unlike in the U.S., because the volumes are so large, you can be an intermediary, and those thin margins are enough because they keep adding up over a lot of transactions.

In India, those thin margins can’t take you that far because the volumes are not large. In India, there’s a tendency for startups to become full-stack. Where they become private label purveyors, they actually do everything from sourcing to delivery themselves, as opposed to the U.S. model, where you want to outsource everything. I think this will happen in manufacturing too.

Gross Fixed Capital Formation

RAJAGOPALAN: That’s super interesting because, India’s basically a services-driven economy. About 50% plus comes from services, not from manufacturing. Though 50% of the population works in the rural agrarian sector, they’re only about 15% of the GDP, and that’s really also the difference between India 1, India 2 and India 3.

The other weird thing here is, a very large proportion of Indian GDP is consumption-driven and not investment-driven. There are two trends here. One is investment is relatively low compared to other developing countries, and the second part is it’s declining. As economists, one of the things we really pay attention to is gross fixed capital formation, especially gross fixed private capital formation. It has been in secular decline since 2011. Well before people started worrying about the slowdown, when people were still saying it’s the after-effect of the global financial crisis and so on. But India’s never quite recovered there.

Is that another reason you see the end-to-end of all transactions being done by the same startup like Nykaa or someone else, which is, you don’t parcel out different aspects to different people? One part is the spreads are really small; the volumes are not large enough. But the other part is the investment is just lacking in all the other sectors that one needs to rely on. The best way to reduce risk and also become slightly more profitable on the margin, but reduce uncertainty and risk in your delivery times, is to control the entire process end to end.

This is very much Ronald Coase’s nature of the firm that you were talking about, but in addition to transaction cost, there’s a risk and uncertainty aspect which is built into that. Is this also related to investment or lack thereof?

PAI: Partly. I think the real incentive comes from trying to give a positive service experience. To that extent, yes, you may be right that there is some uncertainty. As someone, you want to control that end to end. Barring logistics, which you have reasonably good providers, the deliveries of the world, the e-commerce processes of the world, everything else you try and control. Sourcing, initially, is brands. Brands, typically you pay higher prices. The margins are low; then you start trying to manufacture or own the product yourself. This is driven by experience, then margins.

Third would be, I would say, what you said, uncertainty. Then you try and bring in more and more things in house. This is how the hierarchy goes. Uncertainty would be the third. With specific regard to gross capital formation, I don’t know if that plays this kind of a direct role, but I did find that—I didn’t do enough work to conclusively come up with a set of findings or insights and relay that into slides—one of the interesting points was, you’re right, 60% of India’s GDP is private, like China, which is 38%, and our gross fixed capital formation is around 29%.

China’s is typically one and a half times that. Even what our government spends, as a proportion of GDP, is much lower. It’s only about 10% or so. Again, the gross fixed capital formation, almost two-thirds to three-fourths is private. The government is much smaller. The finding was that the ability to drive more and more investments into private is limited by the fact that they are beginning to get crowded out of credit markets. If at all the next version, I would probably want to look at that a little bit, that India is an undersized credit market, and a lot of our ill, so to say, can be found in that.

Credit as a proportion of GDP is like 55% globally, and it’s 48% here, and I can go into more numbers. Fact that the government also competes in the credit markets and leads to, A, the rates going up higher, B, certain investments not happening where they should, even though private is the larger contributor to gross fixed capital formation. Freeing up the credit market, having a bit more, I would say, not competition but a bit more supply into it, would probably ease gross fixed capital formation creation a lot more.

Decentralized Credit

RAJAGOPALAN: On the credit aspect, how much do you think India Stack can really help? Here I’m really talking about, one part is of course the identity aspect of the India Stack. This is Aadhaar and KYC and eSign, DigiLocker, those sorts of things. Then you have all your UPI-related, everything linked to individual bank accounts, which is also linked to individual identity, which can now be used for peer-to-peer pay. But the third part of that, which is relatively underused, is the credit-enabling infrastructure.

Do you see much growth in credit that is being driven by that sector? Both because now the startup economy has added almost a third of the new formal workers coming into the salaried class, but these are also people who are for the first time—they may not be from families that are historically rich or have strong access to credit, but they are for the first time using this India Stack. Is that driving some of the uptick in credit?

PAI: As of now, no, Shruti. As of now, the growth is not there yet. There is a very ambitious initiative, of which the first stirrings are there. It’s called OCEN, Open Credit Enablement Network. It is essentially flow-based lending. Essentially what it means is, let us say you are a business which sells on Zomato. You could be a baker, you could be a pav bhaji vendor, et cetera, not a very large sole proprietorship, or maybe three or four people doing that.

For various reasons, because you’ve just begun in the last six months or 12 months, or maybe you don’t have an asset; you’re shut out of banking circles because they do asset-based lending. But what OCEN can do, thanks to related initiatives such as an account aggregation framework and all that, is enable what’s called flow-based lending, where someone can look at your flows through a tokenized fashion. It’s encrypted, so you don’t know who it is. You can look at that and say, “Hey, there’s 16 months of continuously increasing revenue. I think this chap is good for so much,” maybe not the last month but maybe the sixth month or seventh month.

RAJAGOPALAN: Absolutely.

PAI: Then you can probably look at some kind of, how the person repays and then probably increase it. We will see some of those beginning, and there is effort underway. We had a chart in the Indus Valley annual report which told you about how, for example, that graph is growing, about 4 million accounts come under this account aggregation framework, et cetera. Will that solve the problems? Well, they will solve some problems. It’s sort of like cloudfactory. It will solve to a certain extent. At the margin, it will expand it.

Startups are great at expanding boundaries, a little bit of bringing in boundaries, so to say. They can expand boundaries. At the margins, they are terrific, but for sustained change and for permanent change, I think you will need a lot more intervention by the state, by the government in expanding that market. I would say that while I do think we’ll have growth, I don’t think on its own, India Stack will be able to shift everything.

Even with India Stack, Shruti, I was surprised that as much as half of India’s UPI payments, for example, are done by about 7% of the people. 6.5% of people account for like 45%. There is a small group of people who are really driving all of this consumption, but is it completely democratized? Well, I don’t know. It’s only 260 million people who use it. It’s about a sixth of the country or 20% or so. I’m not 100% confident that, by itself, it can solve credit problems.

RAJAGOPALAN: The areas where I am particularly optimistic or excited about the Open Credit Enablement Network is in education. This is hard to recover. There is no collateral except the individual’s human capital. But on the other hand, for education, it’s quite easy to get a report card, quite literally, of the past track record, and the decision to say this is a good candidate to invest in.

I think it will penetrate education loan markets. It will penetrate what you were talking about, which is someone who is brewing coffee at home, and now you can see through Zomato month on month exactly how they’ve been growing, or someone’s baking. Those things, they’re also part of the formal economy increasingly, thanks to GSTN, UPI, all of that put together.

PAI: Yes.

RAJAGOPALAN: It’s becoming much easier for that tiny sliver which was earlier kept outside by the gatekeepers to now have access to more credit. I very much appreciate your point of how tiny that group is, the group that will first capitalize on that benefit. And to truly democratize, something else has to get figured out, like when can we buy cars with it, for instance?

PAI: Yes. The metaphor I think always is, if you look at James Scott’s wonderful book, “Seeing Like a State”—

RAJAGOPALAN: “Seeing Like a State.” Yes.

PAI: —India Stack is sort of the binoculars of the state. Just helps you see more and bring in more people. Yes, that’s sort of the metaphor that I’d like to use.

RAJAGOPALAN: We’ve made everything more legible, right? The legibility factor has increased through Aadhaar, Know Your Customer, so on. We’ve reduced transaction cost through UPI and peer-to-peer payment systems and so on, but fundamentally, the factor markets, the public infrastructure, the public goods, the law and order, all the classic things that the night-watchman state has to provide, which can’t easily be solved peer to peer. I think you’re absolutely right, that until that gives, it becomes very difficult to do this.

Different Kinds of Startups

RAJAGOPALAN: One question I had about the broader Indian startup story is I see three things going on, especially in the growth sector. One sector is trying to cater to as many of the India 1A, India 1, India 2 group. This is Amazon, Flipkart, now increasingly BYJU’S and so on. These are startups that were made in India for Indians, and for a large number of Indians at scale.

There’s a second group that I see coming in, which is Indian startups made in India but to solve very niche Indian problems. This is not Netflix. This is trying something like Loo Cafe, which is trying to create smart toilets across India. This is deep tech combined with the UPI system, combined with the infrastructure problem that smaller towns are struggling with, something like that. Or all your climate change deep tech, which is now taking off in a big way.

Then the third part, which is quite remarkable and relatively new, is the India startups who are catering to the rest of the world. This is really the SaaS-driven companies; at least that seems to be the first push. We’ve spoken a lot about the first category; we haven’t spoken much about two and three.

How do you see the second and the third in terms of the growth story? Can India really build for the rest of the world? Can it crack the cultural codes, the way those people transact, the way they pay? Or are we past the point, and only going to be good at building out the interface and doing the backend and things that we’ve been traditionally good at? And also India building for Indians but in very niche areas? How do you see these two playing out?

PAI: Let’s take the first, which is really India building for the West. SaaS is pretty big. A lot of venture investors will do just SaaS investments, which is really built in India for the West or built in India for the globe. Just like we have the Infosys and the TCSs and the Cognizants of the world, a set of startups emerged in the early 2000s—Zoho, Freshworks, then Kissflow and Chargebee, and all of these—all out of Chennai, and pioneered a certain kind of model, which used content-led inbound and addressed through inside sales.

Sitting in Chennai, they would sell them to Mississippi or San Francisco, et cetera. Numbers now are quite large. On an annualized basis, they do about $8 billion, $9 billion of revenue every year. There is a smaller India-only SaaS market, which would be a third of this. All put together, they would be about $12 billion, but really, built in India for the globe, the Freshworks of the world is listed, that’s a larger part.

Within that, there are two streams. The first is the value stream, which is that, “Hey, Freshworks is doing what another company did, but we’re doing it one-fourth cheaper, because we have India sales teams or India engineers, et cetera, similar to the Infosys of the world.” But they don’t stay there. It’s an entire Claytonian disruption whereby they slowly start out, and they keep getting into bigger and bigger accounts. They started doing that. However, of late, a number of very innovative startups have emerged.

Postman is a great example. BrowserStack, LambdaTest, Hasura, these are all very, very innovative startups which could have been born anywhere, just incidental that they were in India, and they’re creating global products. This number is very small, and they use what’s called product-led growth as a way to drive adoption.

I would say SaaS is a very promising and it’s a very evolved category within venture. There are playbooks that have evolved over many years. There is a flourishing Bangalore-Bay Area corridor, by which I mean to say that if, for example, you start out in Bangalore, you know you directly to go to the Bay Area to set up because some of your investors may be there. There’s a very easy playbook now to migrate through that corridor, so to say. There are a set of firms which help you in all aspects.

That part is a very large segment. I don’t have the numbers ready, but I would say anywhere from a third to maybe 40% of venture investments might be SaaS, 25% to about 40% might be SaaS. I’ll have to go a little deeper. I’ll probably give you the exact numbers later. With the second category that you mentioned, that’s an interesting one. I didn’t see it like that. I would say that that is really small, Shruti. Will it remain small? No, it will grow.

Really people using tech to solve uniquely Indian problems, which are not in the consumer space, so to say—I think while there are companies—for example, there is a company in the materials space called Log9. We have a company called Pixxel, which is in the agritech, space tech space. I would say, really, I do think many of them look at selling still to the West.

Unless someone is uniquely focused on selling into, say, India for various reasons, I think many of them using deep tech are still looking to sell into the West, and hence they come into the first category somehow. They may not be SaaS, they may be more hardware-led, but they’ll come into that. I think this is broadly how I see it. The first category is really large, which is built in India for the globe SaaS, but the second category, I think it’s very tiny still.

RAJAGOPALAN: Yes. I see a lot of the second category when I look at Emergent Ventures grants. This is of course very, very early stage. I mean, Blume is too big for us in a sense, and I know you guys enter at a very early stage.

PAI: Yes.

Do Indian VCs Lack Imagination?

RAJAGOPALAN: What I have found—I’ve learnt a couple of things about the Indian venture capital scene, and you can tell me if I’m being very uncharitable. One observation is, it is very focused on growth and scale and not focused on broadening the base of good ideas the way U.S. venture capital is. That is I think one very big problem.

That gap in India is filled through family and friends and Angel, and that kind of money, but we just don’t have enough incubator programs, enough accelerator programs. Y Combinator, even those things come at a later stage. Like, we need a program in every university that will fund at the 2,000 USD level, 5,000 USD level to get 1,000 ideas blooming. Most of them won’t bloom, but that’s fine. I really feel like I work more in that space.

Indian venture seems not that interested in that space. Is it a lack of context? Is it a lack of networks in smaller towns and Indian universities and understanding what’s going on with that kind of innovation? Or is this again a case of elite imitation? Anyone who is credentialed by the West, we are going to imitate that and then scale that. We know the growth and the scale model, so we just pump money into that. How do you think about why the Indian venture scene is not investing in early-stage ideas as much as one would like?

PAI: You gave interesting probable reasons. I think a lot of it has to do with the nature of financing and the fact that when it comes to funds like us, it’s good to see it as a treadmill, so to say. That once we invest in a particular company, in 18 to 24 months, we need to take care of that first level of risk and probably pass it on to the next round of investor, who then takes care of the next round of risk and passes it on to that.

When we look at, for example—and when we come in, it’s not in isolation. We have to understand that, “Hey, 18 to 24 months from now, will they be in a position to have a Series A investor come in?” There, I think, we found out that with many of these ideas, it’s hard, because in India, as investors, we understand market risk, but as a fund, for example, we might say that product risk is hard to figure out.

We probably want to see some MVP by the time it comes to us. When it comes to SaaS, still the feeling is that, “Hey, get three, four engineers and one designer into a room, give them pizza and coffee. And three weeks later, they’ll come out with some working prototype.” But with hardware, how can you predict timelines? In the West, there is, for example, Lux or Khosla Labs. I’ve read about them, and they have a very clear understanding that the product can be solved like this. They understand how product risk can be mitigated. Their challenges are with market risk, maybe.

In the U.S., for example, because the venture industry is so evolved and because it’s much larger, even the niches are large enough to have a certain set of funds. Deep tech funds only do deep tech. They only do space. They only do biotech. They only do, for example, Beyond Meat-type products, so they’re able to take that product risk. In India, I think, fundamentally, Shruti, it comes down to those two reasons. One, product risk is hard to figure out in India. You don’t know how much it will take to have a working prototype. B, we look to people who are downstream of us, and we look to see, will it be ready for them to come in 18 to 24 months later?

Typically, the answer is no. We’ve invested in hardware, we’ve invested in some of these, and it’s taken a lot more time than it should have. As a result, now I think the internal understanding is that if you’re doing hardware deep tech, we want the product to be reasonably ready with at least one customer having seen it, else I think it becomes very hard to time it, to say in 15 to 18 months to raise the next round. I think this is a fundamental reason, but yes, the venture industry in India is large, but large because of growth, the growth capital being large.


PAI: Seed one is still small. When the seed one (right now it’s around $1.7 billion) becomes $5 billion, $6 billion, $7 billion, you will see a lot more players who are at that end, who will look at an idea and say, “Hey, I want to put $50,000 or $100,000 there.” But at this moment, I think it’s a little out there. It’s still not mainstream enough for venture.

Too Many Quants?

RAJAGOPALAN: No, I understand, but the way I think about it—maybe this is counterintuitive, but I feel like because there is so much product risk, it’s all the more reason that all the big venture capital firms in India should have accelerator programs which de-risk, right? You start with incubator, accelerator programs which will only risk $5,000, $10,000. Then out of those 10, 15, 20 ideas, you get one really good product where the MVP is at a stage that can be deployed for manufacturing and so on.

I completely agree with you, but that surprises me even more that that investment hasn’t come in. I have a reason for it, and you can, again, tell me if I’m being uncharitable. I think the reason is that the Indian venture scene is full of quants; it’s full of engineering, MBA, finance-type people. Here I’m being really uncharitable because my perception of them is they wouldn’t know a new, innovative, good idea if it bit them, because they like what is tried and tested, right? “Oh, I know this, this feels like that other SaaS company. This ed tech company feels like that other ed tech company with this kind of a growth story, or this e-commerce can be compared to this other giant.”

They’re more willing to take risks in what is the path that has been walked before. But when it’s a new, crazy, out-there idea, when there is product risk, they don’t understand the product, they’re just not willing to invest because they have not been founders. The Indian venture scene is not full of second-, third-, fourth-generation founders. It’s really just full of quants. Again, is this me being very uncharitable to the engineer-MBA category? Or are there other reasons that I’m completely missing, and I’m focusing too much on the individuals who are saying no to my very dear Emergent Ventures winners?

PAI: What you’re saying is really interesting. I think I would struggle to agree with you there. I would struggle because I do feel that typically the folks who come into venture, and I’m speaking of the folks I know, they’re all passionate about invention. They’re passionate about innovation. They all typically try and do one crazy investment in every fund, as much as they can. There are some very deep tech funds who like to take crazy risks, like deep tech funds, for example, pi, Merak. Blume also takes odd risks sometimes once in a while.

I think the challenge, Shruti, is (I alluded to this earlier) that you can be contrarian, but then you need to be right. Contrarian wrong is, “Really, you invested in that?” Again, if you are noncontrarian, if you’re convention-led and right, then you only get what everybody else gets, right? There’s no alpha there. You’re doing beta, you’re driving the beta wave. We all really make big money when we are contrarian and right. To be contrarian means others don’t agree.

There has to be certain time window of being contrarian involved. You want to be contrarian for about a year, and then suddenly everyone has to wake up and say, “Hey, that’s interesting, and how come I didn’t do that? And now I need to do that investment.” I think there are very few opportunities like that. What happens is whenever you look at an investment, because you know that you are only investing for about 18, 24 months and then somebody else needs to come in, you’re continuously looking at what your peers are looking at and how do they view this.

The truth is, in a way, very outlandish ideas sometimes get weeded out because of this, because you try to second-guess them and say, “It may not work unless it really takes off, and I don’t know if it’ll really take off.” There are occasions when you had to drop certain ideas, but we do—for example, Pixxel in space tech was one. Initially there were two founders out of college, out of BITS [Birla Institute of Technology and Science]. They were students there, and we said no to them. But then after a year—and they went out on the market, they raised a little bit of money. They were trying to build a prototype. When they came to us, we said, “Wow, they haven’t given up.” There was interest from a couple of other funds. Then we said, “Sure, let’s go ahead.”

I think the challenge is, unfortunately in venture, you’re not the only one. It’s not like you do an investment, and after that, no other person is going to invest. Because it’s a staged risk, you are continuously looking at how the person next after you is looking at it, and you’re trying to second-guess on their behalf. That leads to some of the behaviors that you said. I hope it changes. I would love for it to change. Your idea about accelerators is very interesting. I now need to think of a reason why. There are reasons, specialization and a few other things including signaling risk and all that, but let me not get into that now.

Investment Incentives

RAJAGOPALAN: Here I have a question for you because I’m not well versed with the way the incentive structure works within a venture fund. Is this about poor management structures and poor HR structures, where constantly people who are making these bets have to justify to the person senior to them, and therefore it is easier to go with the flow of what everyone else is doing and what everyone already recognizes is the right thing?

The other reason I say this is, a lot of times I find that if someone has gotten a little bit of money from a Silicon Valley fund or like Emergent Ventures or any of the small accelerator programs, then an Indian venture is much more willing to invest in them because they’re like, “Okay, this has been credentialed.” I feel like the only time people rely so much on credentialing is when they have to explain to their bosses why they made a particular move. Their bonuses and their salary and their upward mobility is completely reliant on that, right? Do you think something needs to be figured out within the venture fund, like employment structure?

PAI: Not really. I do feel that venture, the way the incentives are designed at a senior level—not at a junior level, at a reasonably mid to senior level—a lot of the money you make is in what’s called carry. Which is, when the fund ends and then they tot up all of the winners, deduct all of the losers, and you look at what is the total number and then what is the fund size, you deduct the two. And then you keep 20% of the surplus with you and send the rest back to the people who gave you money, who are called the limited partners.

That 20% is sizable. Sometimes it can change lives, and all of us are working for that effectively. The money that we get as salaries, it’s fine, but it’s not excessively generous the way it is in private equity or investment banking or anything like that. I would say incentives are well designed. But I think the challenge, Shruti, is perhaps a combination of not having seen enough outlier hits, the fact that if you invest very early on, product development takes a lot of time.

RAJAGOPALAN: Takes a long time, yes.

PAI: So in 15 to 18 months, you can’t take it to the next investor and get an up round, which at least validates you in the eyes of the boss. So a combination of these two are I think the real culprits. I think it’s beginning to change. We’re beginning to see, like I mentioned about the two funds, which are digital tech funds—so as deep tech funds emerge, B, many of us are working in partnership with universities. We do have interaction programs with universities like IIT Madras, for example. We do interact with them. They have incubators.

Sometimes venture funds also give money. Venture funds have scouts now, and you kind of alluded to it, but you said something very interesting. You said that not enough venture folks are previously founders. In a way, there is this mention of truth to it. It’s changing, and so what you have today, it’s a very interesting concept where there are a lot of founders who today are beginning to invest. Sometimes they are second-time founders who’ve had one exit, but they have some money.

Sometimes it’s founders or even very senior operators who have got some ESOPs [employee stock option pools] sold, et cetera, and they’re beginning to invest, and they are taking bold bets. They are actually taking bets on very outlier things, so what you are alluding to will happen. Will it happen soon? Yes, it will happen very soon. I don’t think it will take a long time to happen, whereby some of these bets happen.

It will never be enough, because there will always be more ideas, more Indian engineers, Indian founders coming out, but it will happen in the trend that you’re seeing. And my only quibble, so to say (and “quibble” is a very nice way, yes?) I don’t think it’s got to do with the internal incentive structures on the HR front as much—so what is the structural nature of the industry, which is the driving factor.

RAJAGOPALAN: No, I completely understand. And the long pipeline for deep tech, hardware, that’s the other part of it, right?

PAI: Yes.

RAJAGOPALAN: We have mastered because we have a 30-year, 40-year head start, thanks to the Infosys, Wipros which were the first gen, then building for banks abroad, then building SaaS products at a global platform. We have a 30-year head start in that industry, in software, in tech. We have a 15-year, 10-year head start in e-commerce. We have a 5-year head start in ed tech, so we are at point zero when it comes to hardware in deep tech. I think that’s another part of it, which is, there is some learning by doing, and we will eventually get there.

I think the other thing that will probably drive this space is the saturation in ed tech, in e-commerce. Once the growth model saturates in these areas, then again, if there’s capital going around, then they have to put it somewhere, and then suddenly you will see a thousand accelerators bloom. I’m really hoping that that happens sooner than later.

How Does U.S. Monetary Policy Impact Indus Valley?

RAJAGOPALAN: Last question for you, and I think you alluded to this right at the head of our conversation when you were talking about how the capital comes into the startup ecosystem in India. How is the Indus Valley scene impacted by U.S. monetary policy versus Indian monetary policy? Are these very big predictors? Have we completely bypassed it because it’s largely private money? What’s a good way to think about this?

PAI: Yes, I would say that’s certainly impacted by U.S. monetary policy. But just because interest rates have gone up, we’re not likely to see a reversal or anything like that because now the dam is truly opened. But I would say the spur for allocating more and more money to alternative assets in India certainly was a zero-interest-rate regime.

I joked that as a venture investor, I think the biggest tailwind behind me was really the U.S. Fed rates and that I should wake up in the morning and do a puja for the Fed governor because fundamentally it’s their policy that’s leading to more and more money coming into India. But that joke was only true to a certain extent because what happens is once the money comes in, once the money finds actors on the ground to deploy, and once they start deploying and some returns start coming back, then people there believe that there is more money to be made. And now just because rates are going up, they’re not going to say, “Hey, let’s just stop.”

Of course, it also now helps that China, for example, is persona non grata in the U.S. context, so they are looking at India to a great degree. But returns, exits from India have started happening. While yes, the spur was certainly ’13, ’14, ’15 we saw, for example, zero-interest-rate regime country, that really opened the floodgates. But now that it’s been on for the last six, seven years, we’re unlikely to see it going back.

If I take, for example, Blume’s case itself, we started out as a $20 million fund in 2011. We nearly did a $300 million—$290 million is the exact amount—and the interest that we’re seeing now with the U.S. folks reaching out to us, I think there’s fundamentally a sense that things have changed. And I talk about this line that it’s India’s moment, and for various reasons, I think that’s true, and all of those trifectas of factors that I alluded to are coming together.

I think, yes, U.S. market policies will probably evolve, and I can’t see it going up very much more, but let’s see what happens. As far as India monetary policy is concerned, I don’t think it plays much of a role, Shruti, to be very, very honest. The RBI is a very, very important actor, especially as fintech goes, because they—

RAJAGOPALAN: They control the approvals.

PAI: Yes, and many other factors. They change policies. There’s a joke that, “Scare an Indian fintech founder in three words,” and you say, “New RBI policy.” But I think Indian monetary policy has no impact because bulk of the money that’s coming in is coming from the West.

RAJAGOPALAN: Coming from abroad.

PAI: And while of late, a small Indian family office, a set of contributors and investors are emerging, still are not comparable to what we get from the West. So that is really my answer.

RAJAGOPALAN: Yes. Hopefully that changes, and there’s less crowding out by the government of India investment, and there’s more money coming in from India to develop products for Indians. I’m particularly hopeful about that because if India can manufacture, keeping in mind developing country problems, infrastructure problems, then not only is India making for itself right now, it’s also making for sub-Saharan Africa 60 years from today. Because Africa is going to add a couple of billion people in the next century, right?

PAI: Yes.

RAJAGOPALAN: It’s going to be the next India, so if we can crack that problem for ourselves, we’ll also crack that problem in the future for the world. The manufacturing, the infrastructure problems—at least I’m very hopeful that that matures in India over the next 10, 15 years.

PAI: Well put. I was in Egypt about six months back, and I met a bunch of Egyptian founders, and they all told me this, that “we actually study India”—perhaps much the way we study Valley—“that we look at India and say, ‘India has been here. India is five years ahead of us. What happens in India will come to us five years later.’” You’re right. India has been seen as a champion for much of the Global South, and yes, so we do have a responsibility. You’re right in that degree.

RAJAGOPALAN: Yes, and if we can make money off of fulfilling our responsibility, what can be better than that? That’s the loveliest thing about what’s happening in the Indus Valley ecosystem.

Sajith, I can’t thank you enough. Thank you so much for coming on the podcast and for the report that you guys put out every year. I can imagine how much work that takes, given the lack of aggregate data in India at a granular level. And hope to chat with you again soon. This was a pleasure.

PAI: Thank you. Thank you, Shruti. I really enjoyed this. Thank you for having me on the podcast.

About Ideas of India

Host Shruti Rajagopalan examines the academic ideas that can propel India forward. Subscribe in your favorite podcast app