印度下一代科技巨头为何近在眼前、却常被忽视 | Paavan Gami

Investing in Carvana, we didn’t own the stock at the top. We thought we were smart and sold. Then a business we really liked became available at a very, very attractive price. The stock was already down 90%.

So you sold at, say, around $160 or something like that?

I think it’s very hard to run into a burning building if you don’t already know the structure of the building.

Today, we’re doing something a little different, which I’m very excited about. My guest is a liquid fund manager rather than a venture manager, but I think you’ll see pretty quickly why it makes sense for this show. Paavan Gami is the founder and managing partner of Ras Partners, a concentrated, long-term-focused Indian public equities fund.

Before Ras, he spent 5 years at Greenoaks, one of the truly legendary firms of the last decade, famous for extremely concentrated bets, like putting roughly 40% of its first fund into Coupang, a bet that returned about $8 billion. It’s also known for holding its winners long after the IPO instead of exiting the way most VCs do. Paavan sat in a very interesting seat, working across both venture and public markets. That crossover DNA—venture-style underwriting applied to public markets—is exactly how he invests today.

He left one of the best seats in investing to bet everything on one idea: that India’s great tech companies go public at a few billion dollars while they’re still basically Series C businesses, and that the people buying them don’t yet know how to read them. His ambition is to own the Indian FANGs, the handful of companies that will define that market for the next 20 years, through what he calls “inevitables”: founder-led businesses with moats so deep their dominance feels basically inevitable.

He’s also one of the deepest researchers I know, and I’m really excited to dig into Greenoaks, the Carvana story, the India bear and bull cases, and how AI has changed his thesis and his process. We had some interesting chats about that last time. Full disclosure: I’m an LP in his fund. Very happy to have you here, Paavan. I’m excited to get into it.

This will be really fun. Thanks for having me.

To start, let’s talk about your background, because I think not many people will have heard of Ras yet. I’d love to hear how you got here, your time at Greenoaks, and all of that story.

The influences around investing for me were a bunch of different threads coming together, which resulted in a pretty investing-obsessed kid at age 17 or 18. Part of that was that my family had an operating business in India. I was born there; we’re from a tier-two city called Baroda. I moved to the U.S. and mostly grew up as the son of middle-class immigrants.

I think there’s a culture that comes from being part of a “business family.” It’s kind of an Indian idea, I suppose, and it resulted in me being pretty interested in business from a young age. On top of that, my grandfathers on both sides spent time in the markets. One of them retired at age 35 and traded equities for the rest of his life, keeping his household afloat that way. The other, after selling the business, spent all of his time trading too. I call it trading, not investing, but it gave me some exposure to markets at a pretty young age.

I went to Yale as an undergrad and dabbled in a bunch of different things. I was a pretty mathy kid, and at first, margin-of-safety-type, pick-up-50-cents-on-the-dollar investing was intuitive to me. I spent a summer doing arbitrage-y convertible-product-type stuff. It was stimulating; we can talk about that if you’re curious, but it didn’t really click for me.

Then the version of investing that was much more humanistic, much more about people and cultures, clicked for me. Those are the machines that make the machine, with the machine being a competitively advantaged business. As an undergrad, I read some investment letters that approached investing that way. It wasn’t really Buffett; it was more the Nomad letters. Those blew me away and set me down this path. That’s painting with a broad brush, but that’s how I got into investing.

I worked at a fund called SPO Partners in San Francisco, which at the time was a very storied firm. It had been around for 40 or 50 years and started as a private equity firm, doing a bunch of creative things. One of its most famous investments was buying a bunch of timber in Washington State in the ’90s. It was also doing cable roll-ups in the ’90s. After the dot-com bust, it started making investments in public markets.

By the time I got there, it was mostly focused on public equities, but it had the DNA of a private-market firm: find a business you can own for a really long time. As the youngest person there, I became its first internet analyst. I spent time on large-cap tech and quirkier tech, and the way I got to know Greenoaks was through our work on Naver, the dominant South Korean search engine. We had a hypothesis that its version of Google Shopping could become the dominant e-commerce business in South Korea. At the time, it was competing with Coupang, a Series B startup.

We were doing risk diligence on Coupang and met Neil and Benny. Greenoaks was a very young firm then, maybe 3 or 4 years after being founded, with 3 or 4 investors. I was just blown away by them. At some point, SPO returned capital, and I returned to the family office. Most of my colleagues went on to join public-equities-focused, Tiger Cub-type firms. That conversation with Neil and Benny rang in the back of my mind. I just think the action, if you’re going to be a technology investor, is very obviously in private markets. I wanted to try it, so I ended up at Greenoaks in late 2018.

I’d love to hear more about the Greenoaks period.

Yeah.

It’s kind of become legendary, probably since the Colossus profile. I think that really put them on the map. For me, it was the first time I really knew the lore. They’re famously low-profile, and I’m curious what you think makes their approach different. What did you learn there, what did you decide to bring to Ras, and what did you decide not to bring?

I think it’s a really special firm. I’m biased, of course, but it’s a place where I grew up as an investor. It’s an investing system, and many parts of that system are different from the way you’d typically think about a venture firm. Venture firms are typically a federation of GPs. You have 6 or 8 GPs, divide the fund by 6, and have 6 mini-funds that roll up into a shared marketing function. That’s typically how venture firms organize. Each of those GPs has around 8 bets, so you end up with a pretty diversified fund.

Greenoaks is sort of the opposite. It’s ultimately a single decision-maker. Your conviction largely comes through Neil’s point of view on a founder and a business. The volume is very low: they make a couple of investments a year, and they’re pretty opportunistic about what those investments look like. They can be very early-stage investments, hundreds of millions of dollars into a late-stage company, public-market investments, or, if the circumstances require it, something more structured.

It’s about concentration, a focus on their perspective on what an exceptional company and an exceptional founder look like, and a lot of flexibility and creativity in the mechanism of investment. You could describe that as a crossover fund, but spiritually, I think it feels pretty different. A crossover fund typically has a growth equity fund and a public markets fund, plus a shared marketing function. Greenoaks doesn’t operate like that at all.

I think the firm is the first-principles creation of a vehicle that allows Neil and a couple of the other people there to be the best investors they can be. That was something I took away when I thought about designing Ras: Who am I as an investor, and what are the ways to amplify my strengths? That translates into selecting an opportunity set.

That translates into a process-design piece of things. Eventually, it’ll translate into an org-design piece of things. But, yeah, I think there’s a lot to learn from Neil on that front.

For sure. I want to get to that.

Yeah.

Maybe one last question on Greenoaks, which I’d be remiss not to touch on: the Carvana story. That’s one of the most epic stories in the Colossus piece. You were buying this position in public markets, watching it go down 80% or 90% in your face, and buying it all the way down. I know you were pretty involved in that whole thing. Maybe you could tell that story from your perspective: what happened, what it was like, and what you learned from it.

Yeah, certainly one of the most formative things in my career so far. We had made investments in used auto in private markets before investing in Carvana. The premise is pretty simple: this is a very broken consumer experience. We talked about the lemon problem. Used-car dealers are selling you a car, but you don’t know the underlying quality of the vehicle, and it’s a pretty important, expensive asset for the person buying it. You can improve the customer experience by giving a guarantee and investing in the vehicle.

But there are also a lot of scale economies if you build a vertically integrated system instead of a distributed system. The used-auto models in e-commerce tend to be auto factories that refurbish cars, procure parts at massive scale, and run logistics themselves. It’s very different from a 20-car lot.

Ernie Garcia, the founder of Carvana, was clearly the most special entrepreneur in the category. He invented the category. If you spent time with him versus the dozen or so entrepreneurs executing on the same thing in different geographies, Ernie would stand out.

We were investors from mid-2020 to the end of 2021, and then we sold. I thought we sold pretty well—not quite at the top, but maybe 20% or 30% from the top. The valuation had just gotten expensive.

Then a bunch of things happened. It’s a very cyclical business. Demand is cyclical, and it’s also rate-sensitive: your customer is buying a car and financing it, so if rates go up, their monthly payment goes up. From a P&L perspective, Carvana is financing all of its inventory, so when rates go up, its interest expense goes up. They had also made a very large acquisition, fully financed through leverage, around January or Q4 of 2021, before rates went up.

A bunch of things fell apart: demand went down, they had a lot of fixed costs, and they had this large fixed-interest-expense burden. The business went from being unprofitable but burning a manageable amount of cash per quarter to burning, I think, $150 million to $200 million in cash per quarter. There was a period when it looked like it had 3 or 4 quarters of runway, and it was increasingly priced for bankruptcy.

We didn’t own the stock at the top, so we thought we were smart and sold. Then this business we really liked became available at a very attractive price. I can’t remember the exact details, but I think we built a position at around $20 per share, in the $20s. The stock was already down 90%. Over the subsequent 6 to 8 months, it went down another 75%.

That’s crazy.

You think you bought well, but you still have a 75% drawdown.

So you sold at, say, around $160 or something like that?

Yeah.

Then it went down to around $20, and then it went down to $4 or something like that, right?

In that range. There were a lot of things from a process perspective that I thought we did well to be able to pull the trigger even in the $20s. I think it’s very hard to run into a burning building if you don’t already know the structure of the building. If you’re totally new to the story, there’s enough hair from a balance-sheet perspective and other risks that you wouldn’t have been able to get there. But this was a business we thought we knew pretty well and had a prepared mind around, so we could roll up our sleeves and feel like this was an underwriting challenge we could get over.

There were a dozen bear cases, some realistic and some verging on conspiracy theory. We tried to seriously consider every single one of them before putting capital at risk. I think the biggest question mark—

Yeah, what was the biggest question mark? Did you think there was a chance that this business would go bankrupt?

I’d say there was a balance-sheet question: it was priced for bankruptcy, but was it going to go bankrupt? What was strange was that after all this had fallen apart, the company wasn’t cutting costs very much for 2 quarters. The burn wasn’t going down nearly as much as you might model in Excel. The logical question is whether it was impossible to cut costs, and therefore whether this was a sinking ship.

I think that’s why the stock went from $20 to $5 a share. Internally, they had so much confidence in their ability to pivot on a dime that they were focused on a bunch of structural solutions instead of just cutting headcount. An important case study was COVID in 2020. It really felt like the world was going to end, and between March and May 2020, they cut their fixed costs dramatically. You could look back at their disclosures and see that when put in a corner, they did have a lot of control. Despite not seeing evidence of that control over those 2 quarters, we felt that if necessary, they could do it.

From a balance-sheet perspective, we had to spend a lot of time thinking about who the creditors were, what their incentives were in the situation, what leverage the company needed to restructure the debt, and what that could look like. There were a bunch of balance-sheet-oriented questions.

But I thought the more fundamental question was whether the demand was real. You can grow an e-commerce business, or any consumer internet business, very quickly if you subsidize the unit economics. The question is whether the demand will stay real at sustainable unit economics. In this case, it turned out there wasn’t much subsidization going on, meaning they were selling cars at market prices rather than 10% below market prices. You can sell Infiniti cars 10% below market prices, right? Really pushing ourselves to verify that they were selling at a price that made sense, and that the unit economics could get to a reasonable place in terms of cost to serve, was a big question mark.

There were also questions about the financing business they had built and its sustainability. I think that again verged into conspiracy theory, with a lot of worry about subprime lending. If you really dug into the ABS filings, it turned out that Carvana wasn’t any more subprime than its competitors. There were details you had to unpack. The competitors sold their subprime loans wholesale, while Carvana kept them, so it looked like Carvana’s mix was worse.

Long story short, there were a lot of thorny questions to unpack, and the only way to unpack them was to do the work. It was the only thing—

Mm-hmm.

—I spent that year doing, more or less, beyond my board commitments on the private side.

Interesting. So it was a year-long process. Did you buy a small position and then buy more as you got more conviction? What was that like? I’m also curious: it sounds like there were a lot of different questions and risks for the investment. What did the investment committee look like when you discussed this, assuming you had an investment committee? What was the simplest thesis for saying yes? Or was there no simple thesis? Was it a risk-by-risk, probability-weighted assessment, or was there a simple thesis you had that said, “This is why we’re pulling the trigger here”?

Well, I think the simple thesis was that you had category-killing retailers and an amazing entrepreneur available at about 5 times equity to free cash flow.

And so you kind of didn’t need to do work on what 2030 or 2027 earnings were going to look like. It was kind of like, if they weren’t bankrupt, you were going to do well.

We make money.

Yeah.

Yeah.

And then act two is: if they’re not bankrupt, how big can this be? What are the unit economics? What are the long-term, sustainable margins? We had done much more work on those questions before the balance sheet risk, so we already had strongly held opinions, but wanted to pressure-test them.

I think the committee—it’s a pretty small, flat firm—was doing a lot of work on these things, relaying the answers on those topics on a pretty real-time basis. It was trying to build a deeper perspective on Ernie as an entrepreneur and on the culture of the company. Some of that also involved Neil figuring out ways to spend time with him and learn.

It’s a scary time for a company. Employees can leave. How do you retain them and maintain morale? There were many of those kinds of topics to work out.

In terms of how we traded, I don’t remember the exact details of the timeline. It’s sort of a blur at this point. We built a position in Q2 and then added to it later in the year. I think there was probably a de-risking moment after the debt renegotiation, when they turned the interest expense into PIK interest and the bankruptcy risk was kind of off the table.

The stock had gone to $30 or $40 a share. I think that was probably the best risk-reward over the saga. We’d already deployed enough capital that we added a little bit around then, but didn’t add a lot more. One reflection I’ve had is: could we have saved a bullet? But it’s very hard to know. You’re making decisions in uncertainty, and I think overall we did a pretty good job.

Yeah. You definitely did.

Yeah.

Let’s move to Raas now. You have one of the best seats in investing. You love the culture and what you were doing at Greenoaks. Why did you decide to leave to do an India-focused public equities fund by yourself? What was the thesis?

1. Why Bet on India

I think it goes back to what I was alluding to earlier. People talk about founder-market fit. You could use that rough analogy for an investor, too. There’s an opportunity set, style, and process that fit each investor well. As I matured, I started to have a sharper opinion about what that could look like for myself, and I wanted to build a firm that would allow me to execute at a high level and with duration.

I wanted to make a 20-year bet on something, and then the question was just what. You and I have talked a lot about the unique things with the India opportunity set, but I was really excited about it as a fun pond to fish in as a stock picker. I’m also from there. I’m a believer. I think you have to be a believer in what you’re investing in, and I very much believe in India.

It’s also a fun place to be. The people are inherently optimistic about the future, rationally or irrationally, I’m not sure. But Indians believe their best years are ahead of them, and that’s also part of why I wanted to do this.

And what is the bull case for India? For everyone listening, I think people are broadly bullish on India. The demographics are great, and all this stuff. There are some challenges, too, but I’d be curious: what do you say to LPs and everyone else who asks, “Why are you doing this?”

Yeah.

Well, yeah.

I’ll say a couple of sentences about the big-picture bull case for India. I’m more excited about the stock-picking bull case—why it’s a place where you could have a lot of outsized winners for structural reasons.

On the India bull case, you have 1.5 billion people, half of whom are under the age of 30. Just process that: it’s 20% of humanity and 4% or 5% of global GDP. There are a lot of institutional problems. The rule of law could be better, and there’s a populist-ish thread in Indian politics that results in bad policymaking sometimes, in my opinion.

But despite that, real GDP has grown 6–7% a year for 25 years. Think of it as GDP growth with ankle weights on, and I think the direction of travel on policy is getting better and better. Markets are becoming freer. India liberalized in 1991, not too long after China, so it’s really like a 30-year-old country. I’d say it was half liberalized back then, and then you have two points of additional liberalization every year since. That’s a rough mental model, and it’s exciting.

On the stock-picking front, which is what I’m more interested in, you have such an immature country. Take any category. I spend a lot of time in consumer, so we can use consumer examples: pharmacy, home improvement, retail, QSR, hotels, hospitals, grocery, auto parts. The vast majority of the supply side will be mom-and-pops. Ninety percent of them will be independents.

There are very strong economic forces for why that won’t be the case 20 or 30 years from now. The merits of consolidation are very clear. The only reason consolidation has been slow historically is that there are regulatory diseconomies of scale in India. If you’re big, you have to operate compliantly, pay people not in cash but through bank deposits, which means you have to pay Social Security taxes. It also means you probably can’t accept cash from your customers, and so on.

But those diseconomies of scale are going away as the regulatory regime gets better, and even the mom-and-pops are being forced to comply. From a stock-picking perspective, you have these big categories growing with Indian real GDP. In the 10% of the category that isn’t mom-and-pops, I think a leader has likely emerged. That leader might have 5% or 6% market share, and we’re betting on that leader going from 5% market share to 30% or 40% over time.

We know from mature geographies which kinds of categories have those kinds of outsized winners. In building materials, for example, you know there’s something special about Home Depot. Maybe you could find a Home Depot in India. That’s one interesting thing about stock-picking.

On the technology side, which you need to marry with the shift from unorganized to organized businesses, there are a bunch of technology businesses that I think are the dominant companies in the country, or will be, that have just come public. They’ve come public while margins are still negative and they’re growing pretty quickly. I think Indian public markets don’t really know how to value those things.

You’re able to buy these super-durable FANG-type businesses at what I think are 15–20 times long-term earnings. That stock-picking opportunity—the first bucket and the second bucket—is what I’m really focused on. It’s something you can’t get in a mature market. Home Depot is already at 40% market share in America. All the growth in public-market investing in a mature market is in a technology business taking market share. You don’t have this organic growth piece that you do in India.

Yeah. I want to touch on both of those: the macro thesis for India and the stock-picking—your edge within that. Maybe starting with the macro, I’m really curious to get your thoughts on this. We spoke about this a few weeks ago.

2. How AI Reframes India

A lot of the bull case for India was 1.5 billion people and some of the best demographics in the world. I think one question a lot of people have is: does AI actually invert this from a blessing to a curse? With automation and stuff like that, if you’re really AGI-pilled, population is just sort of a burden that you have to support, right? Because they’re not going to be able to provide meaningful value, assuming everything is automated or whatever.

Obviously, that’s an extreme take. The other side of things is that IT outsourcing is such a huge part of the Indian economy. I think it’s 8% or 10% of the Indian economy, right? Leopold famously has a short in an Indian IT outsourcing company.

There’s a notion of the post-scarcity sci-fi future where the population is dead weight. The way I parse that world is: whatever the distribution of capital is in today’s world, the return on capital goes up a lot. If you have a low starting point of capital—and India has a low starting point of capital and a low starting point of fixed assets—you’re going to be left behind. So I think that’s a tough scenario for India.

But in most non-sci-fi worlds, it should be a place that really benefits. Education is poor. If you have something like The Diamond Age: Or, A Young Lady’s Illustrated Primer—and hopefully Alpha School is creating that—who will really benefit from it? A place where the baseline of education is really inadequate. The human capital that can be unlocked in a place like India is significant, even at the primary education level.

Who does a coding agent make better the most? I’d say an existing San Francisco 10x engineer becoming a 100x engineer is one piece of that. But a junior developer in India who can now deliver much more in an IT services context could potentially see a higher revenue-per-headcount uplift than someone in the American context. That’s speaking very abstractly, but I’d be skeptical of the idea that all these people are permanently left behind, rather than being able to use some of this stuff for the better. That’s point number one.

On IT services, I have some quibbles about the cultures of those companies. We’re not shareholders in any of the Indian IT services companies. But if I have a worldview about software, it probably looks like much more custom software being built in the world, rather than software vendors selling to enterprises. Every enterprise has its own workflows and proprietary data. You should home-grow your own systems, and I think people are increasingly starting to do that. The FDE motion is kind of that.

The challenge of whether an enterprise will be able to hire great technical talent will continue. So the spend in IT budgets is probably going to shift from spending on Oracle to spending on a guy to build your own ERP, I would think. And the COGS of paying that IT services guy to build your own Oracle are going to be largely tokens. So there’s an IT services bull case that’s still kind of a bear case for employment. I’m reasonably worried about that. But the IT services companies could evolve in a way that lets them survive. That’s how I think about it.

The most important sort of defense, I think, is that our portfolio is trying to buy businesses where a lot of the return comes from market share. I don’t think Indian GDP is going to fall off a cliff, but whether GDP grows 7% or 4%, the grocery business I own—an e-commerce grocer that I think will end up becoming the Costco of India—has 2% market share today. I think its growth is totally in its control and not so much a function of what happens to India.

I’m curious how you parse that. I think if you’re pilled enough, there are only a handful of assets worth owning in the world. But if you’re 90% pilled, then actually a lot of value will be created outside of the AI supply chain. The question is: are you 90% or 100%?

Yeah, I’m probably 80% or 90% pilled. I still think there’s a lot of amazing talent in India. It’s just that AI probably made me less bullish on India, and I’m curious how that affects you, because you started Raas at an interesting point, before AGI-pilling was a thing.

Yeah.

Before all this happened. I think it was around 2022 or something, just as ChatGPT came out, and then we started seeing scaling laws and all this became mainstream. I’m curious how AI changed your thesis on India, if it did. When you went in, you had a thesis, and then as you became more AI-pilled—I know you’re pretty AGI-pilled yourself, and you’ve spent a lot of time in software land—how did it change your thesis on India?

At the bottom-up level, it changed what kinds of companies I’m spending time on in a big way. There are businesses I think are going to do just fine. They’re physical-world businesses. Said differently, I’m not worried about them being disrupted by AI, but I don’t think their futures have accelerated because of AI.

Largely, that’s a function of culture. The things you need to do and the capabilities you need to have to be the best discount jewelry retailer aren’t necessarily capabilities that will translate into home-growing a bunch of software. I’m of the view that if your future isn’t accelerating because of AI—and that largely means using AI to build a lot of technology in-house—then it’s probably not interesting.

There are a bunch of traditional businesses I was excited about. If you read the launch letter I wrote about Raas 2 years ago, a good chunk of it was talking about those companies. Maybe we’ll revisit them later, but the opportunity cost of investing in companies that are accelerating is too great to ignore.

The second change is more of a learning thing. I think of myself as a pretty good consumer and investor, and I have a lot of intuition for what a great founder looks like in that world, and what a business in the zero-to-one, one-to-ten, and ten-to-100 phases looks like. I don’t have a lot of intuition about industrials, for example. I’m also probably constitutionally a bottleneck investor, a bottleneck-stock skeptic. But I think there are some exceptions.

So the challenge I’ve set myself is to learn how those businesses work, learn what greatness looks like in them, and then over time potentially have capital at work in some of those ideas. We’ll go slow, and by going slow, maybe we’ll miss some of this. But that’s okay. I think the goal is to learn and put ourselves in a position to find businesses exposed to what’s coming.

I think there’s going to be an inference data center buildout in India that’s pretty exciting, which may be worth talking about. Power costs are low in India, land costs are low in India, and there’s a lot of skilled labor. If you want a political alternative to post-midterms America and you’re thinking about where to put up compute, I think India is a pretty compelling place. India and Australia, maybe. So we’re trying to find businesses exposed to that. I don’t know if 18 or 24 months ago I would have been spending time there. Those are the bottom-up changes.

Yeah, you asked how I see it. I think AGI, both cognitive and physical, and its coming are the most important force happening in the world. In general, you want all your investments to be somehow feeding AGI, whether you’re a bottleneck investor or you’re investing directly in businesses that can accelerate themselves with it. I think there are a lot of ways to play it.

But the discount jewelry business or whatever becomes less interesting to me. I can see a thesis for things that don’t change with AI, but I’m more interested in things that benefit from AGI, or in hedges like defense, commodities, or things that hedge against geopolitical and other risks. To me, it feels like India is an interesting challenge, depending on how fast this goes. And I’m obviously an investor, so I’m bullish.

Like you said, in a totally sci-fi world, it’s sort of the worst place to be, actually, right? In some sense, you have the most people, and they’re the youngest, so you have to support them through this. If these people can’t be productive, you have a pretty bad situation. You can also look at debt-to-GDP and the amount invested in government services.

I’m not personally a believer that that ever really happens. I think if it ever happens, it’ll be decades away. So that’s why I’m kind of—

Yeah.

—more bullish. Yeah.

And then, two quick thoughts. In a macroeconomics 101 kind of framework, you could say that in your production function, if you have labor and capital, so long as you have labor in that production function and you believe the productivity of labor is going to go through the roof with this stuff, that’s a world in which India benefits, right? You have workers upskilling and so on and so forth.

At the limit, if there’s no labor in your production function at all, then it’s a bad world. But that’s a bizarre world to contemplate. So long as labor and people are still useful—and I think they are—even in the context of robotics, the break-even cost of replacing an Indian driver with an autonomous vehicle is pretty far away, as an example. So that’s thought number one.

Thought number two: on the discounted jewelry example, I also think we place too much emphasis on the business model versus the founder and culture. One of the exciting things about India is that it’s a young economy, so any business you look at, even in the traditional world, is a young business, often still run by an entrepreneur aged 40 to 60.

Imagine Sam Walton, imagine Walmart is 20 years old, and Sam Walton is running it. I think Sam Walton would be pretty AGI-pilled, and Walmart would be deeply leaning into using this technology in all parts of the company. I think there are traditional-ish businesses in India that are themselves AGI-pilled. Maybe the density of AGI-pilled traditional businesses is higher in India because the entrepreneurs are younger. Some of these traditional companies have come out of the venture ecosystem, for example. What you kind of lack in frontier labs, I think you’re going to gain in traditional businesses that are more pilled in India than their grown-up peers in America or Europe. And so—

Yeah. No, that makes sense. I think that makes sense. I want to move on to stock picking—

Yeah.

—but on the capital-labor thing, I think the real AGI-pilled case is where capital just generates labor, right? There is no labor because capital can literally just generate it. The server racks generate labor with very little need for humans. I’m not personally a believer in that world.

The other challenge that’s become more of a challenge with AI is that the world has recentralized around San Francisco, at least in the tech world. Abhishek, who we both know, the founder and GP at Grad Capital, recently told me he thinks the window to start a generational technology company in India is closing, that there’s maybe one year left in his “nerds thesis.” This is the idea that India has super-smart technical talent, but not enough of it.

If you skim the best talent off the top and they all have to go to San Francisco because that’s where the compute is, where the customers are, where all of that is, then you’re left with still very interesting businesses, but not the trillion- or $10 trillion businesses that AGI might generate.

I think there’s a different idea about causality there, but I think empirically that’s true. One way to think about it is: How many exciting late-stage startups are there in India? How many private companies worth $1 billion to $3 billion are there? I think it’s a pretty low number.

There’s a really exciting set of companies worth $3 billion to $10 billion, and one very large one is worth $25 billion and is in public markets now. They have distribution scale, cash-generative core businesses, and great talent density. I think those companies will end up being internet conglomerates or technology conglomerates that win a lot of the new TAMs that emerge, including the AI TAMs in India. The Sierra of India is a subsidiary of one of these conglomerates, right? So there’s that piece. But then the part that I—

That’s true.

The part I push back on is early stage. We don’t invest in it, but I hang out with people in that space, and there’s a lot of energy. That early-stage energy looks more like defense, aerospace, and biotech. I think the founders motivated to execute on those ideas are also partly motivated by a nationalist impulse: they want to build for the country.

If you’re trying to build an AI-near-the-frontier business where you think you need to raise hundreds of millions of dollars over time, it’s just very hard to do that in India. I think traditional venture firms, for the most part, are somewhat cynically just investing in Indian founders in San Francisco now. But I don’t think that’s the only thing to do in the world.

When you think about terminal value, these early-stage companies are doing stuff with more terminal value than building a Lovable competitor. Those businesses will eventually mature into my opportunity set, into my TAM. That’s the way I parse it.

I like that. I think now let’s move to the stock-picking side, because whatever you think about India, there’s clearly a big edge on the stock-picking side. When we spoke, you gave me a lot of good arguments for that. I think you told me the median IPO valuation—actually, I’ll let you say it because I don’t want to butcher the number—but it’s very low compared to the U.S.

You can basically invest in Series C; you can do growth investing as a public-markets investor in India. And there are these misunderstood tech businesses, where the earnings multiples look very high because they’re reinvesting to grow, and investors in India don’t necessarily understand that. I’d love for you to dig into all the factors that make your edge as a great stock picker in India so much bigger.

I think there are many layers if you think about structural mispricing, and then you hope to find a great business and an exceptional entrepreneur that intersects with all those sources of mispricing. That’s a great setup.

One is the thing you were just alluding to. This is a two-year-old stat, but the median IPO has a $350 million market cap, and the founder or founding family owns 70% of it at IPO.

What that basically means is the private capital market is pretty underdeveloped, both venture and more traditional private-equity-type stuff. Companies are just entirely bootstrapped. Maybe you raise some debt capital if you need to build a factory, and the first equity capital you ever sell is at the IPO. The family continues to execute capital efficiently into perpetuity, and the companies are run in such an aligned way because the family has such a high stake.

On the tech side in particular, I think the American idea of staying private for longer is the opposite of what happens in India. Often, if you want to raise $100 million or $200 million, it’s easier to do that by going public than by raising your Series D. That’s only become more true as venture balance sheets focus on these Indian founders in America.

Category-leading technology companies with a relatively small absolute market cap will go public at a $600 million to $1 billion market cap, get a pretty good reception, and have analysts cover them. Then all the compounding from there to a much bigger number happens in public markets.

It’s so different from America. A second thing is that people have a lot of reference points from the Nasdaq. You would think the index is a very good passive instrument, but the index is not a good passive instrument in India. If you look at index constituents, there are tons of public-sector businesses, low returns on capital, cyclicals, metals and mining, oil and gas, and so on. There’s only 1 technology business among the 50 biggest market caps in India, and it’s around number 40 right now. If you roll forward 10 years—

That’s a crazy stat. I think both of those are crazy stats. The average—or median—IPO valuation being $350 million, which is kind of a seed-round value. There are new labs raising at multiples of that in San Francisco.

Yeah.

And what you just said about 1 out of the top 50 companies in India being tech companies is an incredible stat for sure.

Yeah, it’s 1 out of the top 50, 3 out of the top 100, something like that. The magic of technology is that these are increasing-returns-to-scale businesses, right? They can and will compound at high rates for the next 5 or 7 years. Then you have the index-inclusion and liquidity-flow benefits that are all to come. That’s not the reason I invest in these things, to be clear, but in terms of the structural setup in India, that’s what’s really fun about being a technology investor there.

Yeah, that makes a lot of sense. And it’s also a smaller community, right? You mentioned you can actually go to dinner with these founders, meet them, develop these relationships, and have a very good founder view that you’ve honed through venture and through Greenoaks, and really apply it to the public markets. I’m also curious, because you’re very research-oriented, as you have to be as a public-markets investor: how do you weight the founder, and what do you look for in these public-markets founders?

Probably the biggest evolution in the way I think about investing over my time at Greenoaks is this: it’ll sound obvious to you, and Buffett—like everybody says it—but to really internalize that the value an exceptional entrepreneur creates versus the value a merely great entrepreneur creates is night and day. You have to believe that, and then you have to believe that you can do real work to differentiate the two. That’s a research workstream. Some of that research is vibes and pattern recognition, and some of it is doing tons of reference checks.

Ultimately, I think the point you were alluding to is earning the right to spend a lot of time with them, and not just with them, but with the N-1 leaders in their organization. You do all of that to triangulate and get a perspective, and it’s very important. Our goal is to marry a handful of businesses and probably start by being a great thought partner to their founders. Then, as the firm scales and we build a high-quality capital base, we can become a high-quality capital partner to these businesses.

Dreaming a bit, but also thinking about the future: at some point, these internet technology conglomerates will want outside capital to mark a bunch of the businesses they’re building. Silver Lake got to invest in Waymo. Google didn’t need Silver Lake’s capital, but they wanted to incentivize the Waymo team with Waymo equity, not Google equity. I think that’ll happen in India, and we have a shot at doing that over time. But that starts with the work today around being a high-quality thought partner. That’s worldview number one.

I think the bar for the kind of companies you marry needs to be extraordinarily high, and “marriage” is a strong word. When you marry a business, you give it the benefit of the doubt. When something goes wrong, you have enough conviction in the founder and the culture to lean in and buy more in those circumstances. I think of it as: we’ve married 2 businesses so far in the 2 years of Arras, and that might be our hit rate. It might be 1 or 2 of these a year that we find. Everything else below the very top of our book is creative destruction that we’re trying to iterate through, through things that are great in case they’re kind of exceptional. I don’t know if that makes sense, but that’s the way I think about the system we’re building.

3. Finding India’s FANGs

Yeah, it makes sense. Your goal is to find the Indian FANGs, as you once described it to me, and own them through their compounding. Who do you think—maybe we can go into specific companies now—that you’re excited about and think have the potential to be that, and why?

Not investment advice.

Not financial advice. Yeah, yeah.

Not financial advice. I’m biased; these are things that we own. But we own them for the reasons you described. There are 2 in our portfolio today, and maybe 2 or 3 other candidates that we don’t own because they’re expensive or for other reasons. It’s a pretty short list.

One that we own is a business called Eternal, which is a punchy name. It used to be called Zomato, the dominant food-delivery app in India, and has evolved into the leading grocer in the country. It’s doing a bunch of other stuff. The second is a business called Meesho, spelled M-E-E-S-H-O, which is a dominant value-commerce business in the country. Both are e-commerce-ish. Should we dive into one of them? What’s the best way to chat about them?

Yeah, I think so. Maybe we can start with Zomato—

Yeah.

—or Eternal, which, in people’s heads, is sort of like Deliveroo or something like that.

Maybe, yeah. I think DoorDash is the best analogy. The history of this company is that they started as Yelp in the late 2000s. They were the food-classifieds company and won that category. I think the founder is the best entrepreneur in the country: very ambitious. In the early 2010s, he realized this food-classifieds business wasn’t going to be very large. Food delivery hadn’t emerged as a category yet, so he tried to expand into food classifieds all over the world. They acquired Urbanspoon, if you remember that brand in the U.S., and at some point they were in classifieds in 25 countries.

It’s actually the Portuguese classifieds. I use Zomato a lot, weirdly, and I knew of Zomato because—

Is Zomato currently the Portuguese classifieds?

—of the Portuguese classifieds.

I actually didn’t even know that.

It is.

Yeah.

Yeah, we use Zomato a lot.

Interesting. They were marching in that direction, and then food delivery happened. There’s a venture-backed business called Swiggy in India that raised a bunch of capital. At one point, there were 20 different food-delivery apps, and I think this is the only classifieds company in the world that successfully evolved the model. It was a function of a paranoid, customer-centric entrepreneur at the helm who thought, “This is the better thing for customers. We need to figure it out.” They entered food delivery maybe a year late, but entered it well. They ultimately won food delivery with less capital than their competitor, Swiggy, had access to.

Today, food delivery is a 60/40 duopoly, and it’s the most profitable food-delivery category in the world. Both businesses earn 5%-plus margins in food delivery. Over time, they’ve entered new segments through acquisitions and organically. I think their mental model is Tencent: you have a bunch of cash, you have distribution, and you should be the preferred growth-equity investor in the country. I think that’ll become more and more the case as venture leaves India or isn’t paying attention to it.

Their second category came out of an acquisition and is now much larger and more valuable than food delivery. It’s called Blinkit. It’s quick commerce. If you remember during COVID, all these 10- or 15-minute delivery apps emerged, like Gorillas in Europe, or Gopuff and Getir. Blinkit was in that category.

India is the place where that category has worked by far the best, for a bunch of reasons. The most important, I think, is that there are no incumbent grocers of scale. It’s very hard to compete with Aldi as an upstart. No brand is going to give you a decent procurement price for a very long time. These guys became the number two grocer in the country within 3 years. They’ll probably be the biggest grocer in the country in the next 12 months.

That’s crazy.

Yeah. Their percentage market share is still very low, so they’ve gone from zero to $10 or $11 billion in top line in 3 and a half or 4 years. It’s an insane customer experience. The other thing they did very differently, which required insane operational complexity to manage, is offer many more SKUs than these global quick commerce businesses.

Normally, Gorillas or whoever was offering 15,000 SKUs in 10 or 15 minutes. In their most mature geographies, these guys offer 80,000 SKUs in 15 minutes. You have basically the entire—

Jesus.

Maybe not the entire Walmart—Walmart has 300,000 SKUs—but you have something like a Target available to you in 10 or 15 minutes. If you’re in India, you’ll realize traffic sucks and parking sucks. It might feel frivolous to ask why you need these things in 10 minutes, but the reality is that it’s compensating for the lack of urban infrastructure.

The cohorts are incredible. In mature geographies, about a quarter of households are using it. People think of it as a premium service, but it’s not. The price points are comparable to or better than the mom-and-pops, who make up the majority of the category in India. Anyway, that’s business number two, and it’s much larger than food delivery at this point.

They have businesses three through seven in incubation. The business system is that you have CEOs or P&L leaders running each of these companies, but talent is very fungible. Think about an Elon-ish model where you have a founder who is almost like the head of HR, and then you have a Gwynne Shotwell equivalent running each business unit.

When a new business unit starts to look really promising, the young, 25-year-old product manager in food delivery gets the opportunity to go lead a P&L somewhere else. That results in great talent density because young people get so much opportunity in this system. Traditional India is otherwise very hierarchical. A 30-year-old managing a billion-dollar P&L is totally unheard of in India. It’s kind of unheard of anywhere, but that’s what Zomato/Eternal offers people.

I’m very bullish about that business. From a price perspective, we built a position a year and a half ago. I think we were paying 1.3 or 1.4 times forward GMV. Roughly speaking, I think the business will end up with 5% free cash flow margins across all its business units over time. Let’s put a footnote in about the thinking behind that, but if you do the math, that implies you’re buying the business at something like 30 times free cash flow. It was growing 70% at the time.

More importantly, the offline retailers were trading at 3 times GMV and 60 or 70 times earnings. You had the online business growing much faster, likely to end up dominant, and trading at half the multiple of the offline guys. That was a sweet setup, and I think there are a number of these.

Yeah, super interesting. Who’s the founder? I’m not familiar with him. What’s the story? He seems like an absolute animal to go after these businesses. He must also be older, right, if he started the business in the late 2000s?

He’s in his early 40s, yeah. He started it right after—

Okay.

He started it at 23 or 24, in the late 2000s, and he’s in his early 40s now. It’s too flattering an analogy, obviously, to compare him to the GOAT, but his operating style is very much like Elon’s. He has lieutenants running the day-to-day of each business unit, and in some ways he thinks of himself as the head of HR. His main thing is nurturing talent, making sure promising people have paths forward, and retaining them.

Occasionally something goes wrong and shit is hitting the fan in one of these business units. Then he shows up, scares people, and creates real urgency and intensity. I can’t remember if I told you this story, but they went public in 2021, and he found that his whole leadership team, who had gotten liquidity for the first time, had become complacent. Ten out of the 15 senior leaders of the company were basically pushed out and forcibly retired. He promoted a cohort of young people to replace them.

I think that’s emblematic of who he is: very high expectations around intensity and urgency, but he doesn’t execute in the details of any of the businesses anymore.

Awesome. Let’s do the other one you’re also bullish on. I’d be curious to hear.

Yeah. Meesho. The analogy would be PDD in China or maybe Shopee in Southeast Asia. The category would be value commerce.

The basic premise is that you have 2 versions of e-commerce that work in every market. You have Amazon-style e-commerce, with a lot of fulfillment infrastructure. They own their own trucks, and the North Star is getting products closer and closer to the customer so they can offer two-day delivery, next-day delivery, et cetera. I think the quick commerce business is that. I think they’ll end up being the Amazon of India, but with the category mix tilting more toward grocery.

Then you have value commerce, which is about offering a ton of selection and the lowest absolute prices. If the trade-off is delivery times, that’s okay. That version didn’t exist in America for a long time, but you saw the market caps of Dollar General and Walmart, and those cohorts weren’t big Amazon users in the U.S. When Temu emerged, you realized there was demand for this value segment even in America.

Outside of America, where the income distribution tilts more and more toward lower-income people, my view is that value commerce is likely to be larger than the Amazon version. That’s what Meesho built over time. It took a very circuitous path to this business model. It started as a business enabling WhatsApp resellers. A neighborhood mom might have a catalog of 20 products to sell to other neighborhood moms. Meesho enabled that catalog, then enabled those moms to buy from wholesalers. Then they discovered their wholesale prices were so good that end customers would download the app anyway and buy directly from wholesalers. So they converted to a marketplace and have evolved over time.

Today, it’s a monopoly. It has on the order of 70% to 80% of all e-commerce in Tier 3 and Tier 4 cities in India. That’s where 1.2 billion of the country’s 1.5 billion people live, and all of that e-commerce happens on Meesho. The vast majority of trucks going from a big-city warehouse to these towns are Meesho trucks. They don’t own the trucks, but their dominance means that anyone entering these categories would be starting with very low logistics utilization and would have no ability to acquire these customers.

It’s very counter-positioned versus the Amazon-Flipkart duopoly because the products are, on average, 250 rupees. What is that? About $3. The average order value for the competitors is $15 or $20. These guys’ average order value is equal to the logistics costs of the bigger guys.

The founder is very young; he’s in his late 30s. He started the business in the mid-2010s, again right after college. Meesho has very different DNA from Deepinder. They’re super customer-obsessed. The North Star for the business is to lower prices every single year. They lower prices by getting sellers in the marketplace to compete and by making logistics more efficient.

Their customers are rural-ish customers in small towns—“village” isn’t quite the right word. So they hire IIT engineers who live in Bangalore and have no idea who their customers are. The solution is that the first 2 weeks for any employee at the company are spent as an intern to a seller who lives in a small town.

They have to go to the small town, live with or near the seller, and deliver some business value to the seller. You come away with some business intuition. The leadership team also spends a week every quarter with customers. There are a bunch of these habits around understanding who your customer is and what they care about, and that translates into a lot of dominance.

I think over time, their next act will likely be in financial services. There’s a lot of demand for credit among their customers. They have 300 million customers, largely unbanked, and you could perhaps start with BNPL and then do unsecured consumer lending, and so on and so forth. So that’s Meesho in a nutshell.

Yeah, really compelling.

Yeah.

I think China’s a good comp for some of these businesses.

Absolutely.

They end up taking a more similar shape to Tencent and JD. But you also mentioned an India data center opportunity, which I’m very curious about. What are you looking at in terms of that? Are there interesting plays there? You mentioned a little bit about how you see India as one of the best places to build them, but is there a way to bet on this right now? What are you looking for?

4. India’s Data Center Bet

I started doing work on this over the last 6 or 8 months, and I’m kicking myself for not having started 18 months ago. It started with the question of whether there would be enough sovereign demand, both from the government and from financial services businesses that have data localization rules. I pretty quickly realized the demand isn’t just sovereign demand.

One problem with inference in India is latency. But if you put that aside and think about long-horizon tasks that could be run in India, the cost per megawatt in India is half the cost per megawatt in the US. The statistic I’ve seen is about $6 million in India and $16 million in the US to put up a megawatt. Could India end up with substantial inference market share? I think that’s the hypothesis.

It’s starting from a very low base. The whole country has 1.2 to 1.5 gigawatts of data center capacity, and very little of that is GPUs today. They’re growing quickly; they’re adding 300 megawatts this year. But based on announced capacity, they should be adding 1.5 gigawatts a year in 3 years. The second derivative is going to go up 5×. But that 1.5 gigawatts gets you to 8 or 10 gigawatts by 2030, which is still a very low percentage of global compute.

That number implies India-for-India demand. If India does end up serving global demand, that number could be 20 gigawatts, not 10 gigawatts. In terms of opportunity, the vast majority of the spend is going to go outside of India. It’s memory and compute and so on. But 30% of the bill of materials is HVAC, piping, and even liquid cooling. I think it makes sense to localize, so you could imagine suppliers attacking those categories.

They might be suppliers to Schneider, ABB, Vertiv, and so on. If they do a good job in India, could they become global suppliers to those companies? Maybe an Indian data center supply chain business could look pretty good. We’re building a position in one of these that I’m really excited about, but I’ll maybe tell you offline if it’s a couple-hundred-million-dollar market cap. I’ll keep the name private for now.

Yeah.

What’s exciting is that there’s a lot of compute that was going to come from the Middle East that I think isn’t going to happen anymore. That’s shifting toward India. The company I’m describing is now becoming a supplier to some of the Middle East build-out and a lot of the Europe build-out. If your TAM is the COGS of Schneider, that’s going to be a pretty big TAM over time, or the COGS of any of the large data center OEMs.

5. AI as an Investing Tool

That’s really interesting. I hadn’t considered that. I’m excited to hear about the company. Maybe the last section I’d be really curious to ask you about is how AI has changed your investment process. I know you mentioned a while ago that you would have hired a data scientist, but now you can just build all the scrapers yourself. In general, what’s your investment process like, and how has it changed with the inflection in capabilities we’ve seen over the last year?

I don’t think there’s a lot of rocket science yet, but I’m trying to put the scaffolding together to increasingly enable agents to start doing things for me. The model has to be that I have all the context about a company, both to make a decision and to have a sharp enough world model, if you will, about that company so that when the facts change, I can update my model quickly. Outsourcing the thinking to an AI model that writes a memo for me, which I read and then make decisions off of, isn’t so useful.

I’m bandwidth-constrained. I can look at a couple of companies a month with a lot of focus, maybe. Are there ideas six through 20 on my watchlist that I’m getting through sequentially, where idea number 18 is actually really awesome? On my current schedule, I’m going to get to it in November, but really I should be spending the time on it now. That’s the problem statement I’m trying to solve.

The level-one thing you do is feed a bunch of context, like company disclosure documents, into a model, and it creates a bespoke equity research report. I don’t think that’s that special or that useful, honestly. The layer-two thing that’s useful is monitoring companies I’ve looked at and passed on, but that could become interesting if something changed. The simple version of that is price, and people have price alerts. But what about a fundamental change? If this new business unit started to really take off, it would totally change the distribution of outcomes, and I should revisit it. I think I and most fundamental investors are pretty bad at monitoring those kinds of fundamental inflections in businesses we’ve already spent time on.

Mm-hmm.

I’m trying to get to a place where, after I’ve looked at a company, I have a set of questions that an agent then goes and reads a bunch of news and earnings calls to answer on a monthly basis. It tells me, “Hey, I think question number four, which you flagged as critical to the thesis, had an answer of no. Now maybe it’s yes. You should revisit it.” So it’s really just a fancy alerting system, but I’ve found that to be somewhat useful.

The biggest source of insight for our model is still spending time with humans, like spending time with the founders. There are a lot of private competitors to our companies, and I find it so fun to spend time with those founders because they’re trying to poach senior employees from my portfolio companies. I get to hear whether that’s easy or really hard, and that tells you something. They’re actively attacking our portfolio companies’ moats, so you learn a lot about whether the theory is real or not.

Often, execution is all that matters, obviously, and you learn the detailed anecdotes about what makes the execution of these winners so distinctive. I’m not sure how to make that piece something I can outsource. The long-term framework is to try to automate more and more of the desk work so that I can spend all of my time in the real world building these relationships and getting insights from them.

At some point, enabling an agent to do expert calls and customer surveys autonomously isn’t in the realm of science fiction. I think that’s probably possible with today’s technology if you stitch it together right. That would be really powerful, because then you can imagine I have this question list and the agent thinks, “Okay, the answer to question four is in the brain of this customer. Let me do a call with that customer,” and then I’m giving that agent a budget to do it.

That’s futuristic. I’ve dabbled a bit, and the person definitely knows an agent is calling them. The agent doesn’t do that good a job, and I’d rather just do it myself today.

That’s the world we have to prepare ourselves for as fundamental investors. The things needed to make that work are really good scaffolding around companies and what matters about a company, at least for me to make a decision, and then empowering these things to fill in the blanks as much as possible. That’s kind of where I am on using AI.

That’s very interesting. Are you still hiring new analysts, or how do you see the role of an analyst? Do you have an internal system that everyone sort of uses?

I think I would have hired somebody this year but for all of this stuff. I think we will still hire people, and part of that is just that the value of a human I trust—their judgment, ability to communicate, and so on—is going to be enormously more valuable than an agent telling me something for a long time.

Yeah.

And the idea of a partner who disagrees with me on something and then argues with me about it for a week to sort of get to the truth—the AIs are too sycophantic to do that. That’s A. And then B—

Yeah.

There’s something about the nature of the sloppy prose that makes me skeptical of it. I feel like it’s kind of an idiot just because of the language it’s using, you know? So—

Yeah.

Having somebody on the team who can form views and then convince me of their views—I almost feel like the latter piece, reading text—

Yeah.

—is going to take a long time for that to happen for me, right?

For me, yeah, I totally agree. How do you think about your time spent between… You told me you’d rather spend sort of 70% of your time on your existing companies and 30% on the new ones, or something along those lines. Is that still the case? And maybe AI is helping you scale the amount of time you can spend on new ones, or do what you said before, figuring out which ones you should spend time on. Is that still roughly how you spend your time?

Yeah. So I think it’s—

Maybe important context is that you run about 10 positions, so you’re relatively concentrated.

That’s—

Important context, right?

That’s right. The top of the portfolio is 10 to 12 positions at 10 to 12% at cost, and then we have some smaller starter positions. Our model has a lot of upfront CapEx per idea, if you will. A small percentage of that is desk work, like understanding what the company does, the competitive dynamics, and the financials. Much more of it is going out in the world and building relationships with people who matter to these portfolio companies.

You can passively sit back and ask an expert network to source you a former employee, and sometimes that works. But it’s much better to get to know an important senior former employee, befriend them, and add value to whatever they’re doing. Often they’re starting their own company. The CapEx for a name, if you’re going to build a network around each portfolio company, is pretty high, and it takes time. You can’t compress the time it takes to do that; it happens over six or 12 months.

I think that’s what I was alluding to with that comment. In order to get to mastery around a portfolio company, that means understanding the business inside and out and having intuition for the people—their strengths and weaknesses. When they’re doing a new initiative, you can recognize that this is a very product-oriented company, and this new initiative is going to require some on-the-ground ops, so it’s a new capability for them and the odds of success are a little bit lower. Being able to have that mental model and a network of people around the company just takes a lot of time.

I try to do all of that work for these starter positions. In the process, I become more and more convicted in the idea, the position size grows, and it becomes one of those companies we marry or not.

Right.

That’s kind of the process.

That’s awesome. It makes a lot of sense. Last question: I’m curious about both the bear case and the bull case 10 years from now. If everything plays out exactly as you hope for India and for RaaS, what does it look like? Paint the picture for us. And if it doesn’t, what would you say is the biggest risk to the thesis? Maybe you can start with the risks so we can end on—

On—

On the positive. Yeah, yeah.

I think the biggest risk is that I don’t execute well. The cop-out answer is that India falls apart and the opportunity set just isn’t attractive. Certainly, that could be the case, but that’s not controllable. The controllable is that it’s such a narrow model that if the blinders I’m putting on myself to increase the odds of success on the things we’re focused on blind me to where I should be focused, that could be a problem. I’d put that in the category of execution.

I’m pretty paranoid and self-critical. I have a list of potholes that I think are structural to the model, and then potholes that are just related to who I am. That’s a pretty long list, and I try to be aware if I’m stepping into some of those potholes. I think it’s that. There should be great companies that emerge out of this country, and it’s incumbent upon us to find those companies, right?

Awesome. And then maybe the bull case. What’s the bull case? If everything really plays out, how do you see RaaS? What do you see India doing? Maybe the most bullish case for India.

I think the most bullish case is that you have a young country, and the youth of that country—I don’t mean demographics, but the youth of its entrepreneurs—ends up being a good thing. The parts of the economy that aren’t explicitly tech actually end up being much more adaptable, and there are probably policy unlocks connected to that. I think the direction of travel is quite strong, and that continues in a good direction as well. That’s the India piece.

On the RaaS piece, I think there’s a way of operating in the world as an investor that I call founder-centric underwriting. Ultimately, the most important thing is to find exceptional entrepreneurs who are going to stay hungry over 10 or 20 years and are in situations that allow them to be very, very ambitious. That looks like a core business that’s already very profitable. I think those founders deserve a certain kind of investor. It doesn’t just mean being dogmatically long-term and owning a stock for a long period of time; it means sharing things with those entrepreneurs about how the world is playing out. They’re very focused on day-to-day, and I think you can add value in those ways. Ultimately, you can add value as a capital partner.

Going back to the Elon example, and there are a number of other versions of these, there are great investor-entrepreneur relationships where the investor, over five or 10 years, develops such intuition for how this person operates that when that person picks up the phone and calls them and says, “I have a new idea,” the investor is able to invest in a matter of days. Not because they’ve done a ton of diligence in the subsequent 48 hours, but because they’ve done a ton of diligence over the last five or 10 years, right? I think the richest version of RaaS for me is that—

Yeah.

—we build that high-quality relationship. It doesn’t even need to be with 10 founders. It needs to be with one or two of India’s generational founders over the next 10 years. If we do a good job, it probably starts today as a thought partnership. I write these detailed letters and try to be reflective.

And part of that is for me, and part of that is for them, honestly. The audience is the founder. If we do a good job at that thought partnership stage, and then eventually do a good job building a high-quality capital base, I think we can graduate into becoming the capital partner that these people sort of deserve. I’d be really excited about that. It’s fun to be investing in a country that will likely be very, very large from a market cap standpoint—

Yeah.

—where very few people are operating this way, right? Versus, it’s not that novel in America.

Yeah, that’s really the opportunity I saw with you. First of all, you’re a great writer and a great thinker. We’re going to put some of these things in the show notes, some of the letters and the Tiger Feathers stuff, too.

Great.

But also, the opportunity—and I think your style really lends itself to it—to build a concentrated portfolio and concentrated relationships with some of the most important founders in India, who go on to do more and more stuff, so you can be an amazing partner for them in a way that doesn’t really exist yet in India. I really enjoyed this. I always enjoy chatting with you. I think it’ll be interesting for people to hear this because, especially now, all people talk about is AGI and superintelligence. I think your perspective will be super interesting for people, just seeing the opportunity in this giant market that is India. I really appreciate the time, and thanks so much for sharing your insight.

I really enjoyed it.

Cheers, man.

Bye.