Full Transcript

Utsav Somani: then you did media and you've co-produced a movie as well. And a VC at K Capital. At what point, and you've written a book also, I think, about poetry. And we'll get to the naming part later, but at what point did you realize that you wanted to do your own fund? And how do all of these things connect?

Natasha Malpani (Boundless Ventures): Yeah, interestingly, I actually think the journey connects really, really well. So initially I studied immunology at Oxford, thinking that I was gonna be a scientist. Then decided that life in the lab really wasn't for me. So I ended up investing in healthcare for five years for a billion dollar fund in London. Went to Stanford to do my MBA, very much thinking I was gonna come back and be an investor in India. That's when I took a detour and actually moved into the world of media. So I've been an operator and founder in the media space for five years. I actually came back to VC though when the Chad GPT moment happened, N22. So first I worked for K Capital and then set up my own fund last year called Barbliss Ventures. But actually, interestingly, when I set up the fund last year, people were saying, you're crazy, you're confused, this is in the valley, our consumer brands could go. But obviously the tide has turned quite a lot in the last year and AI and frontier tech has become very hard to invest in. But actually, my journey of healthcare and science to storytelling, to investing is actually the thesis of the fund. So we say we invested the intersection of science and story. And interestingly, I'm seeing that actually the best founders now are interdisciplinary. So that mix of left brain and right brain thinking is increasingly what we are looking for.

Utsav Somani: And you mentioned that deep narrative support as an offering to the portfolio founders or founders that you invest in. Share a little bit more about that.

Natasha Malpani (Boundless Ventures): Yeah, I mean, I think now, as increasingly anyone can build anything, right? Like given the world of generative AI, whether it's code, it's design, companies at different layers of the stack, what is actually causes attention? So I think that's what I'm saying. It's very important for founders to understand and be able to tell their stories and be able to stand out from the crowd, right? Because at early stage, we come in at pre-seed seed, right? Like often before the company is formally set up, what do you have? You basically have your belief and you have your vision and you have to actually make that reality true, right? So to be able to convince investors, your customers, your employees, you have to be able to tell that story. And then of course you have to execute, build a momentum and then bring that world to life. So we do help with, I hate the word content. We help with storytelling.

Utsav Somani: And I mean, Dhruv, sorry, I'm gonna like, one more question that comes to mind, the new media question. I mean, A16Z has popularized this word new media. We're sidetracking a little bit, but what is new media to you?

Natasha Malpani (Boundless Ventures): Yeah. I mean, I think actually, interestingly- You hate the word content, so yeah. No, new media is old media. We thought that with generative AI and AI slop or whatever, people's attentions will be more constrained. So you need more short form, but actually the world is moving back to long form. So the better your analysis, the more organic your content, the more true it is. And I actually think offline network does an excellent job of trying to tell honest stories of what's actually happening on the ground. The more your words stand out. So it is interesting that you said that, right? Whether it's A16Z's new media team, or it's the Hummingbird hiring the generalist founder, or various VC funds actually hiring media partners. I do think even VCs becoming more interdisciplinary, right? So it's not just founders are trying to marry left brain, right brain, combine execution with narrative. VCs are also actually spending a lot of time now on their branding and differentiation, and actually hiring a partner level writers and storytellers.

Dhruv Sharma: Actually I have a somewhat related question, Natasha, which is you've now been an active participant in the ecosystem for a while, and likely an observer before that. How have you seen founder archetypes evolve, and especially founders who are building at the frontier?

Natasha Malpani (Boundless Ventures): Yeah, yeah. See, we only look at AI and frontier tech, right? And we're a cross-border fund. So we also look at India, we also look at the Valley. Very, very interesting to see how founders are evolving, both in India and the US. I actually think when people talk about the technical talent gap or capability gap, there's none. Honestly, like the quality of talent across both ecosystems is really high. What I'm finding interesting actually is the level of risk-taking and creativity, right? So I see with under 25-year-old founders, you have much wider ambition and ideas. So if you see even our own frontier tech or physical AI portfolio, some of the outlier teams are very, very young teams, right? Aspera trying to build amphibious autonomous airplanes or Armatrix trying to build snake-like robotic arms for engine inspection. So seeing a big age difference, one, in terms of younger founders having more audacious ideas, that are category creation, whether it's in consumer AI or physical AI. And then two, we already discussed the interdisciplinary thinking. Three, I also think because the bar for building or putting something out there is lower, it's actually, people are just building a lot more and you're just seeing a lot more people attracted to entrepreneurship because the entire archetype of thinking, okay, you needed 10 years of experience or whatever to build and create, you had to be from an academy or whatever, is gone, right? So, I mean, you can come from any field and build in any field, as long as you take your time to actually understand your consumer. So I think we're seeing a lot more ambition and more cross-disciplinary teams.

Dhruv Sharma: And if you take a step back from, what's a year ago? 90, sorry. I was just saying like, so we just spoke about the founders, but then if you look at the two respective opportunity sets that you're going after, you said AI native companies and then companies like frontier tech companies. So maybe we'll speak about AI native companies for a bit. Where are you seeing the opportunities in Russia and where do you think they're overstated? Where do you think they're understated? I know you've written a very coherent narrative around this and put it out there, but who better to sort of break that down than you just doing it for us, yes.

Natasha Malpani (Boundless Ventures): Yeah, no, I'll answer. I'll give you a longer answer, right? Because I think the conversation can be very shallow when you talk about AI, because it's one word to describe many, many things. So I'd break the stack down in two ways. There's one way of looking at the stack, which is you say, there's the chip layer, then there's a foundational model layer, there's the agent tech or tooling layer, and then there's the application layer and where do you play it, right? And at the chip layer obviously you've got companies like NVIDIA, foundation model, you've got an open AI or Anthropic, tooling, you've got companies working on memory, context, governance, et cetera. And then application layer can be consumer enterprise, right? So that's one way of looking at the stack. So in India, I'm very, very bullish on consumer AI and physical AI. Consumer AI is pretty obvious, right? I mean, we've got the population, we've got the openness, the willingness to try AI. We've got the engagement. You know what I mean? There's a reason, right?

Dhruv Sharma: That open AI and Anthropic, et cetera, working so hard to- So I also think we have unsolved problems that are uniquely our own and no one is gonna come from outside and solve them for us, so.

Natasha Malpani (Boundless Ventures): Yeah, exactly. So like, exactly, right? So building in health or education or travel for India is gonna look very different than, you know, building that for a US customer. So consumer AI is very, very interesting and very, very early also, right? Because what breakout companies can you name yet? And there's reasons for that that we can come to. And then the second is physical AI. So now see whether you call it frontier tech or deep tech or physical AI. Again, right, this used to be a very unpopular sector that's now suddenly become a very popular one even with investors, because there's a reason for that. One is, of course, the quality of talent. Two is the government support. Three is the policy support. Four is the cost advantage. Five is the supply chain advantage. So it genuinely is, it's not hype, it's actually a very interesting time to build in manufacturing, robotics, space tech. And you are gonna see breakout companies both for India, for India, and from India for the world in these categories. And then we spend a little bit more time when it comes to AI and for our like security, governance, compliance, or enterprise AI, if it's not financial services, looking at global markets.

Utsav Somani: So let's talk a little bit about your two investments that you've mentioned, I think in the very recent interviews as well. Pearsight, which is doing ocean intelligence via satellites and Armatrix, which is doing what you just mentioned on the show as well. The investment, what goes on while making these investments, you're running a 200 crore fund. Do you think your solo GP fund of 200 crores is apt for making these pre-seed bets? And what stages do you wanna play at? And how are you supporting these founders? And also the awesome work that these founders are doing.

Natasha Malpani (Boundless Ventures): These two companies. Yeah, so just for the fund actually, we're a 300 crore fund, oversubscribed. And we've done nine companies so far. And then we also have, we have access to a billion dollar prop book. So we are positioning this first check to IPO, right? Like we actually can come in really early and then double down at every stage, which is why we do Frontier AI. When I say Frontier AI, one is of course the deep tech physically. I'm gonna talk about Pearsight, Armatrix, Espera in a second. But I also mean Frontier AI in terms of actually spending time with labs. So we actually spend time with AI researchers at IT Madras or ISC and say, how can you actually commercialize research? And this is also where my healthcare background does help play a role. Because of course, bio into AI now is very interesting. So Pearsight is, so whether you see Pearsight, Armatrix, both are fascinating companies because they're full stack players. We're not talking about intelligence in the sense that they just need to build software. Pearsight actually needs to launch 32 constellations into space, 32 satellites into space, right? Like a constellation of satellites to be then able to collect the data they need to be able to visualize the earth's oceans, combining two techniques, SAR plus AIS. So there is a very unique intelligence layer they're building by fusing these two technologies, but also they're taking hardware risk, right? In terms of actually building and launching the satellites. And this is the same for Armatrix. What they are doing is they are trying to build snake-like robotic arms, which are basically these flexible arms that can operate in constrained environments. Now that is actually also, from a physical perspective, very challenging because they need their arms to be three meters and of a certain level of flexibility. So this is hardware IP also that has not been created anywhere in the world. Plus, of course, there's a software layer of optimization where they have to figure out how that arm can move in constrained spaces. And let me tell you one more interesting thing. What I found interesting with these companies is one, obviously, you know them. Founders are stellar, generational founders, doing the hardware, software, building through IP, again, India for the world plays. But I'm also, it's very interesting to see the commercialization of how often these companies that you think would be so niche actually have inbound interest from government. So that's my point, that it's not, this is not just hype. There's this actual interest in defense, navy.

Utsav Somani: And the thing that also stands out to me is that they're building such unique solutions that the datasets that they need as an Indian company, I think those will be fairly unique as well and how they have access to that or accumulate that over the course of the journey.

Natasha Malpani (Boundless Ventures): See, to be honest, unless you're now building, you know, at the Chiplay or you're building data centers or, you know, you're building at the foundational level, the answer has to be that your IP is in the data you collect, right? I mean, even that's my point about consumer AI. You know, it used to be that if you were building in consumer, you said, oh, you had consumer insight, you had consumer understanding, you understood the psychology, you could build an interesting consumer company. Those, that is not true. Like my consumer AI companies, I have a company in local discovery, I have a company in healthcare, Glide is the company in local discovery, SuperHealth is a company in healthcare, you might know Varun, right? So these founders are also very technical because especially when it comes to building at the application layer, there's a lot of infra issues, right? It's not that you can just build agents and deploy them in the world. What do you do with memory? What do you do with context? What do you do with governance? So all, you need this mix of being technically proficient to be able to launch consumer products now. The bar is very, very high.

Dhruv Sharma: I think speaking of data, Natasha, this is a great opportunity for us to pick your brains on what's happening with respect to memory. Was it considered a bottleneck? What are the recent breakthroughs? And then also what's happening on the edge? Like where are the smartest applications today and where is the puck, so to speak, going in the future?

Natasha Malpani (Boundless Ventures): Yeah, yeah. See, now memory again, like I think so much of AI, memory is a misused word. You've got memory, which- Software or hardware. And then you've got short-term memory, you've got long-term memory, you've got episodic memory. Basically all we're saying is that when, for example, even you as a consumer, when you talk directly to a foundational model, right? Which are GPT or Claude, you will notice that in every window your conversation resets. That means that, of course, these LLMs don't actually understand your own context. And that, of course, then, if you are trying to build a consumer app, you're trying to help someone shop better, you're trying to diagnose someone's illness, you're trying to suggest where they go next for travel from a consumer perspective. Or if you're trying to help an organization's workflow, you're trying to help wealth managers suggest portfolios. If these agents don't have memory or context, how can they make you, how can they help make better decisions? So it's not enough for AI to be generative, right? The point is to say, how can you automate decisions? How can you actually remove friction? How can you actually execute on your behalf? Which is why memory or context is this key constraint that honestly no one has solved. I'll tell you, there's some buzzwords now, right? Like people are talking about context graphs. The new hot thing is Obsidian. By the way, you wanna trend on Twitter, you say Obsidian plus Claude and you will trend. But this is still, it's still very early. Like Obsidian is not necessarily the answer, right? There's very few people that as a consumer, you can't expect me as a consumer to create a vault for my memory, right? Most consumers aren't even using Notion. So memory and context needs to be solved at an infra layer, which I think honestly, a lot of the times the foundational models are gonna take that layer. I would bet more on the foundational models to actually improve memory and context, right? They know this is a key sticky point. What was the second part of your question?

Dhruv Sharma: Well, the second part was what's happening. We was talking about data, remember? So we spoke about memory and we were also speaking about what's happening on the edge with respect to edge devices and edge computing.

Natasha Malpani (Boundless Ventures): Yeah, so that's the thing, right? So edge is gonna be very, very interesting, again, very early days, but the obvious question then becomes, okay, if you actually have my data, or you have my memory, or you have my context, what about privacy? So one way is to solve it through governance saying, okay, how do you guard real AI? Again, that's a portfolio company of ours, Altor. That's trying to say, okay, who in your organization has access to this agent? Who in your organization knows what data the agent has access to? And how do you guard real and control that, right? So there's all these organizational problems that are now gonna exist in terms of saying, how do you know what agents are there? How do you know what agents are doing? How do you know what data these agents are pulling? So that's a whole question on security, governance, guardrailing. And then the other way to say it is that, maybe you build on-edge just to solve the privacy concerns, right? Because there's two camps, honestly. Some people are saying that MCP is already taken off. It's one.

Natasha Malpani (Boundless Ventures): People are saying MCP sucks, and CLI is the future.

Dhruv Sharma: Connectors of the future.

Natasha Malpani (Boundless Ventures): And some people are saying that, well, actually, none of this is gonna make sense. People are not gonna give you data, and things are gonna move to either on-edge or on-prem, and you're gonna need to create different data source. So, as with most of AI, I think you have different camps, and things are still to be proven.

Utsav Somani: And where are you placing your bets in terms of, will Apple, the new Siri, and everything that comes out, which is supposed to be magical and have context about everything on your phone and your life? Because phone is pretty much one device where you have, I mean, given all the context that you need for most of the daily tasks, like at least from a personal consumer use case. Are you betting that?

Natasha Malpani (Boundless Ventures): Exactly, but personally, it's interesting. On a personal level, it's interesting. I don't know how you solve it at an organizational level. So that's how I always divide. Like, from a consumer perspective, it's interesting, but from an enterprise perspective, it might not be interesting.

Dhruv Sharma: So maybe let's ask that question. Like, I mean, we'll talk about personal agents. By the way, Natasha, both Utsav and I, over the weekend, have spent time with friends who've shown us what the future might look like, and in my case, told me a lot of lore also. Maybe I'll share that joke quickly. Apparently, there's people running agents. Agents have OKRs. Agents have to send performance reports. Agents are getting fired. Agents are getting token budget increments. It's wild out there.

Natasha Malpani (Boundless Ventures): So, agentic commerce, like, so that's the right thing. The future of the web is gonna be, there's gonna be more agentic traffic than human traffic. So you are gonna have to build a world for agents. That is a given, I think.

Dhruv Sharma: But the question we have for you right now is not one about- For agents to hang out.

Utsav Somani: Oh, yes, oh, yes. That's true, yes. So agents are posting pictures from their vacations and stuff.

Dhruv Sharma: Oh, my God, yes. Agentic vacation, yes. Yeah, they were supposed to send us on vacations, going on vacations of their own. But no, Natasha, the serious question we were gonna ask you is, so, I mean, let's set personal agents aside for a second, but enterprise agents, what's it gonna take to increase their adoption safely in the enterprise?

Natasha Malpani (Boundless Ventures): Yeah, I mean, so I'm sure you all know this stat, right? That 95% of agents fail in production. So this is what I'm saying about narrative and hype versus reality. The future is gonna be agentic. A lot of the web traffic is gonna be agentic, as we are saying. You're gonna have agentic commerce, agentic marketplaces. I mean, I feel like the open book or mold book or whatever was a small little marketing gimmick that actually did show people a future of agent coordination. So agent coordination is definitely a massive area of interest. The problem is today that we don't actually, see, software used to be predictive, right? So all our software testing, evaluation, et cetera, has been designed for a predictive world. Now, the amazing thing about generative AI is not just that we can talk to machines in natural language and they can talk back to us. The amazing thing is that actually you're living in a probabilistic world, right? So you don't know, you genuinely, like we've never worked with- It's like a bowl of muesli, you don't know what the next spoon is gonna bring. Yeah, right, so exactly. It's like, I don't know what you're gonna say next, right? So it's the same thing. I don't know what my agent is gonna do next, so how do I control that, right? Like, so this is the issue with organizational deployment. If I don't even know what agents exist, if I don't know what data they have, if I don't know what action they've taken, it's like working with an employee. If I'm completely blind, is my employee actually doing what they're saying they're supposed to do? Are they making mistakes or not? How do I give that employee feedback? How do I tell them, no, this was wrong, please roll back that action? As you said, how do I maybe say, okay, this is not working out, like what are the, okay. You are gonna have to manage agents, it's totally true, it's not a joke. So, you know, those are like, we basically need complete organizational redesign. This is what the sticky point is. It's not saying, oh, we need more intelligence or even that we need a better infra, like this is not a problem that is gonna be solved by developers and engineers. Organize this list.

Utsav Somani: For our listeners, I think what might be interesting is, suppose I run a company which employs 200 people, we're selling D2C products. Many of the Indian founders were funded and have hit certain scale, fit in that category. What are the steps they can take across their organization to see what workflows they can identify to be automated or even get to an AI native company or AI enabled company? And not just use JIT PD and stuff at just the bot level.

Natasha Malpani (Boundless Ventures): Yeah, see, that's the difference, right? It's different to say AI native versus AI enabled.

Natasha Malpani (Boundless Ventures): So, that is genuinely what's gonna change the game for people. So to say that, okay, these are my current workflows as an organization, say you're building a consumer brand and all orgs will have HR, marketing, customer support, et cetera. It's very possible today to say, okay, how do I actually just get all my employees to just work directly with foundational models to streamline our workflows and have them use cloud co-work or projects or whatever so people can share their data, just record meeting notes, structure the organizations data better, use drives. There's a lot of low hanging fruit that every single person can do today to significantly streamline workflows and make things more efficient. And you know, that is by the way, one thing, like you can't have a verbal culture anymore, right? That's a very basic thing. You just need to have more things written down, more things recorded and then stored well for AI to get your context and for you to be able to share that data. That is possible today. Definitely, if you're not doing that, your organization is behind. What we are saying about AI native is that you genuinely say, okay, this is my org chart. What functions do I maybe not need anymore or what functions will I be able to automate over time and how do I build not just individual agents, but agents that coordinate with each other to be able to work across workflows. Now that I'm not gonna say you should do today, or these are the 10 vendors you can use because like I'm saying, agents fail in production.

Dhruv Sharma: I've heard another interesting perspective.

Natasha Malpani (Boundless Ventures): The move is to agent coordination where all orgs will have multiple agents orchestrating and coordinating with each other.

Dhruv Sharma: Yeah, yes. I've heard another interesting perspective on this, which is this is a great opportunity and maybe it's the final opportunity for founders to just take a pause and also ask the question, if I was refounding the company today with what I know now, what would that second company look like? And in a sense, be building two companies at once in parallel, one which you're kind of moving away from, one that you're moving towards. And that's easier said than done, but I think the need is the hard, yeah.

Utsav Somani: Probably philosophical way of looking at things. I mean, all of these are such contextual stuff, like how your AI needs context, you also need context about your own companies. I think it's super contextual.

Dhruv Sharma: You do, but if you wanna walk away from your code base and not be beholden to the interfaces you built over two, three, four, five years, that's to my mind, that's one of the questions to ask yourself. Like if you were to do a rewrite from scratch, what would it look like?

Utsav Somani: Actually, let's bring up that novel tweet as well, where novel recently tweeted, very simple, this thing, software is, pure software is uninvestable now. Natasha, do you believe in that?

Natasha Malpani (Boundless Ventures): No, see, I think there's three AI camps, right? One, this is the first AI camp to say, oh, you know, now all the value is in physically AI, you know, atoms, right? Sexy, new sexy word also, to say, oh.

Utsav Somani: That was named the company atoms.

Natasha Malpani (Boundless Ventures): Yeah, exactly. So to say we only look at atoms, so that's one camp saying, okay, forget software, now all the alpha is in hardware. The second camp is to say what I'm saying, that actually, listen, a lot of software doesn't work, it's broken in production, even a lot of consumer AI, what is actually taken off, we need to fix the infra for, we need to fix the infra for, you know, AI to work in the real world. And then there's this third camp that says, you know what, forget, actually a lot of the value is already being taken by the foundational model companies and hyperscalers. So even if you build very interesting products, actually, in the end, most of the value will accrue to the foundational model companies, you know, because they control the models, they control the margins, they control the pricing. There's three very different camps. And you know, there's truth, there's some truth to all, but definitely, I mean, we will, so I'll give you another example, right? Like super health is a mix. You know how you said that one thing is to say, have founders completely rethink their workflows versus saying, how can they be more efficient? Now, super health is in the opposite. They're saying, let me actually create a offline workflow that works first, right? So if super health is multi-speciality, small format hospitals, 50 to 100 bed hospitals run by Varun Dubey, you know, again, I think outstanding founder comes from Apollo Hospitals. What he said is that, listen, it's not, and he is building, you know, AI native hospitals, rethinking workflows using AI, but he's saying that first, let me actually rethink the workflow itself. So that's, I think, the beauty of AI is you stop and you're saying, listen, what was the work I was doing? Because a lot of knowledge work, right? Like whether it's finance, law, obviously, I'm a VC, a lot of the work was coordination. Like, what do we do? We make slides, we send emails, we coordinate with each other, we share knowledge. Now that the world's knowledge is differently stored and accessible, that coordination economy is gonna be compressed. So that's what I'm saying. Orgs will be redesigned, where you don't need so many levels. Isn't it amazing that you and I, we already have the world's intelligence at our hand? Like how are my under 25 founders building, you know, satellites? Like who taught them? You know, like they can just take, they can ask more than the smartest person in the world directly.

Utsav Somani: What happens to the education system? Like, I mean, should people go to college in the next four years? Like when you can literally ask a chatbot, I think the only thing that's left is figuring out interpersonal skills, maybe just asking better questions, driving your curiosity up, and then everything else is just, you can learn pretty much on the fly.

Natasha Malpani (Boundless Ventures): Yeah, I don't know if you can tell, but I'm a massive nerd, right? So like for me, that is the point of college is to learn how to ask the right questions. That is learning. Like that is actually what's broken, right? Like university or learning or knowledge is not getting the right answer or memorizing it. It is saying that what is the right question to ask? It is imagining what is possible.

Dhruv Sharma: You also have an affiliation with Plaksha, right, Natasha? Yeah. Are you comfortable sharing what's happening there, for instance, in terms of just changing?

Natasha Malpani (Boundless Ventures): Yeah, universities themselves are changing. I'd say one is, of course, you know, I think people are just questioning going abroad, right? Because of all the visa and immigration issues. And then two, the quality of education is changing in the country. So, you know, whether it's Ashoka for liberal arts, Plaksha for science and tech, you are seeing these interdisciplinary universities, exactly like I'm saying, where they're saying that, listen, our goal is to help our students think better. And that is also, like, see, again, we all have the world's intelligence that I have. Everyone has different levels of using AI, right? It's the same starting point. I mean, to be able to ask the right questions, to be able to imagine what's possible, to be able to build those systems, no, that is what a great education actually helps you do. So I do think one thing that's gonna be interesting is the Socratic method, which is that you keep asking questions and you go deeper, deeper, deeper, right? That is gonna be a very popular way to learn. But also, by the way, again, like in edtech, edtech is exploding. I don't know if you guys have seen this or not, but, you know, edtech is hot again, right? Because everyone's asking this. And it's not just how will you learn at a college level, it's how will, you know, how will toddlers learn? So AI toys are popular. How will teenagers learn and interact? How will you, even, I'll tell you like one joke actually, is if you really wanna understand someone, is you look at their prompts. Like I feel like the future of dating or hiring someone, your prompts will tell me a lot more than what you ever will.

Dhruv Sharma: Because you're having a Socratic dialogue with the AI.

Utsav Somani: I never want anyone to read what I prompt or my chat GPT data, if it gets leaked, boss, I'm screwed. As a final question, Natasha, the Boundless name, we've not discussed that. You've kept the name consistent in your journey. You did Boundless Media, you did Boundless as your book of poetry, and now Boundless Ventures. And as a bonus ending, maybe do you have a verse from your poetry that you'd like to recite on the show?

Natasha Malpani (Boundless Ventures): Honestly, I wish I had my book. I would be happy to read you a verse.

Utsav Somani: Let's pull up your notes and see if we can end the show on a fun note.

Natasha Malpani (Boundless Ventures): Yeah, but just to answer the first question, yeah, so, you know, Boundless is interesting exactly for that reason, right? Like Boundless is a concept that actually we are all boundless. Like there is no limit to our imagination and creativity. That's what I wrote that first poetry book with the idea of saying that, you know, whatever the constraints are only in our imagination and anyone is able to do anything they want. And then Boundless Media, of course, was a continuation of that brand. And then Boundless Ventures, actually this is twofold, right? So one, obviously my founders are boundless and they're creating reality. But two, Boundless is also interesting because we're a cross-border fund. So we are fund across borders. So that name is really stuck. And then again, just to tie the whole conversation together, that is the takeaway from the conversation, right? Then what are we constrained by today? It is just our imagination.

Utsav Somani: Awesome. And are you fulfilling our second request as well?

Natasha Malpani (Boundless Ventures): I just don't. And by the way, I've written two poetry books and a children's book, but I don't have the notes.

Utsav Somani: Maybe I'll share it with us and we'll share it in our community. Thank you so much for coming on the show, Natasha. Really a pleasure. Thank you. Really fun chatting with you. See you. Bye-bye. All right, listeners, that's our one and only guest today, but we've got one interesting news that happened, dropped on the weekend. So let's break it down. Elon Musk unveils TeraFab. It's the $25 billion chip factory with huge capacity, almost unimaginable for a mind. Dhruv, you tell him, sir.

Dhruv Sharma: No, no, you have to tell him. But I was just thinking, what's it like? Is Elon the biggest newsmaker in tech or what? Is anything new happening? He has to jump in head first. If or...

Utsav Somani: You know, like, I think all of this is timed very... I mean, you never know if it's because of China, Taiwan, if it's because his SpaceX IPO is coming up, he's merged these two companies together. Of course, he needs, I mean, a whole bunch of compute for different reasons.

Dhruv Sharma: Companies are morphing into becoming this super organism, right? Like, what's the... It's going to be called the X-Corp.

Utsav Somani: The Musk company or the Musk holding company. They have a different name going around.

Dhruv Sharma: Ironman, yeah.

Utsav Somani: It is insanity. Like, I mean, he's merged all of these together and now he claims that if this whole project fortifies, he will basically be 70% of TSMC's current global output. And all the existing fabs can only supply, what, 2% of our Tesla SpaceX.

Dhruv Sharma: They're also talking about vertically integrating every aspect of manufacturing semiconductors from design to lithography to fabrication testing, everything in one place.

Utsav Somani: And it has to happen because what he needs, that one trillion compute, like, I mean... Sorry, what's the number there for the compute capacity?

Dhruv Sharma: I'll just look that up. But he needs the chips, inference chips for the cars in Tesla. And then, of course, the orbital data centers will have to be commissioned at some point in time. So they'll need silicon there as well. Everything.

Utsav Somani: And he thinks that it's always sunny in space, so 80% of his chip compute and everything. He's basically placing a bet that energy on Earth will go down and will not be able to support the future compute capacity. So he wants to put a lot of his resources towards space, and that's the vision that he's pitching towards. And for some, I mean, they might be skeptical of this whole thing, but I think Elon Musk has a way of willfully bringing these sci-fi stuff to reality, right? And that's what we've... You'd never bet against Elon.

Dhruv Sharma: Yeah, he's a little off on the timelines, but who cares? Things that finally he wills into existence, the Starship, for instance, I think most of those things are marvelous. But there might be a supply chain angle here as well. Which is, I mean, given what's happening on our West, everyone's thinking, hey, where's the next trade of Hormuz and one such just bottlenecks so happens to be in Taiwan with the SMC. And I think people...

Utsav Somani: But a lot of people are saying that 25 billion won't be enough. Like I think we need a lot more in terms of capital. And all of this has not been accounted in the Tesla spend capex projections.

Dhruv Sharma: Well, they're saying 20, 25 billion right now, that's a trillion dollar company. But also like, I mean, like TSMC is led spend over four or five decades. And so to expect someone to front load all of these costs and just like, we'll see. We'll keep an eye out for all.

Utsav Somani: Elon Musk, yeah, he's played with physics before and come out winning. So why not?

Dhruv Sharma: I think he must be poaching talent from left, right and center just to, yeah.

Utsav Somani: I think that's a good techno-optimist note to end the show on. And we'll see you on Wednesday. Thank you for tuning in. Bye-bye.