Physical AI is more than robots, says the Nvidia veteran now investing in it

0:00Hello and welcome back to Equity, TechCrunch's flagship podcast about the business of startups. I'm Dominic Midori-Davis and this is the episode where we bring on industry experts to help us explore a trend in the tech world and dive deep. In the past few weeks, news broke that robotics startup Field AI is reportedly eyeing a$700 million raise at a$10 billion valuation, while a German robotics startup hit a unicorn valuation earlier this month. The money pouring into physical AI is getting enormous, but before any of these massive rounds, someone has to write the very first check. Today's guest is making her bets much earlier, and she's been in this space longer than most.

0:33Back in 2018, years before anyone was even calling it physical AI, she was marketing NVIDIA's first robotic software.

0:44Today, we're joined by Amulya Vishwanath, founder and managing partner of Techable VC, an early stage firm investing in physical AI and frontier scientists. Amulya, welcome to the show. Thank you, Dom. and on that note I'm curious do you prefer Dominic or Dom? Oh um either is fine you can call me Dom it's like shorter. Awesome. Something that's actually quite interesting is that you were at NVIDIA in 2018 which doesn't seem like that long ago but I guess in AI years actually is kind of like a long time ago and well my first question of course is like what made you leave like what did you see that was happening at NVIDIA that made you say wait I should probably break out and do my own thing and start backing like the next generation of this?

1:28So I've been studying deep neural networks for about 16 years now. And my engineering background has been in electrical engineering. And I also ended up doing an MBA. But really, over the years, I've noticed that the semiconductor industry has evolved quite a bit from just being known as semiconductors to computing. And my very early career, I started out as a signal processing engineer designing radar systems and then went on to joining Altera, which got acquired by Intel. So I ended up leading technical product marketing for a lot of the signal processing products. And this was back in 2016 and very early stage of thinking about AI deep learning.

2:08So I also was fortunate to launch OpenVINO toolkit for FPGAs, which was Altera's strategy in the overall portfolio by Intel. And from there on, NVIDIA came calling in 2018, and it was one of the best opportunities that would have showed up in my career, I would say. And since 2018, I've been fortunate to lead product marketing efforts for some of the most foundational and core products, including all of the libraries, pre-trained models that we brought in mainstream. And back in the day, it was more research focused. And now you see that people are no longer talking about models, They're talking about core applications that can really enable newer trillion dollar markets.

2:51And so was this some of the stuff because like because what made you what was the decision or what was the moment that made you want to leave and kind of launch and do your own thing? Yeah, I always had an entrepreneurial spirit and it was a matter of time. And I'd already seen and built a lot of the community that is now building and developing in super intelligence or artificial intelligence, as you can think of. But really, it was a point where I realized that I've taken on some tremendous challenges in my career since the beginning. My first capstone project was actually working on neural networks that can do image processing.

3:29So since then, I've seen AI really get out of the R &D phase to becoming mainstream. And as a result, I realized that if I've seen and built a lot of these things and I've supported a lot of the corporations and startups from a very long time, I think it's time for me to be my own Jensen and really go off and support the ecosystem from ground up. And the combination of building the ecosystem, working with a lot of the engineers and researchers from ground up, and to even launching one-of-a-kind, first-of-a-kind products, including NVIDIA's Isaac Robotics platform, all the way to leading, expanding, emerging ecosystems while NVIDIA was going to get much bigger and expand its footprint into emerging markets.

4:14I think these are some tremendous opportunities that I never imagined I would get in this lifetime as a professional in the computing industry. And I've been fortunate to be in the front row and leading a lot of these efforts. And now I take all of that experience and the learnings that I've seen and absorbed from leaders like Jensen, Ian Buck, Deeputala at NVIDIA. And now I started Techable Ventures. What is physical AI to you? Like, what was physical AI back in 2018? And what is it now? because it seems like it's everywhere right now. Yeah, so back in the day, it was essentially robotics, right?

4:51And if you think about it at a company as big as NVIDIA, if you go back and look at any of the keynote slides from NVIDIA's conferences, specifically GTC and so on, robotics and autonomous vehicles have always been a part of their bigger vision. And a lot of it back then was centered around data centers. But that said, things have changed since then. And my definition of physical AI is not necessarily using it as a synonym for robotics. It's more than that. We're thinking about how materials are reinvented with AI. We're thinking about how robotics is now becoming more and more mainstream. Back in the day, it was more pre-programmed robots that could do specific things for specific applications.

5:36So now we can think much bigger and broader in terms of the robotics vision by itself. So it's not confined to robotics. It's more than robotics. It goes on to material science. It goes on to even life sciences for that matter. And that's why I built Techable Ventures, because this culminates these three key pillars where you can think of, you know, physically as anything that has been on the digital front now comes into the existence as a physical form. It could be a molecule, it could be a material, it could be a medicine, or it could be a robot that that's in your house or running across a factory line.

6:12Yeah, it's actually quite interesting because I feel like there's been so much. I feel like you talk to like a regular person and they think about the dancing robots. And I just covered, I think Vogue World brought out dancing robots for like the finale of whatever they were doing in Milan. And it was funny because, you know, I was talking to kind of a guy. She's like a consultant or a PR person in the industry. And she was like, this doesn't even like capture what's happening in the industry in terms of like how physical AI is intersecting with all these different industries. And I was like, that's so fascinating to think of just what's happening and what's bubbling underneath all the surface.

6:48Because, I mean, I don't know, regular people might think of like the dancing robots, but it seems like there's just so much happening. I mean, what are some of the most interesting things that that you're seeing right now? Yeah, absolutely. And, you know, part of robotics being so mainstream is, essentially looking at the cool and cute things, as you can think of, right, those fuzzy robots, the dancing robots, and so on. But then also, there's a lot of mainstream robotics coming into play in terms of factory lines, healthcare, and really where there is a need for conscious automation. And some of the investments that I've already made through Techable Ventures has been in industrial robotics, because what's happening is that a lot of the SMB factories, people are deciding to shut them off because there is a lack of labor interest in general, because people have moved on to newer jobs.

7:41And in this AI first era, what is happening is that the choice of younger generation in terms of what they want to do versus the choice of, you know, like the boomers, so to speak, in terms of what they want to pursue, or even retiring for that matter, has really left with this huge gap. And a lot of these family run businesses are now struggling or figuring out, you know, how do we address those gaps? So that's one area. The second thing, and this is an interesting one, and, you know, I've been researching a lot about it and also ended up making an investment in one of my portfolio companies called Andromeda Surgical, is that surgeons face one of the highest burnout rates because what happens is that when you're assembling complex surgical teams, especially, you know, with the right expertise for complex cases, it takes time to attend to patient care.

8:31It takes months before a patient gets attention from the surgeon and their special unit for especially complicated cases related to cancer and so on. And as a result, it's important to bring this conscious automation where it can be as simple as an autopilot for a surgeon so that instead of doing 10 cases, and interesting fact is that in the first minute, you can tell if the surgery is going to be a success or not, which becomes, again, really critical, right? You want to deliver, you know, the most precision care, but at the same time, you want to make sure that you have specialists who are caring for a patient without having to, you know, like really see a miserable state of a patient, for instance.

9:21So with that said, you know, I think, you know, including a co-pilot for surgical use cases might be a very interesting space. And this is something that I've been pretty excited about. Well, I just I want to go back to one thing. What is conscious automation? A lot of the hype has been around, you know, everything is getting automated, right? And that is that should not be the case. And again, like I said, you know, there are huge gaps in terms of the labor market. There are huge gaps in terms of, you know, areas where you need to deliver care in a critical and a careful timeline. And also, there's a lot of aging population, and it's hard to address the needs that any of these categories need by just bringing in human labor.

10:06That said, automation helps bring in these specific robots or autopilots. Again, it could be a software agent or a physical AI agent like a robot that can simplify and really add productivity and augment an existing use case. So it could be in healthcare, it could be in material sciences, it could be moving materials from one place to another, or it could be, you know, attending to a patient, or I just had to go visit my mom. And the fact that, you know, she has her own struggles as she's aging makes me think that, okay, there are so many promising use cases for robotics to help and enable the aging population in general.

10:49No, that's actually quite interesting. It feels a little bit like a hot take too. I feel like you're talking, like you talk to investors and they're like, everything needs to be agentic. Everything needs to be AI right now. But it really is like the idea of finding like the meaningful things to automate, not just like spray automation, just like finding what's important. Yeah, absolutely. And again, you know, like I said, there's so much happening and there's such a huge gap in the market as well that needs to be addressed. And which is why I take the approach of, you know, having automation, but in a more conscious sense.

11:23And everything is tied to a dollar amount. Everything is tied to looking at, you know, can it be economical and can it be scalable? And which is why it's super important to fund these early stage companies that are thinking and reinventing how robotics has been seen. And at the foundational level, there is a meta science reset that's happening. And the reason why I say this is that the way research is being done is now being reset. The way researchers are doing research, the process of it is also changing. And with that said, the type of companies that are coming together, everyone has now become an engineer, right?

12:02If you think about it, English has become a programming language, which enables anyone to build anything and really question the status quo right now. And that's a fascinating time to be in. And as an investor, during this time period, I'm actually quite curious to know how, because you have so much experience in this space. And there's obviously generalist VCs who also just want to get in on the hype. And so how do you kind of use the experience that you've had in this space to kind of cut through all of the hype or the founders who are just, I don't know, saying, I'm just going to build a robot that's like that can do dishes or something.

12:38Like, how do you use your perspective to really cut through things and also have a competitive edge against the other VCs in the industry? Yeah, absolutely. I think we've jumped from talking about perception AI to generative AI, agentic AI, and then right into physical AI. And a lot of companies and experts have been talking about the chat GPT moment for robotics. But I also want to say that apart from all of the exciting things happening in the chat GPT for robotics era, there's actually... Has it happened yet? We've been talking about it. There's been quite a bit of progress. But that said, there's a Lego era that's happening for robotics.

13:18And what do I mean by it? There are a lot of foundational software and building blocks for robotics that are being rethought right now as we speak. And it's super important in order for innovation to scale. I think it's important for these foundational building blocks for robotics, whether it's actuators, whether it's, you know, type of robots and, you know, anything and everything beyond humanoid robotics. And also what type of applications, which are the newer markets that will be disruptive, but also which are the existing unsexy markets that can see conscious automation as well. Right. So that said, there are a lot of foundational problems that need to be solved for robotics, and which is why I see this as a Lego era for robotics.

14:03That's the first thing. And then the second thing is that I've seen how robotics has been reset within companies like NVIDIA. The strategy has changed by the day, and we've seen mainstream adoption now in terms of, again, all the technologies that NVIDIA has built, but also the robotics strategy that is really helping. And one of the core pieces about robotics is that, you know, simulation is going to be more and more critical. And one of the products that I did technical product marketing for at NVIDIA is called Isaac Lab. But I also led the product marketing efforts for humanoid robotics. And you've probably heard about NVIDIA's Groot or GR00T, as we would like the world to know it as.

14:46So I led a lot of these product marketing efforts, so got to work with a fantastic set of founders, researchers internally, externally, and engineers. And I realized that we're just getting started. And in order for this to happen, I think there needs to be more emphasis on open source safety and newer standards that need to be set for robotics in general. and the approach that I'm taking with Techable Ventures is to be a specialist VC firm because I am a product of an engineer and a doctor to begin with and if you think about the thesis it comes together as you know there's one lever which is around robotics the other one around material science and the third one around life sciences so if you look at you know how physical AI is evolving these are the three core levers that are going to see a lot more disruption and that's our fund one thesis and the second thing is that you know there are fantastic investors out there and i love collaborating with all of them and just based on our existing portfolio companies i have a lot of the lead investors and co-investors reach out to me to get my input about companies that they're evaluating how do we help our existing portfolio companies and so on which again you know says that you know i've seen and lived this experience in the physical AI market.

16:04And that allows me to bring that unique insight into not just to the founder community, but also to the investor community. And one of the things that I wanted to do with starting Techable Ventures is to have the founder think that his or her VC is the moat. And that's where I'm headed. And more and more, if you think about it, there is a lot of capital. So if you think about how founders talk about it, there's smart capital and dumb capital. And I want to be at that area where a founder can make a call even before they start building. But at the same time, you know, I am a part of their moat.

16:44That's really interesting. So what does that mean? What does that mean exactly? Yeah. So like I said, you know, I've been a part of the early journey in terms of AI deep learning becoming more mainstream. So back in the day, during my time at NVIDIA, I started out as the founding product marketer to lead a lot of the AI software training, SDKs and models marketing. And that said, you know, I've built this ground up network of engineers and researchers. And as a result, what has happened is that, you know, doing due diligence for me is much faster. Making introductions for customers or pilots to my founder community is much faster because, number one, I've already worked with a lot of these pilot customers or partners that I can make quick introductions.

17:29And at the same time, you know, a lot of the founders are people that I've already met or worked at one of the technical conferences in the past. So this just makes it a lot more easier for me to be the strength or the first call for founders and for them to even just validate the idea way in advance, even before that idea turns into a company. No, you know, what's actually really interesting is I've been talking a lot more to investors who've been saying that, you know, it's not about the money anymore. Like there's a lot of money out there and it's not necessarily even about talent. And they're saying like what they've been refocusing a lot of their firms around is helping with like distribution and networking and all these insights.

18:09I think that's so interesting that, you know, we've reached a time period in the industry where there's just so much money, right? It's founders are looking for so much more. Absolutely. And you're right in terms of the distribution channel, right? it's not necessarily because you're a VC or an engineer, but it's also because can you help your founders tell the story? Can you help your founders find the best talent and so on? So it all starts or goes back to distribution channel. And fun fact is that when I started out in 2016 on LinkedIn, I had a little over maybe 50 connections. So every connection that I've earned on these platforms has been because of the work that I've done in this ecosystem.

18:49So most of my network are these core researchers and developers that I've worked with. I've written a lot of the technical blogs about products that have now become mainstream. And at the same time, you know, I've had one of the unique opportunities where I've gotten to work with Jensen directly in third month of my time at NVIDIA. So, you know, it gives me this unique perspective, but also, you know, to your earlier point about, you know, working at a company like NVIDIA, it's like working at a company that is, you know, running at light years right now. And it's been a fascinating journey to not only support founders while I was in the ecosystem, but I continue to exist in that ecosystem.

19:30And now I get to be an investor and help them do their fundraise or help with their product positioning, even thinking about their partnership strategy, for instance. So all of these components become critical. And in order for them to get the best talent, position their product, and at the same time, you know, work with the right set of investors is going to be crucial. And all of these are so much more important and capital is too, but you know, distribution and strategy behind it, which is why I say that your VC should be your moat. And I hope that founders can think through that very seriously because capital takes you to a certain distance.

20:11But beyond that, you want someone who's lived the problem. You want someone who can help you find the best talent, help retain them. At the same time, you know, can help you tell the product story, can help you identify the right partners to scale and go from there. And all of these things are things that are done. And I'm excited to bring it back to support the founder community. And when it comes to your founders who are building in this space, what are some of the biggest bottlenecks that they're running into? I'm thinking in terms of like, is it manufacturing? Are there regulatory problems?

20:40Is it finding compute and energy? Like what are some of the biggest problems that they're facing right now when it comes to building and scaling? I would say it's a combination of all of these, right? And that's fantastic. These are the problems that are needed in order for us to really build a scalable set of programs and ways to enable the entrepreneurship community. And as a result, you can already see that there's so much happening within the US ecosystem. We're seeing companies that are building actuators. We're seeing companies that are building variety of robotic hands, for example. We're starting to see companies that are reimagining how materials come into play with, you know, the semiconductor shortage that we're talking about.

21:21We're seeing robots being built, general purpose and special purpose robots that are now going to be, you know, US made. So that said, I think it's important to have supply chain within this ecosystem. And at the same time, I think the challenges are one thing, but then, you know, having the right talent, retaining that talent is also going to be very critical. And now is the time where you're seeing a lot of the talent going into some of the bigger corporations. But that said, you know, I see two polarized ways. One is where you're seeing a lot of talent going into some of these bigger corporations like the Anthropics and OpenAI and NVIDIAs of the world.

21:58And then you're also seeing a lot of people leaving these existing corporations to solve the foundational problems across different domains and industries and build something on their own. Same thing is happening with the university ecosystem as well. A lot of people are getting into building and this is such an exciting time. So I'm sure, you know, this is a pretty strong ecosystem. We've done this during the internet era. We've done this long, long time ago with the industrial revolution. I think it's just a matter of time before things fall in place. And which is why I say it's the Lego era for robotics.

22:30But I feel like when it comes to manufacturing, you can't not mention China. And so where is the U.S. actually winning when it comes to a lot of physical AI production versus what do you think China is winning on right now? And how do you see the two ecosystems playing off of each other? Yeah, I remember during my time at NVIDIA, I did extensive research about, you know, where is a lot of work or publications coming from for some of these key areas, including robotics. And we definitely saw a lot more happening in China in comparison to other countries. But that said, in the U.S., there's been a strong talent pipeline.

23:09And we want to make sure that, you know, like we offer them the right set of programs to support if they want to become an entrepreneur and also have channels for them to get into some of these bigger corporations. I think talent has and is being a moat for U.S. in general. China has always had this position where, you know, they've been manufacturing experts for a very long time. But that said, I think the fact that we're bringing in super intelligence by itself and there's so much emphasis within the country, you could say already that there's so much happening. You're probably seeing a lot of it when you see it in the news or when you hear it from one of your friends or colleagues.

23:52But that said, there's so much more that needs to get done within this ecosystem. At the same time, it is going to be a race, but at the same time, it is going to take a while. I think the way superintelligence has been evolving, I'm pretty sure we'll be able to accelerate a lot of those development and discovery to development timelines in general. There's going to be a global ecosystem, so it doesn't have to necessarily be centric around one country. In order for us to have a lot of these materials or life sciences development or robot development, a lot of these big corporations have presence beyond U.S.

24:35as well. So it's important to think about this as a global ecosystem building towards what I see as the newer civilization or the more conscious civilization. That's kind of how the thesis that I built out. You know, there's so much happening, so much happening in the US, happening in China and in general within the ecosystem. Founders are pretty excited. We already see a lot of founders building or thinking about building. We see a lot of enthusiasm, even in the bigger corporations. Back in the day, they were slow to move and would think twice or thrice before working with startups are now taking bets on startups and building this three way, four way ecosystems with university entities.

25:19the startup themselves, a bigger corporation, and a cross-domain entity. It could be like an NVIDIA, a university, a startup, and an Eli Lilly, for example. So we're, again, you know, seeing a lot of these resets, and which is why I say we are going through this Lego era. And it's a fascinating time to see robotics, life sciences, and material sciences being relooked at. That is so interesting. I think I was talking to someone very recently about it. I was like, Like we see coming out of China right now, like a lot of the dancing robots and like, I think the robotic Olympics was over there, right?

25:52They had something. And I was like, no, there's definitely something else happening. Like this is, you know, they're funny to see. I think, what is it? I don't know. The dancing robots are just so funny. I have no idea. But I was talking to some academics and they were like, with humanoids, they were like, I think the best use case for them right now is entertainment. And I thought that was really interesting from an academic point of view, because I I was like, that can't just be the only thing humanoids are good for. But maybe, but maybe, but they can't be. I don't know. Well, I could say this much that I've seen non-humanoid robots also dance.

26:27In fact, back in 2018, I did see this artist enterprise put together a robot powered by NVIDIA Jetson, which was just, you know, like going around the dance floor, you know, with different movements and was not a humanoid. Sure, you can see some entertaining use cases, but that said, humanoid robots in general are a fascinating form factor because we as humans are used to seeing this form factor. And that's nice because a lot of these existing spaces, buildings and outdoors are designed primarily for us. but that said there's more beyond this form factor and there's there are more use cases beyond just thinking about you know entertainment or for the fact that you know it's going to be running through factory lines and so on you know you can see optimus you can see neuro robotics you can see unitary you know releasing fantastic features with their robots or at least show them as demos But there's a lot more form factors that we will start seeing.

27:27And one of the things, and this is something I want to test it out, and I'm sure probably there are companies that are thinking about it, is that imagine if you wanted to start a business, you know, 10, 15, 20 years from now, and you could maybe commission a set of robots from another entity and specialize these robots to either, you know, do all the lawn cleaning. or maybe, you know, move products from point A to point B within the factory line. So I think that there's going to be a combination of conscious automation, but that said, also conscious entrepreneurship that will involve robots working alongside with us to start new businesses.

28:10I guess like my final question is kind of like, what do you see as the most successful near term area when it comes to physical AI? Well, the definition of near term could be shifting based on who you asked this question to. Right. And also the definition of physical AI has been shifting. And like I said earlier, you know, physically, I is more than just robotics. I'm really excited about, you know, where things will lead in terms of solving the foundational problems with robotics by itself. Right. Having a more easy to use, deterministic and safe set of products that can enable the ecosystem.

28:46to build for different industries, whether it's regulated or otherwise too. So that's one area that I'm excited about. Safety standards are going to be more and more important in terms of, you know, how you think about robotics within houses versus outdoor robotics and so on. And the third thing that I'm excited about is healthcare. I think in general, there's so much that we could be doing in that space and which is why the culmination of conscious robotics, material sciences and life sciences put together is going to be one new disruptive space that I'm investing in. And we're just about out of time, but where can our listeners connect with you and find your work?

29:27Yeah, so you can find more information on techable.vc website. And you can always email me amulya at techable.vc. Thank you so much again for joining us on the show. This has been a really fun conversation. I'm now looking out for the chat GPT of physical AI. So I'd like to have my eyes and ears open. And for our listeners, you can find me at Dominic Midori on Instagram. Please follow me on Instagram so I can get more followers. And Equity at EquityPod on X. Thank you again. Talk to you next time. Equity is hosted by TechCrunch Senior Reporters and produced by Teresa Locansolo with editing by Cal.

30:07Subscribe on YouTube or wherever you get your podcasts and Find out what's next at techcrunch.com slash events. Thanks so much for listening, and we'll talk to you next time.