theCUBE + NYSE Wired: Physical AI & Robotics Leaders | Krishna Rangasayee, SiMa.ai

Clips
More from theCUBE + NYSE Wired: Physical AI & Robotics Leaders

Speakers

Krishna Rangasayee

Founder & CEO

SiMa.ai

play_circle_outline Advancing Industry Action: The Power of Generative AI, Machine Learning, Robotics, and AI Convergence

play_circle_outline Shift towards multipurpose robots enabled by generative AI and LLMs

play_circle_outline Move from monolithic to orchestrated systems in robotics environments

Info
Transcript

Krishna Rangasayee, SiMa.ai

Krishna Rangasayee

Founder & CEO SiMa.ai

Krishna Rangasayee, SiMa.ai's Role in the Advancing World of Robotics and AI

Krishna Rangasayee, founder and Chief Executive Officer of SiMa.ai, brings expertise to theCUBE's discussion at the NYSE Wired Robotics and AI Media Week. As key topics such as machine learning, generative AI, and robotics take center stage at this exclusive event, Rangasayee provides invaluable insights into the evolving landscape of intelligent systems. The conversation is hosted by theCUBE Research analysts.

In this video, Rangasayee discusses the transformative ... Read more

explore Keep Exploring

What is the significance of machine learning and generative AI in the robotics and industrial automation industries? add

What role will generative AI and LLMs play in the evolution of robotics in the next decade? add

What are some new possibilities and use cases that could be opened up by the increased flexibility and capabilities of LLM and generative AI infrastructure in robotics environments? add

bolt Powered by CUBE AI

Krishna Rangasayee, SiMa.ai

search

>> Welcome back, everyone. I'm John Furrier with theCUBE. We are here at the NYSC. It's theCUBE East location. This is where all of our action's happening in Wall Street, and in the stock exchange, part of the NYSC Wired community and open community building trust amongst experts. Again, the focus is on robotics and AI this week, coming off GTC from NVIDIA two weeks ago, robotics center stage. And of course, machine learning and generative AI is at the center of the action. Krishna Rangasayee is here, founder and CEO of SiMa.ai, Cube alumni. Great to see you again. Krishna. Thanks for coming back on theCUBE remotely into our studio. Great. Great.

Krishna Rangasayee

>> Thanks, John. Pleasure to be here again.

>> Great to have you. So first of all, we had a great chat on theCUBE. For the folks who want to check that video out, go to the YouTube, go to theCUBE.net, check it out. You guys doing a lot at the chip level, going on with machine learning, the generative AI wave. I think last time we talked about the AI infrastructure, really, the innovation at the chip level. Okay?

Krishna Rangasayee

>> Right.

>> Okay. We were right on that one. Pat ourselves on the back. You've been right. Doing great in business. Hard to call. That's a hot take, I'd say. But if you look at what NVIDIA did at their GTC conference, it was really the center of all the industry. Of course super computing, open computer, all doing great work. But you're starting to see visibility into the systems. I think we called them clustered systems. They're now called AI factories. Robotics was also center stage. Which is kind of a north star for where gen AI goes, because you got to have everything working in real time. Robots got to be intelligent. They have to reason, infer. They have to be intelligent. So the crossover between robotics and what you're doing is pretty significant, because AI's at the center of it. What's your take on that? Because as that becomes accelerated, that's just forcing everything through. What's your take?

Krishna Rangasayee

>> Absolutely. I think you eloquently captured it. I think if you look at robotics and industrial automation, that market segment's been around for 40, 50 years. And the big shift that's happening at AI adoption is really now going to be around generative AI, bringing in really amazing reasoning capability and decision-making capability to traditional robotics. So you're going to not only see a refresh of classic robotic industries and industrial automation, you're also going to see a new class of robotic environments. Call it embodied AI or physical AI, but fundamentally, generative AI enabled infrastructure for self reasoning and for adaptable robots is going to be a new trend.

>> One of the things we're seeing come out of the AI leaders around robotics is the use cases that, I won't say a pedestrian, I would say they're more real world. I interviewed a company that's making a therapeutics device. It's simple device, but it's also got hardware and software innovations. And so, they've essentially built a new kind of ecosystem. So that combination of hardware and software allows them to plug in LLMs-

Krishna Rangasayee

>> Yep...

>> in. This is a new kind of business model venture architecture, where the innovation is looking at the hardware as an enabler. Certainly it's beachhead in a business when you've got hardware robotics, but the system has a hardware software element where they can actually develop a solution, but then use AI to bring software in to run on that. So that's like an operating system running on their device that if you go back and previous generation, Krishna, that's going to be a purpose-built device, and then you buy another one. So you now have infused capabilities coming in from ecosystems powered by open source. What's your reaction to that? Do you agree? Is this a new thing? Is this something that you see as scalable?

Krishna Rangasayee

>> Absolutely. And so I mean, you clearly researched the space fill. You clearly are speaking to everybody in the industry. But you captured it well again. And I would say no doubts, every new trend goes through a hype cycle. But I think this time it's really quite real. And just like as you mentioned, I think people are moving away from single-purpose robots to, if you will, multipurpose robots. And generative AI and LLMs really have opened up a new playground for what robotics can do, and it's super exciting. And in my mind, over the next decade, between 2025 and 2035, most every robotic infrastructure, either classic groupers or robots, warehouse logistic robots, or humanoid robots, that's an evolving field, are all going to be AI-centric. And they're all going to be reasoning capable. And LLMs slash generative AI are going to play a very key role in enabling this new capability. And as you said, the hardware and the software combination AI enable, opens up new business possibilities and new technical possibilities.

>> Yeah. One thing I found interesting is at NVIDIA's conference, I heard the word computer science on stage multiple times, and I also heard the word operating system for your AI factory. Now I'll pause there, just make that statement. But what we're starting to see is that system operating model in software, okay? Not, here's the firmware, or here's the glow-level code. You're starting to see orchestration kicked around. You're hearing words like operating system. We're talking about a system architecture here. Can you share your view on this? Because when you start getting down to now the hardware being intelligent and being, I won't say portable, but compatible with new things, what does that open up for opportunities? Because-

Krishna Rangasayee

>> Absolutely....

>> it seems like a new wave of entrepreneurship. It's a new wave of engineering. What does it mean?

Krishna Rangasayee

>> Great question. So I think if you really look through classic robotic systems, I would call them monolithic. And the desire to really orchestrate across the entire infrastructure of a factory floor automation, or maybe across multiple warehouses, has always been there. But the hardware capability to enable that has been very limited, very limiting. What you're now seeing is really not only the flexibility of what a LLM or a generative AI infrastructure could do, deploying, if you will, very complex functions in a robotic infrastructure. Robotics environments become easier. Now the model is going to move into orchestration. And to answer your question in terms of new possibilities, it opens up new use cases, no doubts. It opens up new possibilities. But also opens up new companies that could really now enable and accelerate a scaling of AI at the edge. And that's part of the thesis behind what we built our company around. But I also would say, this is not just the frontier for the new companies alone. Existing market leaders will also be going down this path. So you're going to see a massive shift in the industry, from monolithic self-contained environments to orchestrated systems, and really orchestration across the factory floor in robotic environments, but maybe even across the entire fleet, entire set of warehouses globally.

>> What's your view on the whole supply chain? Again, two-sided marketplaces exist. You got a supply chain to manage. You also have an ecosystem, because on this thesis you've got an open source, you have developers, so there's a lot of white space for opportunities. Check. But now you've got a supply chain, that with AI, you could have a steady stream of, I would say, distributed components or distributed suppliers. And so, that's key. And then also, you've got potentially disruption with tariffs and other kind of geopolitical things. So I know it's a lot in there, but unpack that. Because on one hand, let's lay out the opportunity. And two, what does the impact to say, regulations and compliance and tariffs?

Krishna Rangasayee

>> Yeah. So we already had a lot going on technically, now there's a lot going on business, and then there's a lot going on geopolitical, and all three are interlinked. So it's really hard for anybody to parse through what the future is going to look like. But what I think is a given, is that absolutely AI and robotics are going to be the next big thing that's going to be shifting. I think we have had from 2015 to 2025 cloud and AI. Between 2025 and 2035, you're absolutely going to see AI scale at the edge. Robotics is a poster trial application, but you have automotive, you have aerospace and defense, you have smart vision systems, medical. There's an entire plethora of physical AI applications that are going to be impacted by it. And given the scale of where it is at, given the scale of how global it is, and also given the scale of it touches human lives, regulation is going to be a large component of what's something we have to pay a lot of attention to. No doubts, geopolitical climate's going to be quite complicated. It's our view that I think the western hemisphere will probably start figuring out how to manage itself. And there will be an environment in China that's really also going to be self-managed. And it's really going to be very fascinating to see how fast things are evolving. And the last statement I would say is, open source is really a democratized AI. The ability for multiple players globally to deploy complex AI systems is really become more accessible. So a lot happening. I would say, there isn't any one human being that has a good picture of what the future's going to look like. Lord knows there's a lot of opinions, but a lot to sort through as you just said.

>> I mean, there's a lot of science and computer science involved, which I love, why I love this market right now. Because if you think about the intelligent edge, the way it was defined, go back 10 years ago, it was basically IoT in factories. Now with robotics, you now have everyday life is impacted. So the hardware-software relationship is critical. I do think your point on open source is so right on, because with open source, that's just more, I mean, developers and domain experts that could solve a problem that could run on something. So when you talk about the real-time edge, whether it's computer vision, or an LLM, or whatnot, they're dealing with multiple contextual data feeds. Right?

Krishna Rangasayee

>> Right.

>> So take us through that state of the art, because I think all the discussion over the past two years has been training. You do a lot of NVIDIA, you train the data. And then you got edge for inference, but now inference at scale and then inference across multiple inputs coming in, including new real-time information coming in from the edge. So you have one, net new information coming into a device. It's never been trained, or it's just new information. It might have some context because it's at the edge. And then multiple data sources coming in. It could be from this group or that group. This is a data problem in real time.

Krishna Rangasayee

>> Absolutely. And I would say, listening to you, my ability to add too much intelligence behind what you already said is going to be limited. But you're entirely right. You're entirely right in that I think we have gone from a training world to an inference world, to an inference at edge world, and now context-aware inference at scale. So that's really been the evolution of other market is, a key enabler for all of this, is really having the right hardware capability at the edge. So when we started SiMa, we knew that the democratization of AI from a software environment like we talked about is going to happen, but there was not really a purpose-built hardware infrastructure for the edge. That's really at the very key of what we are doing. What's going to be eventually important is a scalable hardware system that supports the evolution of AI. AI systems are changing once every six months. Hardware has to keep pace. Not an easy thing to do. The second thing is power. Power is going to be very, very critical as you deploy things at the edge. The third one is really to make sure that there's ease of use and deployment. So anybody that cracks the code on this is really going to be the future market leader in this combined hardware system, hardware, software ecosystem.

>> Yeah. And a lot of development going on. I think it's safe to say that the hardware, software enablement, and then integration is key. Great call out there. Given kind of we're in this stage now, I got to ask you how your business is doing. Obviously, we're seeing proof points that the model works. You got hardware, software systems that can be deployed, workable, but extensible with whatever model you plug into it or whatever input. Check. What is your plans? Where are you guys now? What are some of the successes since we last talked? Can you share some numbers and some milestones, or anything anecdotally you could share from an observation standpoint?

Krishna Rangasayee

>> Sure. So we are really in an exciting phase. We are now six years old. We are very proud that we just recently announced our gen two platform. It's called Modalix, and it's the first platform that really goes from CNNs and mission transformers to supporting generative AI and LLM functions all on a single chip, and for about five months. So that's really, really impressive. What we've been able to do. We're super excited that it's sampling. Gen one has done really well for us, and we've been public about the number of customers we have engaged, think top 10 market leaders in robotics, industrial automation, medical, aerospace and defense, smart vision. We are engaged with all of them globally. What Modalix does is now extends that into a new set of possibilities. So we are bringing in new possibilities to existing customers, but we're also expanding our customer base into new frontiers, if you will. So super exciting for us. And I think I mentioned this to you last time around, we really think of ourselves as a software company building our own silicon. And that software strength really is the key of it. And for hardware systems to scale, software needs to be amazing. And so we are super excited. The combination of what we have done with our software so far, with our gen two platform is now going to take us into a broader set of customers globally.

>> Well, Krishna, congratulations on the momentum. You're in a hot spot. I love chatting with you. I could probably go for an hour, but I know you're busy and we got to get gone with our robotics event. theCUBE and NYSE is an open community of leaders, a trust network that's been forming as a result of just collaborating with individuals, sharing their voices, sharing their knowledge. Of course, long form content. We love that podcast style. And again, bringing that to you here from the NYSE studios at CUBE Plus, NYSE Wired, and John Furrier. Of course, we got the Silicon Valley, Palo Alto office connecting tech and finance. Thanks for watching.