theCUBE + NYSE Wired: Robotics & AI Media Week | Sviat Dulianinov, Bright Machines

Clips
More from theCUBE + NYSE Wired: Robotics & AI Media Week

Sviat Dulianinov

CEO

Bright Machines

play_circle_outline Revolutionizing Manufacturing with Integrated AI Backbone and Robotics: A Bright Machines Perspective

play_circle_outline Leveraging NVIDIA Omniverse for simulation in product design process.

play_circle_outline Maximizing Server Manufacturing: The Crucial Roles of Precision, Consistency, Flexibility, and Software Efficiency

play_circle_outline Series C funding of $126 million led by BlackRock, Nvidia, Microsoft, JBL

Info
Transcript

Sviat Dulianinov, Bright Machines

Sviat Dulianinov

CEO Bright Machines

Sviat Dulianinov, head of strategy at Bright Machines, joins theCUBE at the NYSE Wired Robotics and Artificial Intelligence Media Week to share insights on the evolving landscape of robotics and AI. The event focuses on how companies such as Bright Machines harness technology to drive the robotics economy forward.

In this video, we explore the expertise of Dulianinov, who is instrumental in refining Bright Machines' approach to full stack robotics solutions. During the discussion with John Furrier of theCUBE, Dulianinov highlights the importance of s... Read more

explore Keep Exploring

What does Bright Machines offer and what is their specific focus as a company? add

What are some examples of how digital twins can be utilized in product design and assembly processes? add

What are the advantages of using automated robotics and software in manufacturing processes compared to human labor? add

What was the amount raised in the Series C round and who were the lead investors? add

bolt Powered by CUBE AI

Sviat Dulianinov, Bright Machines

search

>> Welcome back everyone. I'm John Furrier, host of theCUBE. We're live at the NYSE here for our Robotics AI leader series with theCUBE and the NYSE. It's an open collaboration, opensource network of experts coming together, sharing their knowledge. Of course, Silicon Valley and Palo Alto and theCUBE Studio here and NYSE Connects. Technology innovation, Silicon Valley to Wall Street where the money is, and obviously the market is changing very rapidly. It's about Duly Nino's cheap strategy after Bright Machines, a company building full stack hardware, software powering this robotics economy, which is booming as the result of things like better systems, better hardware, better GPU sensors, computer vision, all that coming to the table. Look no further than GTC. A few weeks ago the CUBE had exclusive coverage there where we had Sviat at his event, kind of by accident, very serendipitous. Sviat, Great to see you. We bumped into each other at GTC, you had a customer event there. Great to see you. Thanks for coming on theCUBE, part of our CUBE NYSE Wired network.

Sviat Dulianinov

>> Yeah, really great to be here. Awesome place and looking forward to our discussion.

>> So Brian Baumann, founder of NYSE Wired, he and I were at the Equinix event. Nvidia had an analyst thing, we're like, "Okay, we're going to navigate through here. And oh, they must be back there. We thought it was a room." You were having a customer event, you had all your customers there. GTC is a very important part of your business. I want to get into that, because you guys are doing some pretty cool things around the full stack, I want to get into that. But first, explain what Bright Machines is doing. You guys have a specific focus, talk about the company, take a minute to explain what you guys do, what you guys offer, and the focus.

Sviat Dulianinov

>> Yeah, sure. So Bright Machines is soon to be seven years old. We do what we call full stack, is the robotics lines powered by our software. And we also separate out the data piece, because our platform that we call Brightware, it runs automation, collects all the data, and is powered by what we call smart skills. So there's this AI element behind it that makes it, in a way, autonomous. So flexible assembly for different types from the same product family. And speaking about the history of the company, we started working with different verticals of the electronics market, from drills to coffee machines, to smoke detectors to speakers. Looking back, when you do all of those different products, it's really hard to productize because they're all a little bit different. I joined the company three years ago, and strategy-wise we narrowed down what we call AI backbone. For us it means AI infrastructure and everything that goes to the data center. And it's mostly servers that use GPUs from Nvidia, from that event that you went to, and that we build them with our robotics and software instead of manual labor.

>> Sviat, great call too on that strategy, because what you saw was happening at GTC, physical AI, big storyline there, but also the underlying roadmap he shared, Jensen shared the roadmap. Why? Because people are building long game strategies around this AI factory, which I love the names. Who doesn't love that name? Implies building cool stuff. But when you look at where the money is, it's the digital twins, the simulation goes into everything. That to me is massive. So okay, yes, fire detectors, coffee machines, a little bit more down the road. Maybe you could scale, but right now the market is building the factories, the digital twins. That hangs together as far as I'm concerned. So that's happening. Well, I'm sure you agree. What does that mean for the businesses that you sell to? Are they standing up digital twins? Are they mostly just getting retrofitted? What's the real transformational elements in some of these manufacturers?

Sviat Dulianinov

>> I mean, speaking about digital twins, first of all, when you visited us at our event, that was because we also launched our product that leverages NVIDIA Omniverse for simulation, and you can simulate the whole factory. That's not what we do. We do simulate the product design and that's where we see a lot of potential, actually, speaking about the transformation and how you can improve the cycles. We simulate the design of new electronics in our case server, and then we simulate the process of assembly so they could give feedback to the engineer, how you can design it better. I believe you can do similar ways for factories and other simulation things. But again, speaking about our business, we focus on that element, on the assembly process and the design of the thing.

>> So you're manufacturing servers?

Sviat Dulianinov

>> Correct.

>> That's what your company does?

Sviat Dulianinov

>> We're providing a solution, robotics and software that manufactures servers, yes.

>> For others too? Or other for-

Sviat Dulianinov

>> I mean, we work with the whole ecosystem, so the-

>> So if I wanted to build servers, let's just say I come up with a Cube design.

Sviat Dulianinov

>> Sure.

>> Do I work with you or do you build them for me or do I build them using your technology?

Sviat Dulianinov

>> You buy from us, our technology, to use it on your floor.

>> Got it. Okay. A little nuance, but I want to make sure. So you guys are building servers yourselves or no?

Sviat Dulianinov

>> We run the solution on the floor. We put it on the floor. But for example, the company can own the factory. We don't own the factory, but we're going to come to this company and work with them and...

>> So you enable basically people to put their AI factories together faster.

Sviat Dulianinov

>> Correct. Correct.

>> So that's-

Sviat Dulianinov

>> Faster, much higher quality. Yes.

>> Yeah. Okay. So I get it now. I was a little bit confused there, because server building is the top priority.

Sviat Dulianinov

>> And in the US mostly, or when you outsourced it historically to different Asian markets. Right now everybody is trying to bring it back home, so we can do it in any place in the world and mostly here.

>> Yeah. What's the demand right now? Because I mean, you're starting to see Abilene, Texas, you've got OpenAI, you've got the Silicon Sands in the Phoenix area, TSMC is going to have eight factories there. We've been talking to a lot of other entrepreneurial folks in the robotics areas, that people think that they could build smaller factories, not just for chips, but you start to see a new wave of manufacturing, a new kind of entrepreneurial motion, persona, younger, smarter, modern, thinking about things like hardware, software integration, adaptability, versatility. Buy one robot, multiple uses.

Sviat Dulianinov

>> Correct.

>> Solved some of the things that you pointed out earlier about the changing nature of a smoke detector.

Sviat Dulianinov

>> Right.

>> Okay. I can do smoke detectors today if I wanted to, but I need to switch to something else without massive hit on costs. That's a trend. Do you agree there too?

Sviat Dulianinov

>> 100%. And I think we are here looking at this trend because of several reasons. A, the software became so much more mature than it was before. And B, robotics is getting better. And C, what we're trying also to do, you push the design to be more manufacturing-friendly, automation-friendly, and that pushes all that innovation, as you mentioned, different entrepreneurial minds. And actually, it's really great to see, as you mentioned, to see more young people and more innovative people in this industry, because as you know, manufacturing has been a little bit old school for many years.

>> Yeah, it's been old school because the old model was, build a physical plant, purpose-built it for output they wanted or outcome, a car, a product. Now you have the agility mindset coming in where robotics actually is set up for the first time to be multi-purpose. I mean multi-purpose, not just humanoids, the precision is there, software could be plugged in kind of like the Matrix. An example. Yeah, I want to fly a helicopter one day and then do karate the next. I mean, that's the kind of software kind of injection that's happening with these large-scale or smarter robotic systems.

Sviat Dulianinov

>> Right.

>> Explain that concept. You agree? What's your reaction to that? And explain why that's important.

Sviat Dulianinov

>> Super important. So A, you have different types of robotics, you mentioned humanoids. We don't do humanoids. We do purpose-built applications and robotics per each particular use case, right? On top of it you have also vehicles that could move around the factory. And putting that all together and orchestrating it, that's how you could in the future create potential efficiencies, right? But to your point, you need software that runs multiple things together, because historically you had software that runs each particular product or computer, so on and so forth. What we're also trying to do, like we are doing, we have this platform that runs, conveying several robots, and it all should work cohesively together. And that's how you bring it to the floor and that's how you create results.

>> Well, it's great to see what you guys are doing. Great to run into it here at GTC. Give us an update on the business. What's some of the traction, momentum? Can you just give us... Tell some stories. Stories drive movements. And we're in a robotics software, hardware movement. Tell us a couple of stories.

Sviat Dulianinov

>> Sure. I think one of the interesting things right now is, and people maybe don't know this, but it's not only about assembly, but it's also about what we call circular economy. It's a lot of disassembly, recycling, and reuse happening too, because if you think about the trends, everybody is trying to build data centers. A, you cannot build greenfield new data centers forever. So they also take out older data centers and take out kind of new servers. They're two, three years old. It's a waste if you just throw it away. But they want to disassemble it, reuse, recycle, resell. So that's a big market movement that we also see. And a lot of companies are coming to us in the last, I think half a year and ask us to build solutions for them to disassemble, for example, sort, label, test different elements that they take out. That's an interest angle to the market. And also, I think the most recent announcement of our Bright designer, we call it Design for Automated Assembly, leveraging digital twins and Omniverse. We can see more of this happening with other companies as well, moving to Omniverse, moving to simulation. And as you mentioned, I think it's still early stage, but a lot of companies are really excited to see the potential results when they come up.

>> Yeah, well, that's awesome and I think that's a great observation. I just love the fact that we're here at the NYSE. The folks can't see on camera, but behind us is the option trading floor. Market has been big swings today and it's very loud in there. Active trading floor, the top story here is with the market fluctuations, the puts and the calls are all under water. So what's happening is that it's a feeding frenzy here in New York City, again showing the volatility of the market. We are in a revolution. We are in a significant changeover and I think the market is just... Outside the political issues there are still geopolitical supply chain issues going on. We have a culture shift. I've been saying this for years on theCUBE. Everyone knows my position. It's a cultural generational takeover revolution. The new generation will take over business, all business. And AI is going to accelerate that. Blockchain, robotics, these new forcing functions are a legitimate value that cannot be ignored. Enough value to lever out the existing people who don't adapt. This is a huge dynamic. Share your thoughts on this, because I'm trying to get some consensus. Am I weird in thinking this way? It's almost a complete revolution.

Sviat Dulianinov

>> I completely agree. I think what's happening right now with AI and the speed of, we call it revolution, it could be evolution, whatever you choose.

>> I like revolution better.

Sviat Dulianinov

>> Great. Sure.

>> Evolution for sure, but not this fast.

Sviat Dulianinov

>> Right. It's going to be super fast and I think it's bigger than most of the things we saw before. I really believe that those companies are people that are smart enough to adapt and actually use those instruments like AI, robotics, flexibility that you mentioned. For different industries, they're going to win a lot. And as you mentioned otherwise, those who don't do, they're going to lose, right?

>> I was talking with Chris Stevens yesterday, that is field CTO at Grok. He's new to the company, but he's been around, we talked about our journey. He also teaches at Carnegie Mellon Pedigree School. Obviously great school, it's Northeastern. I went to Northeastern, so of course I can say that. Maybe have a good computer science program. But Chris Stevens at Grok was talking about, and we are riffing on the moment we're in now, and if you think of AI and robotics and blockchains, some of these advances that are generational shifts, it's like electricity, but we're in the 1800s. It's early days still. So the benefits of what's coming is still not yet realized. And that's why I think this generational shift, it's a multi-decade changeover.

Sviat Dulianinov

>> 100%.

>> And I think evolution, for sure, but evolutions don't happen this fast. So the question is, who takes over? What big brands come in next? So in all these market transitions, winners emerge. Some people drop to the side, people kind of will adapt. Never fight fashion, as we say.

Sviat Dulianinov

>> Yeah, as it happened before, right? So I think it's going to be similar. But also to your point, it's going to be decades process. And also, you can also see that, because not every company just jumps right into it and invests tons of CapEx into, I don't know, a data center or new AI tools. Some companies move step by step because they're trying to mitigate their risks. And again, I don't know what's the right strategy, we'll see in five years, but I think that will determine who's going to win and who's going to come to the side, right?

>> So two years ago, Dave Vellante and I started doing a research kind of view around what we called "clustered systems". Not the classic SMP-based system, not the old definition, but clustered servers. Because what was happening with Nvidia, we saw that clearly last year at GTC, was the iFactory. Okay, we saw that. We saw what was going up prior to that with Broadcom, other chip manufacturers. Pretty obvious a new system architecture around a data center. And the data center kind of gives people the comfort that the old weight still exists, but it's just electricity and power and cooling. It's a room where servers used to live. Now it's server that's a collection of other servers and components, subsystems, a whole nother system level, like a mainframe. I say mainframe, because mainframes are big iron systems. So Nvidia's factory is essentially a large mainframe as far as I'm concerned, in the sense of big iron capabilities. Not definitionally, but it takes up a lot of space. You've got to engineer the connect interconnects, you've got to engineer the fabrics, storage change its role. So as you guys look at some of these build-outs, a lot of enterprise, high-end enterprises are starting to adopt and starting to think about, "What does my footprint look like in my data center?" Not just a rack and stack of hubs and switches and servers. It's like, how do they all work together as one thing?

Sviat Dulianinov

>> Right.

>> So I imagine that as you guys design all this, if no one server is going to look the same, maybe they look kind of the same or not. Tell me what you think.

Sviat Dulianinov

>> So a really interesting angle, because we don't design, we help design, like other companies design servers. But to your point, what we saw in the last several years, the design change or the amount of new servers per, say month or year has increased dramatically. So you can see companies having more than 15, 20 different servers per year, which if you look three, four years back, it was one or two, right? You didn't need that flexibility before that. But now no other server looks the same, as you said. Of course, you're going to have packs with them, but then you have 20 different variabilities, which really interesting, because that you need to meet the demand, you have different needs and that's why you change the design, right? And then you have different components and different supply chains.

>> And that's why you guys are operating on the product, because that system is what needs to be kind of spec'd out. I mean it's almost like spill to order systems. I mean, I'm oversimplifying.

Sviat Dulianinov

>> Yeah.

>> I mean, I'm simplifying. It's not that easy, but scope, the complexity, what you guys do, what's the alternative?

Sviat Dulianinov

>> The alternative historically is just human labor and you just outsource it to China mostly. And you have the same floor that we see here and tons of people just step-by-step building things with their hands. There's several problems with that. First, the precision at which we operate is around 75 microns, which is really precise and you need to be precise when you put electronics components together because it's going to be a longer life, higher quality, it's going to work better. People are not that consistent and not that precise. That's one, right? Two, it's really hard for standard automation to process different products on the same line. I'm speaking about the same product family. So for example, if we built a line and we build servers, we can build 15 different SKUs of those servers on this line. People can do that. Robotics in most cases, automation historically could not do that. And the secret sauce here is just software. That's how you add this flexibility. I think the other thing is, when you try to build a new product, you're going to fly your engineers around the world. You're going to fly them to China, to those factories, you're going to train people, spend weeks. The way we can introduce new products to the factory floor on our line is basically within four hours, we call it NPI process, new product introduction, we create a 3D model, create what we call a recipe flow assembly, and then you push it through the cloud and it could start assembling. Sounds easy, really hard.

>> But I'm sure once you get that going then, then it's smoother. So the investment on the front end, super valuable.

Sviat Dulianinov

>> Right.

>> All right. What's next for you guys? What's on your agenda as a company? What are you guys looking to do? How's the funding look? How's the customer traction? Give a couple of highlights.

Sviat Dulianinov

>> Last year we closed our Series C around 126 million led by BlackRock. Nvidia that you mentioned, Microsoft, they were strategic investors. JBL was the other strategic investor. So we feel pretty good about the financial part of business for the next year and a half or so. It's more about focusing on making our customers happy because they have a lot of needs right now. As I mentioned, there is a lot of needs for assembling servers, but also for disassembling servers. So we can see the companies we work with, big names, we can see both angles and they ask for both solutions right now. And also, they ask for this design thing. We announced it two weeks ago only at GDC, the design for automation, and we already have two customers signed. That's pretty incredible. For a startup we have two customers who then just do it.

>> Can you tell about the pricing? What do they normally sign up for, total contract value?

Sviat Dulianinov

>> For this one they usually sign up for, we call proof of concept. And they invest, let's say half a million dollars or so into this first stage and then we'll determine... We want to have it on Azure marketplace end of year, so it can go through the marketplace. If you want to buy a solution like a line, it's also going to depend on the line and complexity. So some lines that are complex, they could be several million dollars, more than five. And some lines that are simpler, fewer steps, that could be cheaper obviously.

>> Got it. It depends on what they're building.

Sviat Dulianinov

>> Of course, and how many steps you have, how complex the product is, what you're trying to automate. We built a GPU integration solution, we call it Zero Scrap Operations, which means technically 100% yield. That's pretty impressive.

>> That's awesome. Yeah, I mean, look at just the benefits. I mean, it's clear that the benefits are there, you start to see. It's really not a lot of objects if you can actually prove it and get it into production. So Sviat, it's great to have you on theCUBE. Thanks for coming in.

Sviat Dulianinov

>> Thank you.

>> Good to see you in person again.

Sviat Dulianinov

>> Yes.

>> Again, GTC, where all the action was, you guys doing well there. Of course, theCUBE is bringing all the action here from the NYSE. As the markets are in transition, new leaders will emerge. Robotics, AI coming together as the new waves go faster, more availability components, and of course the applications that are going to sit on the top of it are all going to be a big part of this fast revolution in systems. I'm John Furrier, your host at theCUBE. Thanks for watching.