The $10 Trillion AI Revolution: Sviat Dulianinov on Why Robots Will Reshape Manufacturing

Summary

Podcast Summary: "The $10 Trillion AI Revolution: Sviat Dulianinov on Why Robots Will Reshape Manufacturing"

Podcast Information:

• Title: Liftoff with Keith Newman
• Host: Keith Newman, former journalist and Silicon Valley dealmaker
• Guest: Sviat Dulianinov, Chief Strategy Officer at Bright Machines
• Release Date: June 18, 2025
• Episode Focus: Exploring the transformative impact of AI and robotics on manufacturing and the broader economic landscape.

---

Introduction

In this episode of "Liftoff with Keith Newman," host Keith Newman welcomes Sviat Dulianinov, the Chief Strategy Officer at Bright Machines, to discuss the evolving dynamics of AI and robotics within the manufacturing sector. The conversation delves into how advancements in AI infrastructure and robotics are poised to revolutionize economic models and reshape industries globally.

---

AI’s Transition to the Physical Space

Sviat Dulianinov begins by contrasting the current focus on Large Language Models (LLMs) with the burgeoning integration of AI into the physical realm through robotics.

“AI Infra is an important topic, of course, and we've seen AI changing the world in the last few years. But mostly you think about LLMs... what really changes right now and economics wise, it's about AI enters the physical space and then you do that through robotics and we are on the verge of revolution.”
[00:24]

Dulianinov emphasizes that while LLMs handle tasks like writing and coding, the true economic transformation lies in deploying AI-driven robots to build infrastructure, machinery, and electronics, enhancing human capabilities and productivity.

---

Current State and Future of Robotics

When questioned about the current trajectory of robotics, Dulianinov provides a nuanced perspective on both achievements and future prospects.

“From the models perspective it's really, really early days. Like everybody speaks about foundational models, it's going to take time to get to something really impressive... current robots... pretty well built... use case based applications.”
[01:23]

He highlights that while specialized robots for tasks such as pick-and-place operations are well-developed, more complex forms like humanoid robots remain in nascent stages. The differentiation underscores the gradual progression towards more adaptable and intelligent robotic systems.

---

Infrastructure Considerations for Robotics

Addressing the infrastructural prerequisites for widespread robotics deployment, Dulianinov identifies data acquisition and computational power as critical challenges.

“You need like first of all you need a lot of data. That's the challenge for robotics because... it's simulation, which is not real world, or you do a lot of repetition in real world.”
[02:01]

He draws parallels with autonomous driving companies like Waymo, which rely on extensive real-world data to train their models. Additionally, the surge in AI-focused data centers provides the necessary computational backbone to support both LLMs and robotics, positioning Bright Machines to facilitate faster deployment rates.

---

Balancing Energy Needs with Urban Integration

The conversation shifts to the logistical challenges of integrating massive AI and robotics infrastructures within urban settings, particularly concerning energy consumption and cooling requirements.

“Not every state or city is ready for those massive data centers because of energy and because of other reasons... you need to be smart and also think about cooling.”
[03:10]

Dulianinov underscores the importance of strategic placement of data centers to mitigate issues like energy shortages and environmental risks (e.g., wildfires in California). He suggests optimizing infrastructure by co-locating servers and GPUs within data centers to enhance efficiency.

---

Market Demand Insights

When discussing market demand, Dulianinov notes a significant uptick in the United States driven by reshoring initiatives and supply chain optimization efforts prompted by the COVID-19 pandemic.

“We see most of the demand in the US because of the push on reshoring... supply chain closer to home.”
[04:25]

This trend reflects a broader move towards strengthening domestic manufacturing capabilities, reducing dependency on global supply chains, and navigating tariff landscapes, thereby increasing the demand for advanced robotics and AI solutions.

---

Concerns and Opportunities in the Near Term

Concerns: Dulianinov expresses apprehension about market uncertainty potentially stalling innovation. He warns that:

“Companies become really careful with their choices and that could kind of delay innovation.”
[05:03]

The hesitancy to invest in new technologies during uncertain times could allow competitors, particularly in countries like China, to accelerate deployment and gain a competitive edge.

Opportunities: Conversely, he highlights a monumental opportunity in aligning the manufacturing sector with AI and robotics to resurrect domestic production capabilities.

“The biggest opportunity is to align the sector... to leverage robotics, AI... $10 trillion opportunity to bring manufacturing back in the US.”
[05:30]

Dulianinov envisions a future where automation compensates for labor shortages, unlocking vast economic potential over the next decade, provided there is cohesive effort from industry leaders and government bodies.

---

Vision for 2030

Looking ahead to 2030, Dulianinov anticipates significant advancements in simulation technologies and broader application of robotics across various industries.

“You're going to see much more use... simulation can help a lot... more use of robotics as part of the manufacturing processes and maybe construction processes.”
[06:36]

He predicts that simulations will mature, enabling more efficient design and testing processes, while robotics will become integral to autonomous construction, manufacturing, cleaning, and even medical services. The precision and reliability achieved in fields like electronics manufacturing will pave the way for robotics to handle increasingly complex tasks across diverse sectors.

---

Conclusion

Keith Newman wraps up the insightful discussion by commending Dulianinov's perspectives and encouraging listeners to explore his keynote presentations for deeper understanding.

“We've got to run, but I encourage everybody to check out your talk that you gave on stage here. It was fantastic. Appreciate you stopping by.”
[07:50]

Sviat Dulianinov reciprocates the gratitude, marking the end of an enlightening conversation on the intersection of AI, robotics, and the future of manufacturing.

---

Key Takeaways:

• AI's Physical Integration: The true economic impact of AI lies in its application within the physical realm via robotics.
• Current Robotics Landscape: Specialized robots are advancing rapidly, but broader, more adaptable robotic systems are still evolving.
• Infrastructure Needs: Adequate data and computational resources are essential for the continued growth and deployment of robotics.
• Energy and Urban Planning: Strategic placement and efficient resource management are critical for integrating AI infrastructures into cities.
• Market Dynamics: The U.S. is witnessing heightened demand for robotics due to reshoring and supply chain resilience efforts.
• Future Outlook: By 2030, simulation technologies and robotics will be pivotal in various industries, driving significant economic growth.

---

For more insights and in-depth discussions, visit Liftoff with Keith Newman and explore over 80 episodes featuring industry leaders and innovators.

Transcript

[A]

Foreign.

[B]

I get to welcome Sviat, the chief Strategy Officer at Bright Machines here at Infra AI. Welcome Sviat.

[A]

Thank you Keith. Great to be here.

[B]

Yeah, it's wonderful to have you. And we get to talk about the changing economics that AI and robotics have foisted upon us. Give us your thoughts from a Bright Machines perspective.

[A]

Yeah, it's a fascinating topic by the way, because I mean, first of all, AI Infra is an important topic, of course, and we've seen AI changing the world in the last few years. But mostly you think about LLMs, right? And you think about how we can write poems, can write an email, you can maybe code a mini game, maybe it can create a picture like an image of art. But what really changes right now and economics wise, it's about AI enters the physical space and then you do that through robotics and we are on the verge of revolution. Because if you excel in it, robots can be so capable and they're going to help us, human beings to be more capable in other things. Right. So we can use robots to build buildings for us, to build machines for us, to build electronics for us. And that's what we are excited at, Bright Machines. And we're excited because we're one of those leaders who push for that because we use robotics and AI to build most complex electronics. That's fascinating.

[B]

Where are we at today in the curve for robotics and its usage?

[A]

I think from the models perspective it's really, really early days. Like everybody speaks about foundational models, it's going to take time to get to something really impressive. But from a robotics perspective, I think current robots are, I mean they're really well built. I'm not talking about humanoids, it's more about use case based applications. Like you use a robotic hand and use a special end of arm for pick and place or for screwdriver. I think that is pretty developed right now. And then there is more to go for other types of robotics like humanoid robots. So it's going to take also time.

[B]

So what do we need to see happen from an infrastructure standpoint then to help accelerate that deployment of robotics?

[A]

AI usage, I think that's what everybody's pushing for. You need like first of all you need a lot of data. That's the challenge for robotics because like, you know, when foundational models like you know, ChatGPT and so they learn based on the Internet, right? It's like vast, massive data. For robotics you don't have that much data. So you either do simulation, which is not real world, or you Do a lot of repetition in real world. So you can see that in San Francisco, like Waymo initially was teaching like learning, driving the streets and then training the models. Right? That would be a bottleneck and a challenge. It's not about infra, it's more about like time and data needed to train it. And then on the other hand, everybody's building huge data centers right now for AI infrastructure that's going to help both LLMs, but also going to help the foundation models for robotics because you need the compute power. And again we are helping them build it. So hopefully we're going to be one of those guys who are going to help improve the speed of deployment of robotics and AI.

[B]

I see that happening very quickly now. Talk to me about balancing out the need for mass amounts of energy and power with the need for integrated into a city's planning strategy and like having real world implementations.

[A]

That's a really good point, I think. I mean first of all, let's be clear, like not every state or city is ready for those massive data centers because of energy and because of other reasons, right? So you need to be smart and also think about cooling. It's not only about energy that you need to test those servers based on GPUs, but also you need to a lot of energy sources. Huge issue, right? So even when you think about like oh, let's build a cooling thing in Texas, I guess, like sounds great, but like also in summer it's going to be double challenging, you know. So you know it is about they.

[B]

Have their own grid issues as we speak, right?

[A]

Or like think about California and like British is already like fires and stuff. So you need to be smart of where you want to build it. And to your point, I think you need to be smart if you could combine where you build the infrastructure and where do you deploy it. Because when you build it, you still need that same energy to test those servers and GPUs and you kind of burn them and then you take those racks and GPUs and move them to the data center. So the question is like do you need it to be in several locations or can you optimize and actually have like one location maybe co locate building servers and GPUs and actually placing them into the data center to run them.

[B]

So with bright machines you serve a global market. Where are you seeing the most demand?

[A]

We served the global market historically and we are open to work with any country. But right now we see most of the demand in the US because of the push on reshoring and you know, it has been happening since I think Covid years when the whole world realized that supply chain could be broken easily. And so everybody started thinking how we can optimize to have supply chain closer to home. And right now it's the push and tariffs and all that stuff. So we see a lot of interest and demand in the US market.

[B]

Yeah, well, with the head of strategy, I got to ask, what are your biggest concerns and what are the biggest opportunities looking 0 to 12, 24 months out?

[A]

Yeah, I think the concern is when the market isn't at uncertainty and we can recognize that it is kind of uncertain because many changes are happening every week. Companies become really careful with their choices and that could kind of delay innovation. So it is kind of a concern because some big companies can say, look, we can wait another quarter to see what's going to happen. But for innovation, one quarter can mean a lot because you can push harder and just innovate faster and deploy new technology versus just waiting another quarter or two because other countries, I don't know, looking China, they not going to wait, they're just going to deploy it. So that's kind of a concern. I think as an opportunity, as we discussed, the biggest opportunity is to align the sector I'm speaking about right now, manufacturing, to align the sector that you need to leverage robotics, AI to make it happen and to bring manufacturing back. And you know, when you align everybody, the opportunity is enormous. I was just on the stage talking about like $10 trillion opportunity to bring manufacturing back in the US to the US and again, you know that you don't have enough labor here, so you cannot do it manually. So you can realize on this huge potential maybe not two years, maybe 10 years. You asked about 24 months, maybe 10 years. But for that you need alignment of the industry, leaders of the government. That's the leverage you're going to push. Like robotics, AI software and leverage and data that you create to make it more efficient.

[B]

Okay, so now we're going to go out to 2030. What do I see radically different as I cover the landscape?

[A]

You're going to see much more use, I think of simulation. It's going to become much more mature. I think right now there's also not early days, but day three maybe out of 10. But it's getting better. And I think the simulation can help a lot with simulating physics and how you design tools before you actually bring them to real life and then going to cut the cycles and it can do a lot of repetitions in the simulation when you design a product, for example. So you're going to see a lot more use of this, that you're going to see a lot of more use of robotics as part of the manufacturing processes and maybe construction processes and other things, because it's getting more and more mature also in physical space. I think you're going to see more examples of like what we see in San Francisco with Waymo. It's a little bit further in the curve versus everything else, maybe because of precision, because again, driving a car, when you go 1 inch left or 1 inch right, not a huge deal when you park. But when you build electronics like we do, and we operate like 75 micron level precision, you go left 25 microns, you break the latch, so you cannot go this way. But you're going to see much more of Waymo in other industries like autonomous construction, autonomous manufacturing, autonomous cleaning, autonomous medical services, and so on and so forth.

[B]

Yeah, that's fantastic. We've got to run, but I encourage everybody to check out your talk that you gave on stage here. It was fantastic. Appreciate you stopping by.

[A]

Awesome. Thank you, thank you, thank you.