Sviat Dulianinov, CEO of Bright Machines, joins Nataraj Sindam to explain how a software-first, robotics-first approach is bringing AI data center hardware manufacturing to the United States. Founded in 2018 and backed by BlackRock, NVIDIA's NVentures, Microsoft, and Jabil, Bright Machines builds the compute nodes and racks that power modern data centers using AI-driven robots instead of the massive manual factories of Asia. Sviat breaks down the economics of building $250,000 server units, why quality and time-to-market matter more than labor cost, the role of "physical AI" on the factory floor, and the real case for reshoring strategic electronics.
5 Things You'll Learn from This Episode
- Why Bright Machines bet on AI data center infrastructure before the ChatGPT moment — and how the economics of $10K–$250K server units change what actually matters in manufacturing.
- How a software-first, robotics-first factory replaces the skill and scale of Asian manufacturing hubs with smart software and easily replicable robots.
- Why time-to-market, first-pass yield, and throughput — not labor cost — become the metrics that drive value when each unit is as expensive as a car.
- How Bright Machines' platform, AI "smart skills," data layer, and Bright Designer close the loop between product design and automated assembly.
- Why humanoids aren't ready for the factory floor, and how purpose-built robotics deliver 45–55 micron precision and up to 2x the throughput of manual lines today.
- The real case for onshoring AI hardware — security, speed, and building new capacity — beyond the simplistic "cost of labor" argument.
About the Episode
Sviat Dulianinov, CEO of Bright Machines, joins Nataraj to explain how a software-defined, robotics-first approach is bringing AI hardware manufacturing to the United States. Founded in 2018 and backed by BlackRock, NVIDIA's NVentures, Microsoft, and Jabil, Bright Machines builds the compute nodes and racks that power modern data centers — using AI-driven robots instead of the massive manual factories of Asia. Sviat breaks down the economics of building $250,000 server units, why quality and time-to-market matter more than labor cost, how the assembly line blends robots with humans in the loop, and the real case for reshoring strategic electronics.
Timestamps
- 0:00 — Introduction: Bright Machines and AI hardware manufacturing
- 1:01 — What Bright Machines does: software-first, robotics-first factories
- 2:25 — The original thesis: manufacturing electronics anywhere
- 4:40 — Where server racks are manufactured today: Taiwan, Mexico, the US
- 5:58 — The market stack: chip designers, ODMs, OEMs, hyperscalers
- 8:46 — Contract manufacturers: Foxconn, Jabil, Flex
- 10:36 — Offshoring, labor arbitrage, and two decades of low inflation
- 12:20 — Betting on data centers before ChatGPT
- 16:16 — The first product: CPU compute nodes for a hyperscaler
- 17:45 — How Bright Machines builds a server vs. the traditional line
- 21:50 — The advantage: time-to-market, first-pass yield, throughput
- 24:18 — Inside the assembly line: robots vs. humans in the loop
- 30:02 — Business model: manufacturing as a service, per-unit pricing
- 31:09 — Scale: 3x growth, half a gigawatt, 40–50 units per hour
- 32:34 — Will a CapEx slowdown hurt Bright Machines?
- 34:30 — From CapEx commitment to deployment: the bottlenecks
- 36:19 — The core IP: platform, smart skills, data, Bright Designer
- 38:39 — Physical AI, humanoids, and purpose-built robotics
- 41:16 — How LLMs are changing robotics
- 43:38 — The real advantages of onshoring beyond labor cost
- 45:50 — What Bright Machines will build five years from now
Key Insights
Q: What does Bright Machines do?
Bright Machines builds AI infrastructure hardware — the compute nodes, racks, and data center electronics that power modern AI — using a software-first, robotics-first manufacturing approach. Founded in 2018, it replaces the massive manual factories of Asia with AI-driven robots, enabling electronics to be assembled locally, close to where they'll be deployed. The company is now 100% focused on AI data center infrastructure.
Q: Why is manufacturing moving back to the US for AI hardware?
For expensive, strategic electronics like AI servers, labor cost is no longer the deciding factor. When a single compute node can cost $50,000–$250,000, what matters is quality, first-pass yield, throughput, and time-to-market — plus the security of building critical infrastructure locally. Robotics makes US-based production viable despite higher labor costs, and there aren't enough workers to build the needed capacity manually anyway.
Q: How does robotics-based manufacturing differ from a traditional assembly line?
Traditional electronics assembly relies on large factories with tens of thousands of manual workers moving product down a conveyor. Bright Machines replaces most of that with robotic cells guided by AI "smart skills" for navigation, inspection, and force control, achieving 45–55 micron precision and up to 2x the throughput of manual lines. Around 20% of steps still use humans in the loop, and the system captures data at every station for traceability and continuous improvement.
Q: What is Bright Machines' business model?
Bright Machines operates as "manufacturing as a service." Rather than selling robots or software, it charges customers per unit built — whether that unit is a compute node or a full rack. Its core IP is the software platform that orchestrates the line, the AI smart skills that run the robots, the data layer, and Bright Designer, which feeds manufacturing insights back into product design.
About Sviat Dulianinov
Sviat Dulianinov is the CEO of Bright Machines, a company reimagining how complex electronics are built by bringing software, AI, robotics, and computer vision together on the factory floor. Since 2018, Bright Machines has raised more than $600 million from investors including BlackRock, NVIDIA's NVentures, Microsoft, and Jabil, and deployed in more than 100 microfactories across 13 countries. He is a recognized voice on the future of AI-driven manufacturing and "physical AI."
CEO at Bright Machines
About the Host
Nataraj Sindam is the creator of The Startup Project, a podcast featuring founders, investors, and operators building the future.
#StartupProject #BrightMachines #SviatDulianinov #PhysicalAI #AIInfrastructure #Manufacturing #Robotics #DataCenters #Reshoring #Automation #Entrepreneurship #Podcast #Tech