Samsung Electronics has developed a chiplet platform with Cadence Design Systems aimed at physical AI semiconductors for robotics and autonomous systems, ETNews reported. The foundry initiative targets chips that process real-world sensor data and control actuators in robots, vehicles, drones, and factory equipment.
Context
Chiplet architectures allow manufacturers to combine specialized processing units—sensor interfaces, AI accelerators, motor controllers—on a single package rather than monolithic dies. This reduces development costs and allows customization for different robotic applications without redesigning entire chips. Samsung's foundry business has struggled against TSMC in recent years, capturing roughly 13% of the global foundry market compared to TSMC's 60%.
Industry Impact
Physical AI chips differ from data center AI accelerators by prioritizing low-latency sensor fusion and real-time control over raw compute throughput. Robotics companies currently rely on combinations of general-purpose processors, FPGAs, and specialized ASICs—a fragmented approach that adds cost and complexity. A standardized chiplet platform could accelerate time-to-market for new robotic systems if Samsung can offer competitive pricing and yield rates.
The move puts Samsung's foundry in direct competition with Intel's efforts in edge AI and TSMC's automotive chip partnerships. Cadence's involvement suggests the platform will support heterogeneous integration, mixing different process nodes and IP blocks. Samsung has not disclosed timeline or lead customers.
