SEMIFIVE has signed an 18 billion-won (≈ USD 12.5 million) contract with Mobilint to develop a robotics-focused AI chip under the K-On-Device AI Semiconductor program.
Contract scope and technical focus
The agreement covers end-to-end development. SEMIFIVE will define the architecture, lay out the silicon, manage wafer fabrication, and handle packaging and testing. Mobilint will supply system-level specifications and integration support. The chip will combine next-generation memory with high-speed interfaces. Planned features include LPDDR6 for on-chip memory, PCIe Gen6 for host connectivity, and UCIe-S for scalable interconnect.
Relevance for robotics AI
Robotics systems now need more on-device inference capability. Most AI chips are built for data-center, high-performance computing or edge tasks. SEMIFIVE’s new chip aims to fill the gap by delivering “Physical AI” performance inside robotic platforms. Integrated memory and fast interfaces should cut data-movement overhead, a common bottleneck in real-time control loops. If the targets are met, robot makers can run complex perception and planning models without external compute resources.
Market implications
The contract adds to SEMIFIVE’s portfolio of AI ASIC projects. Recent announcements show the company securing multi-hundred-million-won deals with both domestic and U.S. fabless firms. This suggests SEMIFIVE is strengthening its role as a provider of custom silicon. By moving into robotics, the firm broadens its addressable market beyond the crowded data-center and edge segments. The approach may also attract robotics OEMs that prefer a single supplier for design, fabrication and testing.
Outlook and next steps
The development timeline aims for a first silicon tape-out within 12 months. After silicon validation, SEMIFIVE plans to offer volume production through its existing foundry network. Mobilint is expected to run pilot deployments on partner robot platforms to gauge real-world performance. Success could lead other robotics firms to seek similar solutions, increasing demand for AI chips that blend high-speed interfaces with advanced memory.
For similar updates on semiconductor activities, visit our Press Release Archives.