Semidynamics Secures SK hynix Investment to Advance Memory-Centric AI Inference Chips Following 3nm Tape-Out

Semidynamics, a company developing memory-centric AI infrastructure for large-scale inference, announced a strategic investment from SK hynix, a memory manufacturer. The investment is based on the shared view that memory architecture determines the economics of next-generation AI inference, measured by cost per token.

Large language models are scaling, and agentic multi-turn workloads require persistent context across longer inference sessions. System performance is constrained by memory capacity and data movement rather than raw compute. Semidynamics states that its architecture delivers multiples of the memory capacity available in conventional HBM-based inference systems, supporting larger models, larger KV-caches, and larger contexts. This configuration enables more users per rack, which leads to lower cost per token.

The company is headquartered in Barcelona and has designed its proprietary implementation of the open RISC-V architecture from first principles around the memory wall. The architecture includes Semidynamics’ Gazzillion® memory subsystem technology, supported by a growing patent portfolio. Gazzillion® is the company’s latency-tolerance technology, implemented from the core and tensor unit through to the memory subsystem, to maintain productivity during long memory access times.

Semidynamics completed a 3nm silicon tap...