Semidynamics secures strategic investment from SK hynix to push memory-centric AI chips, signaling memory architecture as a key lever for AI inference performance and cost.

The company is building its own open RISC-V based architecture around memory-wall challenges and integrating its Gazzillion memory subsystem to minimize data movement bottlenecks.

Semidynamics highlights workloads that are memory-bound and data-movement dominated, such as agentic reasoning with long contexts and persistent sessions.

SiMa.ai’s SoMs, embedded with Micron memory, are ready for deployment, offering a faster path from prototype to production with optimized hardware-software stacks for real-time AI workloads.

Micron memory integration enables immediate deployment via SiMa.ai SoMs that fit into robotics, automation, and autonomous vehicle platforms, accelerating prototype-to-production timelines.

A 3nm silicon tape-out with TSMC marks a milestone, placing Semidynamics among a select group of European firms at this advanced node.

The 3nm tape-out with TSMC advances Semidynamics’ roadmap toward high-performance AI inference processors and vertically integrated systems.

The collaboration with TSMC on a 3nm tape-out positions Semidynamics as a European leader in next-generation AI silicon and integrated infrastructure.

SiMa.ai emphasizes hardware-software co-optimization and Palette software to maximize performance-per-watt and shorten time-to-market for customers.

Semidynamics is building a full-stack AI infrastructure platform—chips, boards, and rack-level systems—for data-center-scale inference, backed by a Barcelona-based team of over 150 engineers.

The company’s Barcelona headquarters houses a team of more than 150 engineers focused on proprietary silicon architectures and vertically integrated systems for large-scale AI inference.

The funding includes €45 million in non-dilutive European and Spanish grants to support future chip tape-outs and rack-scale data-center deployments, expanding Semidynamics’ partner ecosystem.