• Scaling plan targets higher density, faster switching, and lower energy
• 40 nm RRAM program tracking to expectations
• 22 nm node evaluation underway, with projected gains in density, energy, and speed

dorsaVi Ltd (ASX: DVL) is pleased to announce the commencement of a formal evaluation program to scale its oxide-based RRAM from 40 nm to an advanced 22 nm process node. The initiative builds on strong device- and wafer-level results at 40 nm and is aimed at delivering higher density, lower energy, and faster switching consistent with the Company's roadmap for embedded non-volatile memory, wearables and AI driven systems. The Company's near-term applications emphasise real-time biosignal processing (EMG, ECG) and always-on motion intelligence for wearables and industrial safety. By evaluating a 22 nm implementation of the RRAM platform, dorsaVi is targeting up to ~3x higher bit density, ~40-60% lower energy per bit, and sub-100 ns switching. These scaling gains are intended to relieve NAND bottlenecks, extend device battery life, and enable tighter closed-loop responsiveness in edge workloads. Over the longer term, the same improvements strengthen compute-near-memory and robotics reflex functions being advanced within Artemis Labs, providing a clearer integration path from today's 40 nm results to next-generation embedded AI systems at 22 nm. Subject to final device- and array-level validation and commercial terms, dorsaVi intends to proceed to a 22 nm test-chip tape-out on TSMC's 22 nm platform, marking the transition from evaluation to fabrication.

Projected gains at 22 nm include ~3.3x density increase, ~50% energy reduction per bit, sub-100 ns switching, and ~1.5 V write operation.

Advancing to a 22 nm RRAM implementation positions dorsaVi's motion-sensor technology to deliver better outcomes at a smaller scale, enabling more responsive analytics, longer wear time in lighter, slimmer devices, and tighter control loops with lower power draw for robotics applications.