The solution combines an SPSRAM memory compiler and standard-cell library for AIoT, wearables and always-on edge devices.
M31 Technology (M31) and Ambiq announced a joint ultra-low-power Foundation IP solution on the TSMC N12e process at TSMC 2026 North America Open Innovation Platform (OIP) Ecosystem Forum. The solution brings together an optimized SPSRAM memory compiler and an ultra-low-leakage standard-cell library for Ambiq’s platforms, to help SoC designers meet increasingly stringent power and area budgets for AIoT, wearables, and other always-on edge applications.
As SoC designs incorporate more always-on capability and embedded intelligence, power targets continue to tighten while die-size constraints and tape-out schedules remain unforgiving. Built on TSMC N12e, this solution focuses on design-ready Foundation IP that improves energy efficiency while delivering the high performance required for complex AIoT and edge computing workloads.
The Foundation IP includes a TSMC N12e ULL (Ultra Low Leakage) SPSRAM memory compiler designed to deliver high density and high performance with extremely low leakage and optimized area, while incorporating essential low-power features such as redundancy, power gating, and BIST. The compiler supports both high-performance and ultra-low-power operation within the same physical memory footprint, enabled by deep co-optimization across circuit design and layout implementation. In low-power mode, its dual-rail architecture achieves close to 500 MHz at near-threshold periphery voltage, while keeping VCCA retention leakage within 1.05-1.1x of the total bit-cell retention leakage. Initial instances have demonstrated strong PPA gains, and full compiler development is ongoing.
Complementing the memory compiler is the TSMC N12e standard-cell library from M31, developed to support Ambiq’s platforms and achieving approximately 50% leakage power savings through design-level collaboration with Ambiq, combining Ambiq’s cascade-style approach with M31’s circuit and layout innovations.
Designed to empower next-generation always-on sensing and edge intelligence applications, these Foundation IP deliverables support faster time-to-market and shorter design cycles, maximize energy efficiency through ultra-low-leakage design, and improve cost efficiency by helping reduce die size and minimizing redesign effort.
Looking ahead, M31 will continue expanding its low-power IP portfolio on advanced process nodes while working closely with ecosystem partners to advance high-performance, low-leakage solutions for next-generation Edge AI applications.
For more information, visit m31tech.com.
