Texas Instruments Integrates TinyEngine NPUs into MCUs, Boosts Edge AI Efficiency 120x
Texas Instruments launched MSPM0G5187 and AM13Ex MCUs with integrated TinyEngine NPU, achieving up to 90x lower latency and 120x reduced energy per inference at under US$1 in 1,000-unit quantities. TI also enhanced its TDA5 SoC with a Green Hills Software production-focused SDK on a virtual development platform.
1. MCU Portfolio Expansion
Texas Instruments introduced two new microcontroller families, MSPM0G5187 and AM13Ex, each integrating the TinyEngine neural processing unit to enable on-chip AI acceleration across both general-purpose and real-time applications.
2. Performance and Pricing Metrics
The TinyEngine NPU delivers up to 90 times lower latency and 120 times lower energy consumption per inference compared to non-accelerated MCUs, with the MSPM0G5187 priced under US$1 in 1,000-unit quantities to address cost-sensitive edge AI markets.
3. Production-Focused SDK for TDA5 SoC
TI collaborated with Green Hills Software to offer a comprehensive production-focused software development kit on the TDA5 SoC virtual development platform, ensuring optimized integration, embedded safety features and security compliance for high-volume deployments.