At Embedded World in Germany, Texas Instruments (TI) officially launched two new microcontrollers with edge AI capabilities—the MSPM0G5187 and AM13Ex—reinforcing its commitment to embedding AI across its processing portfolio.

The highlight: both MCUs integrate TI's self-developed TinyEngine Neural Processing Unit (NPU) . This hardware accelerator is tailor-made for MCUs, slashing edge processing latency and boosting energy efficiency. Compared to similar MCUs without an accelerator, AI inference latency drops by up to 90x, and energy consumption is reduced by over 120x.

The MSPM0G5187 is priced under $1 (in 1,000-unit quantities), offering a cost-effective entry point for bringing AI to battery-powered, resource-constrained devices.

The AM13Ex MCU is the industry's first single-chip solution to combine a high-performance Arm Cortex-M33, TinyEngine NPU, and a real-time control architecture. This integration cuts BOM costs by 30% and enables simultaneous complex motor control and AI functions. A trigonometric math accelerator boosts computation speed by 10x, targeting precision applications like industrial automation and robotics.

TI's TinyEngine NPU runs machine learning algorithms in parallel with the main CPU. It delivers 2.56 GOPS of computing power, supports various quantization formats, and simplifies the development toolchain—cutting development cycles from weeks to just hours.

ICgoodFind: TI's self-developed NPU and new AI MCUs lower the barrier for edge intelligence, empowering the next wave of efficient embedded devices.