TI brings edge AI microcontrollers deeper into embedded design
Texas Instruments has expanded its embedded portfolio with two new MCU families aimed at pushing edge AI microcontrollers into lower-cost and more control-intensive designs, combining dedicated neural acceleration with a broader software push around model deployment and AI-assisted development.
TI expands edge AI microcontrollers across two MCU lines
The launch centres on the MSPM0G5187, a general-purpose Arm Cortex-M0+ device, and the AM13Ex family, which targets higher-performance real-time control. Both integrate TI’s TinyEngine neural processing unit, which is intended to offload inference work from the main CPU and make local AI practical in systems that would previously have relied on standard MCU compute alone.
TI says the TinyEngine block can cut latency by as much as 92 times and energy per inference by as much as 123 times compared with similar MCUs without a dedicated accelerator. For the smaller MSPM0G5187, that matters because it shifts AI from a specialist feature into something that can plausibly sit inside wearables, appliances, sensor nodes and electrical control products, rather than only in higher-end SoCs or processor-based designs.
The company is also pushing on price. TI says the MSPM0G5187 comes in at under US$1 in 1,000-unit quantities, which is clearly meant to make the case that edge AI microcontrollers are no longer confined to premium designs.
Edge AI microcontrollers for motor control and industrial systems
On the higher-performance side, the AM13Ex combines an Arm Cortex-M33 core, the TinyEngine NPU and real-time control hardware on one chip. TI is positioning that device for appliances, robotics and industrial systems where adaptive control, condition monitoring and predictive maintenance increasingly have to coexist with deterministic control loops.
According to TI, the AM13Ex can manage real-time control for up to four motors while running AI-based adaptive control algorithms. The company also says an integrated trigonometric maths accelerator runs calculations 10 times faster than CORDIC implementations, while the overall integration can trim bill-of-materials cost by up to 30% by removing the need for extra external parts.
The hardware story is only part of the launch. TI is pairing the devices with its CCStudio Edge AI Studio, which now includes more than 60 models and application examples, and with generative AI functions inside the CCStudio IDE for code development, configuration and debugging. That gives TI a stronger tools narrative than a straight silicon launch would have done on its own.
The broader message is that TI wants AI inference to become a standard embedded feature across its MCU portfolio, not an exception. That fits with the wider shift in the MCU market, where, as previously reported by eeNews Europe on TI’s C2000 roadmap, the company has already been adding AI acceleration to real-time control devices. More generally, the debate around how AI is reshaping low-end embedded design is now becoming a defining theme of embedded world 2026, as seen in TI’s launch announcement and the wider MCU discussion taking place in Nuremberg this week.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
