Texas Instruments Incorporated (TI), a U.S. semiconductor corporations, unveiled two new microcontroller (MCU) products (model names MSPM0G5187 and AM13Ex). By applying a Neural Processing Unit (NPU), it enables electronic devices to perform artificial intelligence (AI)-based smart functions. An MCU is an integrated circuit (IC) that miniaturizes and integrates a central processing unit (CPU) and related modules into a single chip. It is a component used as a chip that serves as the "brain of electronic devices."
TI unveiled two new MCU products equipped with an NPU at Embedded World 2026, an embedded electronics trade show held in Nuremberg, Germany, from the 10th to the 12th (local time). TI Korea held an online press briefing on the 11th to introduce the technologies applied to the products and its business vision. Heo Jeong-hyeok, director of technical support at TI Korea, said, "The two new MCUs combined with an NPU are designed to enable smart operations at low power," adding, "The biggest advantage is that AI inference functions can be added in parallel without affecting the MCU's inherent performance."
TI set a strategy to deliver results in the Edge AI market with the new MCUs. Edge AI refers to technology that executes algorithms directly where data is generated. Because data collected from sensors and the Internet of Things (IoT) is processed on the device itself to run AI, it is also called On-device AI. Because it does not send data to a cloud server, latency is short and security is strong, allowing for a wide range of uses. As AI that once operated in the digital realm is mounted on robots and devices and evolves to recognize the physical world and make decisions on its own in the opening "physical AI" era, its importance is seen as rising even further.
TI integrated its in-house NPU, called TinyEngine, into the new MCUs to perform Edge AI functions. It delivers performance of 2.56 GOPS (Giga Operations Per Second, 2.56 billion integer operations per second). The company said, "Compared with equivalent MCUs without an NPU, the new products can reduce latency per AI inference by up to 90 times and energy use by more than 120 times." By using an NPU as a hardware accelerator, it reduced data processing time and improved energy efficiency in AI edge environments.
TI is also providing software solutions to help customers using its MCUs implement AI functions on their devices. Heo said, "By leveraging Generative AI capabilities, you can write code and configure systems with simple natural language inputs," adding, "These new products back TI's strategy to implement Edge AI across its embedded processing portfolio."
TI cited wearables, smart home, and robots as target markets for the new MCUs. The aim is to deliver results in fields expected to grow rapidly as physical AI technology expands. As electronic devices with AI functions proliferate, cases of installing MCUs are also increasing. Typically, refrigerators, washing machines, and dryers that offer smart functions contain about two to three MCUs. In the case of a Humanoid Robot, about 30 to 40 MCUs are expected to be installed. TI priced the new MCU (based on MSPM0G5187) at under $1 per 1,000 units to boost competitiveness.
Heo said, "We are discussing adoption of the new products with various companies, including domestic customers," adding, "We expect higher sales given the strong market response."