Edge AI is moving closer to the sensors, motors and embedded controllers that run physical machines. At the D Forum 2026 Microcontroller Forum in Taiwan, Artery Technology showcased its AT32 MCU portfolio across edge AI sensing, robot joint control and high-speed server cooling, highlighting how increasingly capable microcontrollers are becoming part of the hardware foundation for intelligent devices.
Artery Targets the Next Edge AI Layer With AT32 MCUs
Artificial intelligence is increasingly moving out of centralized cloud environments and into the devices that generate data and control physical systems.
That shift was the focus of Artery Technology’s presentation at the D Forum 2026 Microcontroller Forum on August 7, where the company outlined how its AT32 microcontroller (MCU) portfolio is being positioned for edge intelligence, robotics and high-performance motor control.
The company’s keynote, titled “AT32 MCUs Drive Edge Intelligence into a New Era,” was accompanied by demonstrations spanning AI sensing, robot joint control and server cooling fans.
The common thread is increasingly capable embedded computing.
Rather than sending every sensor reading to a cloud or centralized AI system, edge devices can increasingly perform inference and control locally. That can reduce latency, lower bandwidth requirements and improve responsiveness in applications where milliseconds matter.
Artery’s strategy is to bring those capabilities into MCUs that sit directly inside products and machines.
Making Edge AI Easier to Deploy
One of the company’s main demonstrations combined the AT32 Edge AI Sensor EV Board with AT32 AI Studio, Artery’s development environment for deploying AI models on its microcontrollers.
The workflow is designed to cover data collection, model training, optimization and deployment within a single development process.
That matters because edge AI development has traditionally involved multiple tools and engineering stages. Developers may need to collect sensor data, train a model using separate machine-learning infrastructure, optimize it for an embedded processor and then integrate the resulting model into firmware.
Artery is attempting to simplify that path.
The demonstration uses sensors including time-of-flight (ToF) and inertial measurement units (IMUs). Potential applications include gesture recognition, motion classification and equipment anomaly detection.
These are relatively small AI workloads compared with the large language models driving generative AI, but they represent an important part of the expanding edge AI market.
A local MCU can continuously analyze sensor information without needing to transmit raw data elsewhere. For industrial equipment or consumer devices, that can mean faster responses and potentially lower connectivity and cloud costs.
MCUs Move Into Embodied AI
Robotics provided a second demonstration of how embedded intelligence is changing.
The company showed applications involving dexterous hands and robot joints, reflecting the growing interest in embodied AI and humanoid robots.
Modern robots require numerous control loops to operate motors, read position sensors and coordinate movement. The computing requirements are different from those of a cloud AI model, but the underlying principle is similar: intelligence increasingly needs to exist close to the physical system it controls.
Artery’s AT32F435/437 series is gaining a compact 7 × 7 mm BGA100 package, which the company says is intended for space-constrained applications such as dexterous hands and robot joints.
Its AT32M416 motor-control MCU is aimed more specifically at motion systems, combining CAN-FD, high-resolution PWM, a 12-bit ADC, DSP/FPU capabilities and magnetic-encoder feedback.
The combination supports closed-loop motor control, where sensor feedback is continuously used to adjust motor behavior.
That architecture is relevant not only to humanoid robots but also to quadruped machines, industrial automation equipment and other systems requiring precise movement.
The Data Center Creates Another MCU Opportunity
The third application is further removed from AI inference but directly connected to the AI infrastructure boom.
Artery demonstrated a server cooling fan solution designed for AI data centers, where high-performance computing generates significant heat and places demanding requirements on cooling systems.
The solution integrates an MCU, gate driver and DC-DC converter into a 4 × 4 mm QFN32 package.
The small footprint is important because cooling fans operate within tightly constrained server designs. The controller also runs a field-oriented control (FOC) algorithm intended to improve motor stability while reducing noise and vibration.
The system is designed to support startup under reverse airflow and stable operation at high speed and static pressure.
The application illustrates an often-overlooked aspect of AI infrastructure: AI acceleration depends not only on processors and networking but also on the embedded control systems that keep servers operating within safe thermal conditions.
As data centers deploy more accelerators, cooling becomes an increasingly important engineering challenge. MCUs can provide the real-time control layer required by fans, pumps and other thermal-management equipment.
Higher-Performance AT32 Chips Expand the Portfolio
Artery also previewed two upcoming MCU families.
The AT32F406/408 series is based on the Arm Cortex-M4F architecture and is expected to deliver up to 216 MHz processing performance, with 512 KB of Flash and 192 KB of SRAM.
The chips integrate three high-speed ADCs, High-Speed USB OTG and QSPI interfaces, targeting applications requiring rapid data transfer and low-latency control. Artery identifies gaming keyboards, game controllers and USB peripherals as potential applications.
The higher-end AT32F403E/407E series is designed to reach up to 320 MHz, with 960 KB of Flash and 224 KB of SRAM. It supports interfaces including Ethernet on the AT32F407E, USB and CAN.
The company is targeting these devices at more computationally demanding embedded applications, including service robots, smart appliances, electric two-wheelers and industrial control.
Together, the products show how MCU vendors are competing not simply on clock speed but on the combination of processing capability, memory, connectivity, package size, peripherals and real-time control.
Edge Intelligence Becomes an Embedded Hardware Competition
Artery’s strategy places it within a broader semiconductor market where companies are attempting to bring AI capabilities closer to the physical edge.
Major chipmakers including NVIDIA, Qualcomm and NXP Semiconductors are developing technologies for edge AI, embedded computing and intelligent devices, while STMicroelectronics and other MCU suppliers are adding greater processing and AI capabilities to embedded platforms.
The competitive landscape is increasingly fragmented by workload.
A sophisticated vision model may require an edge AI accelerator or application processor. A sensor-class task such as gesture recognition or anomaly detection may be better suited to a low-power MCU.
That distinction creates a large opportunity for microcontrollers that can execute useful AI workloads while maintaining the low power consumption, real-time behavior and cost characteristics that embedded systems require.
Artery’s demonstrations suggest the company is targeting precisely that intersection.
The Bigger Shift Is From Connected Devices to Intelligent Devices
The evolution of the MCU is important because microcontrollers are embedded in enormous numbers of products.
If those devices become capable of performing meaningful AI inference locally, intelligence can spread across everything from appliances and vehicles to industrial equipment and robots.
That does not mean every MCU will become an AI processor. Instead, the industry is likely to develop a hierarchy of edge computing, with workloads distributed between sensors, MCUs, dedicated accelerators, edge processors and cloud systems.
Artery’s AT32 portfolio is designed to occupy several points within that hierarchy.
Its focus on AI sensing, robotics and data-center cooling also highlights an important characteristic of edge intelligence: AI does not have to be the visible product to be commercially important.
Sometimes AI is the system that recognizes a gesture. Sometimes it controls a robot’s movement. And sometimes it quietly optimizes the fan keeping an AI server from overheating.
As AI becomes embedded throughout physical infrastructure, those seemingly smaller computing workloads could become an increasingly important semiconductor market of their own.
Market Landscape
The edge AI market is developing across several hardware categories, from microcontrollers and sensors to AI accelerators and application processors.
MCUs have traditionally been valued for low power consumption, deterministic control and low cost. The emergence of lightweight AI inference is expanding their role into intelligent sensing and local decision-making.
The broader ecosystem includes companies such as NVIDIA, Qualcomm, NXP and STMicroelectronics, alongside Arm’s processor architecture ecosystem.
Artery’s positioning focuses on combining:
- Edge AI model deployment
- Real-time sensor processing
- Motor and robotics control
- High-speed embedded connectivity
- Compact device packaging
- AI data-center thermal management
For enterprise developers, the key consideration is workload fit. MCU-based AI will not replace cloud-scale models, but it can complement them by handling localized inference and control where low latency, power efficiency and reliability matter.
Top Insights
- Artery’s AT32 MCU portfolio targets edge intelligence across AI sensing, robotics and high-speed motor control, bringing more computation directly into physical devices.
- AT32 AI Studio combines data collection, model training, optimization and MCU deployment to simplify the development cycle for embedded AI applications.
- The AT32M416 adds motor-control capabilities including CAN-FD, high-resolution PWM and magnetic-encoder feedback for precise robotic motion.
- Artery’s compact server cooling controller highlights an overlooked AI infrastructure market: MCU-driven thermal management for high-performance computing systems.
- New AT32F406/408 and AT32F403E/407E families expand processing, memory and connectivity options for robots, industrial controls, appliances and embedded devices.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
