Boardcon Launches PICOT536 SoM and EMT536 SBC with Allwinner T536

The embedded systems vendor Boardcon announced the PICOT536 System-on-Module and EMT536 single-board computer on April 30, 2026, according to the company's official announcement. Both products center on the Allwinner T536 SoC, a chip that combines traditional Arm processing with RISC-V coprocessors for edge AI workloads.

The T536 integrates a quad-core Arm Cortex-A55 CPU running up to 1.6 GHz alongside two RISC-V cores: the XuanTie E907 at 600 MHz and the XuanTie E902 at 200 MHz. An integrated 2 TOPS NPU handles on-device neural inference. This heterogeneous architecture isn't uncommon in industrial hardware, but it does require developers to understand how workloads distribute across different instruction sets (which adds complexity to the build pipeline).

Per the official Boardcon product announcement, the EMT536 development board exposes a dense array of industrial interfaces: nine UART ports, two RS485 channels, two CAN buses, dual Gigabit Ethernet, and mPCIe expansion. Display output supports either MIPI DSI or dual-LVDS configurations. The board physically occupies standard SBC form factors, though the connector density means cable management becomes a real consideration during prototyping.

The PICOT536 SoM ships with configurable memory ranging from 2GB to 8GB LPDDR4/LPDDR4X with inline ECC protection. Storage options span 8GB to 64GB eMMC. The module uses a 314-pin MXM 3.0 edge connector that exposes LVDS and MIPI DSI display outputs, support for 2×4-lane or 4×2-lane MIPI CSI camera inputs, Gigabit Ethernet, PCIe, USB 3.0/PCIe 2.1, four USB 2.0 host ports, and one USB 2.0 OTG port. Additional I/O includes UART, I2C, SPI, PWM, ADC, and audio interfaces.

Video processing capabilities listed by CNX Software include decoding up to 4K at 15 fps (MJPEG) or 1080p at 60 fps (JPEG), with encoding support up to 4K at 25 fps (H.264). Optional wireless connectivity arrives via a VS6621S80 module providing WiFi 6 and Bluetooth 5.4. These specs place the hardware squarely in the cost-sensitive industrial segment rather than high-performance AI inference territory.

Boardcon supplies a Buildroot-based software environment with Linux kernel 5.15.147 and U-Boot 2023.04. The company recommends Ubuntu 22.04 for cross-compilation development and provides toolchains and debugging utilities. This kernel version is LTS but not cutting-edge, which matters for long-term industrial deployments where stability outweighs feature freshness.

The boards target industrial HMI, machine vision, robotics, and edge AI applications. In practice, this means factory floor displays, basic object detection, sensor fusion, and control systems where deterministic timing matters more than raw throughput. The 2 TOPS NPU handles modest inference tasks—think anomaly detection or simple classification rather than large language models or complex computer vision pipelines.

Industry observers note that pairing small NPUs with heterogeneous CPU and RISC-V microcontrollers represents a common pattern in embedded compute. The goal is balancing deterministic control with ML inference at the edge without the power draw or cost of GPU-class hardware. For practitioners, the real question becomes whether the T536's driver ecosystem supports their specific camera sensors, display panels, and industrial protocols.

What to watch for integrators: pricing details, long-term availability commitments, mainline kernel support status, and upstream driver availability. Extended-temperature operation and supply chain guarantees matter more than raw specs in industrial deployments. Boardcon's documentation on camera and display stacks will determine whether this hardware integrates smoothly into existing workflows or requires significant customization.

Whether the T536's mixed architecture delivers practical advantages over pure Arm or pure RISC-V alternatives remains to be seen. The hardware exists. The software stack exists. Whether developers actually adopt it depends on documentation quality, community support, and whether the price point justifies the complexity of managing three different processor types on one board.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn