Skip to main content
The Quantum Dispatch
Back to Home
Cover illustration for Boardcon's PICOT536 SoM and EMT536 SBC Bring 2 TOPS Edge AI to the Allwinner T536

Boardcon's PICOT536 SoM and EMT536 SBC Bring 2 TOPS Edge AI to the Allwinner T536

Boardcon launched the PICOT536 system-on-module and matching EMT536 single board computer on May 4, 2026 — pairing the Allwinner T536's quad-core Cortex-A55, RISC-V coprocessors, and a 2 TOPS NPU for industrial HMI, machine vision, and edge AI builds.

Alex Circuit
Alex CircuitMay 4, 20266 min read

A Purposeful Edge AI Pairing for Industrial Builders

Boardcon rolled out the PICOT536 system-on-module and the matching EMT536 single board computer on May 4, 2026 — a credible new entry in the industrial edge AI tier built around the Allwinner T536 SoC. For embedded engineers, industrial HMI builders, machine vision designers, and the broader edge AI maker community, the PICOT536/EMT536 pair is a useful addition to the spring 2026 SBC landscape: a modular, well-specced platform purpose-built for the kind of compact edge AI workloads that have become the dominant theme in the embedded computing market.

The release reflects a pattern we have been tracking across the 2026 SBC cycle. Modular system-on-module platforms paired with carrier-board reference designs have become the standard delivery shape for industrial-grade SBCs, because they let one silicon platform serve a wide range of mechanical and I/O configurations without forcing the silicon vendor to ship dozens of bespoke board variants. The PICOT536 SoM and EMT536 SBC follow that pattern cleanly.

What the Allwinner T536 Brings to Edge AI

The Allwinner T536 SoC at the heart of the platform is a heterogeneous compute design that lines up well with modern edge AI expectations. It packs a quad-core Arm Cortex-A55 cluster as the primary application processor, adds an XuanTie E907 RISC-V coprocessor and an XuanTie E902 RISC-V coprocessor for real-time and low-power tasks, and layers in a 2 TOPS NPU for INT8 inference workloads. The combination matters because it gives builders a clean separation between general-purpose Linux workloads on the A55 cluster, deterministic real-time control on the RISC-V cores, and dedicated neural inference on the NPU.

The 2 TOPS NPU is well-sized for the target workload class. Industrial HMI applications, smart camera vision pipelines, predictive maintenance models running on machine vibration data, and the broader category of compact edge AI use cases sit comfortably inside a 2 TOPS inference envelope, and the heterogeneous architecture means none of those workloads have to compete with general-purpose application code for the same compute pool.

The Module's Specs

The PICOT536 SoM ships with up to 8GB of LPDDR4 or LPDDR4X memory, up to 64GB of eMMC flash storage, an optional Wi-Fi 6 and Bluetooth 5.4 module, and a 314-pin MXM edge connector that exposes a comprehensive I/O surface. The connector breaks out MIPI DSI and LVDS display interfaces, multiple camera input lanes, audio I/O, gigabit Ethernet, USB 3.0 and USB 2.0 ports, multiple UART, SPI, and I2C buses, and the GPIO pins that industrial HMI and machine vision designs depend on. The matching EMT536 SBC pairs the SoM with a reference carrier that exposes those interfaces in a development-friendly form factor.

For builders deciding between a fixed-form-factor SBC and a modular SoM-plus-carrier approach, the PICOT536/EMT536 combination is a useful reminder of why the modular pattern keeps winning industrial design wins. The same SoM that ships on the EMT536 development carrier can be dropped into a custom carrier for a production HMI, a smart camera enclosure, or a robotics control box without re-validating the core compute and memory subsystem.

Where It Fits in the Spring 2026 SBC Landscape

The Allwinner T536 platform lands alongside an unusually busy spring 2026 SBC market. The Banana Pi BPI-OM7 brought integrated 3D vision to the RK3588 ecosystem. The Boardcon EM3326S delivered a $25-class smart-audio SBC. The M5Stack Cardputer Mesh Kit shipped a card-sized off-grid Meshtastic terminal. The ESP32-C5 Mini squeezed Wi-Fi 6 and Bluetooth into a USB-C stick. Where those platforms each chase a different capability axis, the PICOT536/EMT536 pair targets the industrial edge AI middle — capable enough for real machine vision and predictive maintenance workloads, modular enough for production deployment, and reasonably priced for the smaller edge AI integrators who do not have the volume to justify a fully custom SoC platform.

For developers building toward industrial HMI, smart camera, robotics control, or compact edge AI use cases, the PICOT536 SoM is worth a serious look. The 2 TOPS NPU plus heterogeneous Cortex-A55 and RISC-V architecture is the right compute mix for the workload class, the modular SoM design simplifies the path from prototype to production, and the May 4 release timing means working developers can build today against a platform that is shipping in volume.

Sources: CNX Software Boardcon PICOT536 SoM and EMT536 SBC Allwinner T536 Edge AI Coverage (May 4, 2026), Allwinner T536 Datasheet, Boardcon PICOT536 Product Documentation