Congatec Module Adds NPU-Based AI to Entry-Level x86 Embedded Designs

Congatec recently introduced the Conga-TC300 COM Express Compact module, built on Intel Core Series 3 processors. The device targets low-power, cost-sensitive edge computing applications that now require integrated AI capabilities. Congatec claims the module may be useful for systems that need more processing power than legacy entry-level platforms, while still meeting constrained power and thermal requirements.

The Conga-TC300 COM Express Compact module with Intel Core Series 3 processors for edge AI and embedded applications. 

The module adds a dedicated NPU and a hybrid CPU architecture to the entry-level x86 space, enabling systems based on Intel Atom or Celeron platforms to add AI acceleration without a full redesign. Along with scalable performance, standard industrial interfaces, and available software options, it can be used as a relatively straightforward upgrade for embedded and edge systems. 

AI Acceleration Moves Into Entry-Level x86

The Conga-TC300 will feature a dedicated NPU alongside CPU and GPU resources. The Intel Core Series 3 processors combine two performance cores with four low-power, efficient cores, delivering a hybrid architecture that balances compute performance and energy efficiency.

The TC300 platform (datasheet linked) delivers up to 41 TOPS when combining NPU, GPU (Xe3), and CPU acceleration. The NPU can deliver up to 18 TOPS for inference workloads, which is significant for entry-level x86 modules that previously relied solely on CPU or GPU resources. This allows AI tasks such as object recognition, voice processing, or local inference to run more efficiently without overloading general-purpose compute resources.

An Upgrade Path for Existing Embedded Designs

Congatec designed the Conga-TC300 as a practical upgrade for systems currently based on Intel Atom or Celeron processors. Because it follows the COM Express Compact Type 6 form factor, existing carrier board designs can often be reused or adapted with minimal changes.

This approach reduces redesign effort while allowing developers to add higher performance, AI acceleration, and newer interfaces. The scalable TDP range from 12 W to 28 W also helps maintain similar thermal design constraints, which is important when upgrading systems already deployed in space- or power-limited environments.

Memory, I/O, and System Integration

Congatec’s new module supports up to 64 GB of DDR5 memory at up to 6400 MT/s, with optional in-band ECC for improved reliability. Storage options include onboard UFS 3.1, and connectivity via 2.5 GbE, with optional TSN and TCC support for time-sensitive applications.

Block diagram for the Conga-TC300. 

Peripheral expansion is handled through up to eight PCIe lanes, eliminating the need for an external PCIe switch in many cases and simplifying carrier board design. Additional interfaces include USB4, USB 3.2, SATA, UART, GPIO, and common embedded buses such as I2C and SPI, making the module suitable for a wide range of industrial and edge applications.

Software Stack and aReady.COM Ecosystem

The aReady.COM concept aims to reduce the amount of software integration work required during development. Instead of starting from a bare module, developers can use preconfigured operating systems such as Ubuntu Pro, ctrlX OS, or KontronOS, along with virtualization support through "Conga-zones."

This allows multiple workloads, such as real-time control, HMI, AI processing, and gateway functions,  to run on a single module. Additional building blocks for IoT connectivity, remote management, and system monitoring are also available, which help accelerate deployment in connected edge systems. For developers, this reduces the effort required to bring up both hardware and software, particularly in applications where time-to-market is a constraint.

The Conga-TC300 brings a dedicated NPU, a hybrid CPU architecture, and enhanced software support to the entry-level COM Express space. The TC300 maintains compatibility with existing COM Express designs while adding support for newer interfaces and higher performance, simplifying the process of upgrading legacy systems. Congatec’s Conga-TC300 will feature preconfigured software options and built-in AI acceleration, providing a practical way to add AI functionality to embedded and edge applications without requiring a complete platform redesign or major changes to system architecture.

All images used courtesy of Congatec.