Innodisk launched a portfolio of local area network (LAN) modules aimed at edge AI networking workloads.

Its High-Speed 10 Gigabit Ethernet (10GbE) LAN series includes both M.2 and peripheral component interconnect express (PCIe) form factors and are designed for space-constrained edge deployments where low latency and high throughput have been made critical by AI training and inference use cases.

The new series uses Intel E610 and X710 Ethernet controllers and supports technologies including the data plane development kit (DPDK), precision time protocol (PTP), and single root input/output virtualization (SR-IOV).

DPDK is a software framework that accelerates packet processing for high-speed networking applications, while PTP enables precise time synchronization across networked systems. SR-IOV allows a single network device to appear as multiple virtual devices for improved virtualization efficiency.

Innodisk said these capabilities are intended to improve packet processing, synchronization accuracy, and resource utilization for edge AI environments.

The hardware lineup is available in M.2 2242, M.2 2280, and low-profile PCIe formats to simplify deployment across embedded and industrial systems. The products use a daughterboard architecture combined with high-speed shielding cables to reduce integration complexity and accelerate deployment timelines in compact systems.

According to Innodisk, the EGPL-T203 dual-port module is the first M.2-form-factor 10GbE LAN module with wide-temperature support spanning minus 40 degrees Celsius to 85 degrees Celsius ambient temperature operation, while the EGPL-T2F1 offering is claimed to be the industry’s first M.2-based small form-factor pluggable plus (SFP+) LAN module. The product supports both optical and direct-attached copper SFP+ modules for high-speed fiber connectivity.

Additional models in the portfolio include the EGPL-T103, ELPL-T101, and ELPL-T201, which provide different configuration options for customers deploying edge networking infrastructure.

The new lines appear targeted at a growing segment of compact edge AI and small-form-factor workstation deployments where traditional PCIe expansion slots are limited. The company positioned the launch as a response to the limitations of traditional one-gigabit Ethernet (1GbE) connectivity, arguing that older networking standards struggle to support modern edge AI workloads such as real-time inference, industrial automation, video analytics, and multisensor systems.