Xsight Labs introduced the E1L, a low-power DPU designed for edge and embedded systems, with 25W TDP and support for up to 200Gbps networking. The device, part of the E-Series portfolio, will sample in Q2 2027.
Xsight Labs unveiled the E1L, a data processing unit (DPU) tailored for power-constrained environments, marking a significant expansion of its E-Series portfolio. The E1L operates at 25 watts TDP, with a typical draw of 15 watts, and supports up to 200Gbps networking through 4 SerDes lanes. It is built on Arm Neoverse N2 cores and leverages TSMC’s N5 process, enabling it to scale from edge deployments to data center applications.
Detailed Technical Specifications
The E1L offers three core configurations—8, 16, or 22 Arm Neoverse N2 cores—and supports up to 2 DDR5-5200 memory interfaces with inline encryption. Its power envelope ranges from 25W to 40W, with typical draws of 15W, 18W, and 25W, respectively. The device integrates 20 PCIe Gen5 lanes across six dual-mode controllers, enabling flexible deployment as a discrete device, a COM Express module, and a 1RU server. The E1L’s COM Express Type 7 form factor allows system designers to integrate it into existing chassis designs, reducing thermal and space constraints.
Unlike AMD Pensando’s P4-programmable packet-processing engines, which prioritize specialized architectures, Xsight Labs’ E1L relies on standard Arm cores running Linux with DPDK. This approach ensures cross-platform compatibility, allowing software written for the 800G E1 to function on the E1L. This design choice simplifies porting workloads, as both devices share the same Neoverse N2 core and DPDK stack.
Industry Context for Power-Efficient Infrastructure
The E1L addresses a growing demand for power-efficient infrastructure in edge and embedded systems, where traditional 800Gbps DPUs exceed thermal and power budgets. Xsight Labs emphasized that the E1L’s architecture enables deployment in switches, storage nodes, and other applications where power consumption is a critical constraint. The device will sample in Q2 2027, and Xsight Labs will present the E1L at the 2026 Open Compute Project (OCP) Global Summit, October 12 to 15, in San Jose, as an OCP Bronze member.
Yossi Meyouhas, CEO of Xsight Labs, stated that the E1L fills a gap in the market by providing DPU capability within existing power envelopes. Eddie Ramirez of Arm highlighted the benefits of using Arm Neoverse CSS N2 for programmable, efficient compute from the edge to the data center.
These statements underscore the E1L’s role in supporting AI infrastructure expansion, where performance per watt and software flexibility are critical.
Implications for AI and Edge Computing
By offloading infrastructure tasks from host CPUs, the E1L reduces latency and improves isolation for critical systems. The device’s 15W to 25W typical draw provides headroom for adjacent components, making it suitable for dense switch or storage configurations.
Comparisons with Nvidia’s BlueField-2 and BlueField-4 DPUs reveal that the E1L’s 22-core variant, while not matching high-end host CPUs in single-threaded performance, is optimized for I/O-bound workloads like storage control planes and virtualization hosts. This positions the E1L as a strategic alternative for applications where power efficiency and compatibility with standard Linux tools are prioritized over raw processing power.
The E1L’s launch reflects broader industry trends toward scalable, open-standard silicon. By offering a unified software stack from 100G at the edge to 800G in the core, Xsight Labs aims to simplify infrastructure management across diverse environments. The device’s availability in Q2 2027 will determine its impact on markets reliant on power-constrained deployments.