Lanner And Qualcomm Technologies Deliver Integrated Platforms For AI On RAN And On-Premises Enterprise LLMs

Lanner Electronics, a leading provider of network appliances and edge AI computing platforms, today announced the successful integration of the Qualcomm Cloud AI 100 Ultra across two premier Edge AI platforms. The launch introduces the ECA-5555 Edge AI Server, engineered for AI on RAN, and the EAI-I732 Edge AI Workstation, optimized for on-premises enterprise large language model (LLM) inference.

By uniting Lanner’s system-level hardware design with Qualcomm Technologies’ dedicated AI acceleration, the joint solutions provide a scalable foundation for telecom operators, private 5G providers, and enterprises seeking to deploy robust local AI services.

ECA-5555: Bringing AI Applications to the RAN Edge
The ECA-5555 Edge AI Server combines Intel Xeon 6 SoC processing with Qualcomm Cloud AI 100 Ultra acceleration to establish a powerful computing foundation at the radio access network (RAN) edge.

Designed to process AI workloads close to connected devices, the platform effectively mitigates latency and backhaul bandwidth constraints. This architecture is specifically targeted at advancing applications in smart factories, corporate campuses, and connected enterprise environments.

By enabling edge-native video analytics, contextual analysis, and interactive AI assistants, the ECA-5555 equips telecom operators with the infrastructure to transition from providing simple connectivity to delivering comprehensive, local AI services.

EAI-I732: Verified Llama 3.1 8B Performance for Enterprise AI
The EAI-I732 Edge AI Workstation integrates Qualcomm’s acceleration into Lanner’s modular edge platform, enabling organizations to develop and host enterprise LLM services entirely within their own infrastructure.

To validate its on-premises configuration, Lanner conducted a comprehensive sanity check using the Llama 3.1 8B model. The EAI-I732 demonstrated reference-level or better performance across all assessed inference metrics, yielding the following results:

• Inference Performance: Throughput, latency, time to first token (TTFT), and time per output token (TPOT) matched or exceeded their respective reference values.
• Thermal Stability: The system successfully passed PowerStress tests, maintaining device temperatures well below reference limits under sustained loads.
• Benchmark Accuracy: The benchmark run duration differed from the reference configuration by less than 1%.
• Configuration Alignment: All required Llama 3.1 8B test parameters matched the reference configuration precisely.

This validated setup is targeted at enterprise applications such as internal knowledge assistants, document summarization, and retrieval-augmented generation (RAG). By localizing document retrieval and LLM inference, organizations can connect AI services to proprietary internal data while retaining absolute control over information security and processing location.

A Unified Acceleration Architecture
Both platforms share a common, high-performance hardware foundation driven by the Qualcomm Cloud AI 100 Ultra accelerator, which delivers up to 870 TOPS of INT8 performance, 128GB of onboard memory, and a 150W thermal design power (TDP) per card.

The hardware is supported by a comprehensive software ecosystem, including the Qualcomm® AI Inference Suite. This suite equips developers with essential tools for generative AI application management, native support for RAG, and OpenAI-compatible APIs to streamline deployment.

Together, the ECA-5555 and EAI-I732 extend Lanner’s Qualcomm-accelerated portfolio from the network edge to the enterprise data center, providing a highly capable, measured reference point for organizations evaluating localized LLM infrastructure.

For configuration details and solution inquiries, visit www.lannerinc.com.

Featured Products
ECA-5555
Multi-access Edge Computing Appliance With Intel Xeon 6 SoC Processor

For more information visit, https://www.lannerinc.com/products/network-appliances/x86-rackmount-network-appliances/eca-5555.

EAI-I732
Edge AI Platform with Intel Core Processors Series 2 (Bartlett Lake-S)

For more information visit, https://www.lannerinc.com/products/edge-ai-appliance/edge-inference-appliances/eai-i732.