Gigabyte G262-IR1-AAX1 2-GPU L40S PCIe Server SKU

The Gigabyte G262-IR1-AAX1 is a 2U rack-mount server engineered specifically to support dual NVIDIA L40S PCIe graphics accelerators, delivering a high-density GPU compute platform optimized for enterprise AI inference, professional visualization, and data-intensive workloads. Built on Gigabyte's proven G262 server chassis lineage, this SKU pairs the computational density of two full-length PCIe GPU slots with a robust server architecture capable of sustaining the sustained thermal and power demands of the NVIDIA L40S — a GPU built on the Ada Lovelace architecture with third-generation RT Cores, fourth-generation Tensor Cores, and 48 GB of GDDR6 ECC memory per card.

The G262-IR1-AAX1 is designed to serve as a versatile multi-function GPU server, supporting NVIDIA's Ada Lovelace-generation L40S accelerators which deliver up to 91.6 TFLOPS of FP32 performance and 733 TOPS of INT8 inference throughput per card. With two L40S GPUs installed, the platform aggregates 96 GB of total GPU memory and provides substantial AI inference and training bandwidth within a 2U footprint. The server's PCIe architecture enables straightforward integration into existing data center fabric without requiring NVLink-based topology, making it compatible with a broad range of enterprise HPC and AI deployment patterns.

Targeted at datacenter operators, enterprise AI teams, and cloud service providers across the UAE, GCC, EMEA, and APAC regions, the G262-IR1-AAX1 supports workloads spanning large language model (LLM) inference, generative AI serving, 3D rendering pipelines, video transcoding at scale, and virtual workstation hosting. The server's 2U form factor optimizes rack space utilization while preserving sufficient airflow headroom for the L40S GPUs' 350 W TDP per card, and the platform supports standard data center power and management infrastructure including IPMI-based remote management.

Ideal for

• Large language model (LLM) inference serving using both L40S GPUs in a high-throughput, low-latency production environment
• Generative AI application hosting, including image synthesis and multi-modal model inference, leveraging Ada Lovelace Tensor Core performance
• Professional 3D visualization and GPU-accelerated rendering pipelines for media, engineering, and architecture workflows
• Virtual workstation and VDI deployments using NVIDIA vGPU software on L40S to support multiple concurrent professional users
• AI-accelerated video transcoding and streaming analytics at scale within broadcast, media, and surveillance infrastructure
• Enterprise HPC workloads including scientific simulation and data analytics requiring large aggregate GPU memory across PCIe-attached accelerators

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.