Intel Data Center GPU Flex 170 1x32GB GDDR6 PCIe

The Intel Data Center GPU Flex 170 is a purpose-built enterprise graphics processing unit designed for media delivery, cloud gaming, and AI inference workloads at scale. Based on Intel's Xe-HPG microarchitecture and manufactured on the TSMC N6 process node, the Flex 170 integrates a single GPU tile with 32 GB of GDDR6 memory, delivering high memory bandwidth and capacity suited to simultaneous multi-stream processing and inference serving. Its PCIe 4.0 x16 host interface enables broad compatibility with modern server platforms without requiring proprietary interconnects.

The GPU Flex 170 is engineered specifically for the demands of media transcode at scale, supporting hardware-accelerated AV1, HEVC, AVC, and VP9 encode and decode across dozens of simultaneous streams per card. Intel's oneAPI software stack, including the oneVPL media library and OpenVINO inference runtime, provides a unified programming model that reduces integration complexity across heterogeneous datacenter environments. The card operates within a 150 W TDP envelope and ships in a standard dual-slot, full-height full-length form factor, making it deployable in standard 1U and 2U server chassis used across hyperscale and enterprise datacenters.

Targeted at cloud service providers, telecommunications operators, and enterprise IT organizations running GPU-accelerated virtual desktop infrastructure or AI inference pipelines, the Intel Data Center GPU Flex 170 supports SR-IOV-based virtualization to enable fine-grained, multi-tenant GPU partitioning. Its feature set aligns with open-standard frameworks including OpenCL, Vulkan, and DirectX 12 Ultimate, ensuring compatibility with a broad ecosystem of workloads. Omnixon Global supplies this solution to enterprise buyers across the UAE, GCC, EMEA, and APAC regions.

Ideal for

• High-density video transcode for streaming platforms requiring simultaneous AV1, HEVC, and AVC encode across dozens of streams per GPU
• AI inference serving for computer vision and natural language processing models deployed at scale in datacenter inference clusters
• Cloud gaming infrastructure leveraging SR-IOV GPU virtualization to serve multiple concurrent gaming sessions per physical card
• Virtual desktop infrastructure (VDI) deployments partitioning GPU resources across enterprise end users via hardware virtualization
• Media analytics pipelines combining hardware-accelerated decode with OpenVINO-powered inference for real-time content analysis
• Telecommunications and broadcast workflows requiring low-latency, high-throughput video processing integrated with standard server hardware

Technical specifications

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