Intel Data Center GPU Flex 140 2x8GB GDDR6 PCIe

The Intel Data Center GPU Flex 140 (part number GDCFLX140) is a purpose-built enterprise GPU based on Intel's Xe-HPG microarchitecture, designed to deliver efficient, high-throughput media and AI inference performance in space- and power-constrained data center environments. The card integrates two independent 8GB GDDR6 memory stacks on a single PCIe add-in card, providing 16GB of total on-card memory accessible across dual GPU tiles, enabling flexible workload partitioning and multi-tenant deployments at scale.

Engineered for cloud service providers, streaming platforms, and enterprise media operators, the Flex 140 delivers hardware-accelerated AV1, HEVC, AVC, and VP9 encode and decode capabilities alongside support for Intel's XMX (Xe Matrix Extension) AI acceleration engines. Its PCIe 4.0 host interface and low-profile, half-height form factor make it well suited to high-density 1U and 2U rack configurations where thermal budget and slot availability are at a premium.

The GPU is supported under Intel's open-source oneAPI software stack, including the oneAPI Video Processing Library (oneVPL) and OpenCL, enabling integration into existing media transcoding pipelines, AI inference frameworks, and cloud-native virtualized environments. Intel's SR-IOV support facilitates fine-grained hardware partitioning for virtualized media and inference workloads across multiple simultaneous tenants, a capability particularly relevant to operators running large-scale video-as-a-service or GPU-as-a-service platforms.

Ideal for

• High-density cloud video transcoding at scale, supporting simultaneous multi-session AV1, HEVC, and AVC encode/decode for OTT and streaming platforms
• AI inference serving for lightweight to mid-scale models in latency-sensitive enterprise applications, leveraging XMX matrix acceleration engines
• Virtualized GPU deployments using SR-IOV for GPU-as-a-service offerings in multi-tenant cloud and managed hosting environments
• Adaptive bitrate (ABR) ladder generation for live and video-on-demand workflows requiring high concurrent stream density per rack unit
• Edge and regional data center media processing nodes where power efficiency, physical form factor, and TCO per stream are primary constraints
• Enterprise AI analytics pipelines integrating with oneAPI-based frameworks for image classification, object detection, and video analytics workloads

Technical specifications

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