Brand: Intel | Category: GPUs
SKU: INTEL-GAUDI3 | Part #: Gaudi-3 | MPN: HL-338
Contact for Pricing — Request a Quote
The Intel Gaudi 3 AI Accelerator from Intel is enterprise-grade GPUs hardware built for data centers. It features 16 GB HBM2e capacity, PCIe 4.0 x16 form factor. The Intel Gaudi 3 AI Accelerator, a cutting-edge addition to the Intel GPU lineup, is designed to empower enterprises and data scientists with unparalleled AI training capabilities. As part of. Available in brand new condition. Intel gpus products are trusted by enterprises and data centers worldwide. Available from Omnixon Global. Contact our sales team to request a quote.
| Brand | Intel |
| Category | GPUs |
| SKU | INTEL-GAUDI3 |
| Part Number | Gaudi-3 |
| MPN | HL-338 |
| Condition | New |
| Capacity | 16 GB HBM2e |
| Form Factor | PCIe 4.0 x16 |
| Architecture | Intel Gaudi 3 |
| GPU Count | 3x |
| AI Optimized | Yes |
| Process Technology | 7nm |
| Max Memory | 16 GB HBM2e |
| Memory Bandwidth | 1.2 TB/s |
| Compute Performance | 250 TFLOPS (FP16) |
| Power Consumption | 300W |
| Interconnect | Intel Ultra Path Interconnect (UPI) |
| Form Factor | PCIe 4.0 x16 |
| Supported Frameworks | TensorFlow, PyTorch, MXNet |
| Cooling Solution | Active Air Cooling / Liquid Cooling options |
The Intel Gaudi 3 AI Accelerator accelerates AI/ML training, inference, scientific HPC, and virtualization (vGPU) workloads. Typical deployments include LLM training clusters, computer-vision pipelines, financial risk modeling, and rendering farms.
Key specifications for the Intel Gaudi 3 AI Accelerator: 3 Years support; new condition; architecture Intel Gaudi 3; process technology 7nm; memory 16 GB HBM2e; memory bandwidth 1.2 TB/s; compute performance 250 TFLOPS (FP16); power consumption 300W. Manufacturer part number Gaudi-3. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
The Intel Gaudi 3 AI Accelerator requires a PCIe Gen4 or Gen5 x16 slot, server power adequate for the card's TDP, and CUDA/ROCm driver support in your hypervisor or bare-metal OS. Sales engineering will confirm chassis fit (1U/2U/4U), PCIe lane count, and PSU headroom before quoting.