Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray (8x OAM)
Brand: Intel | Category: GPUs
SKU: M50CYP-GAUDI3-8OAM | Part #: M50CYP-GAUDI3-8OAM | MPN: M50CYP-GAUDI3-8OAM
Contact for Pricing — Request a Quote
About the Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray (8x OAM)
For AI training and inference workloads demanding massive parallel compute density, the Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray delivers purpose-built architecture in a compact 2U form factor. This system integrates eight Intel Gaudi 3 accelerators built on TSMC's advanced 5nm process node, each equipped with 64 tensor processor cores and 128 GB of HBM2e memory for a total of 1 TB aggregate accelerator memory. The all-to-all non-blocking RoCE v2-based fabric interconnect provides 2.4 Tbps of aggregate internal bandwidth, enabling efficient multi-accelerator scaling for large-scale model training and inference.
The platform supports multiple precision formats—FP8, BF16, and FP32—to optimize both performance and accuracy across diverse AI workloads. Each OAM accelerator features 24x 200GbE RDMA-capable ports for high-speed network connectivity, while PCIe Gen 5 host interfaces connect to the Intel Server System M50CYP2SBSTD chassis. The system ships with Intel SynapseAI SDK and supports popular frameworks including PyTorch and TensorFlow, making it ideal for AI infrastructure teams and HPC-focused organizations. Designed to the Open Accelerator Module (OAM) specification, this configuration (part number M50CYP-GAUDI3-8OAM) ensures modularity and interoperability within enterprise AI deployments. Contact Omnixon Global for an RFQ on this Intel accelerator solution.
Key Specifications
- Product Name: Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray (8x OAM)
- Part Number: M50CYP-GAUDI3-8OAM
- Accelerator Architecture: Intel Gaudi 3 (8x OAM, TSMC 5nm)
- Total Accelerator Memory: 1 TB (8 x 128 GB HBM2e)
- Internal Fabric Bandwidth: 2.4 Tbps all-to-all non-blocking RoCE v2
- Network Ports per OAM: 24x 200GbE RDMA-capable ports
- Supported Precisions: FP8, BF16, FP32
- Form Factor: 2U Rack
- Framework Support: PyTorch, TensorFlow
Technical Specifications
| Brand | Intel |
| Category | GPUs |
| SKU | M50CYP-GAUDI3-8OAM |
| Part Number | M50CYP-GAUDI3-8OAM |
| Condition | New |
| Manufacturer Part Number | M50CYP-GAUDI3-8OAM |
| Product Name | Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray (8x OAM) |
| Accelerator Architecture | Intel Gaudi 3 |
| Process Node | TSMC 5nm |
| Number of OAM Accelerators | 8x Gaudi 3 OAM |
| HBM Memory per OAM | 128 GB HBM2e |
| Total Accelerator HBM Memory | 1 TB (8 x 128 GB HBM2e) |
| Tensor Processor Cores per OAM | 64 |
| Supported Precisions | FP8, BF16, FP32 |
| Internal Accelerator Interconnect | All-to-all non-blocking RoCE v2-based fabric |
| Aggregate Internal Fabric Bandwidth | 2.4 Tbps |
| Network Interface (per OAM) | 24x 200GbE RDMA-capable ports (on-die) |
| Host Interface | PCIe Gen 5 |
| Server Chassis Base | Intel Server System M50CYP2SBSTD |
| Form Factor | 2U Rack |
| Software Stack | Intel SynapseAI SDK |
| Framework Compatibility | PyTorch, TensorFlow |
| Target Workloads | AI Training, AI Inference, HPC |
| OAM Standard Compliance | Open Accelerator Module (OAM) specification |
Frequently Asked Questions about Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray (8x OAM)
What server platforms accept the Intel Server System M50CYP2SBSTD with Gaudi 3 OAM Tray (8x OAM)?
Reference servers include Dell PowerEdge XE9680 / XE9712, HPE Cray XD670, Lenovo ThinkSystem SR685a / SR675 V3, Supermicro AS-A21GE / SYS-821GE, Gigabyte G593 / G894, ASUS ESC. Share your target platform in the RFQ and we will confirm chassis-to-GPU compatibility and recommended NIC pairing.
How long is the lead time on AI GPUs?
Highly model-dependent. L40S / RTX-class: typically 3-6 weeks. H100/H200/B200 in SXM form factor: 12-16 weeks for whole-platform allocations. We quote genuine-channel ETAs only — no grey-market promises.
Do you supply matched networking (Quantum InfiniBand / Spectrum-X)?
Yes — Omnixon stocks the full NVIDIA networking lineup (Quantum-2 / Quantum-X InfiniBand, Spectrum-X Ethernet, ConnectX NICs, BlueField DPUs) so we can quote a complete training-cluster BOM, not just the GPUs.
Can you help with NVIDIA AI Enterprise licensing?
Yes. We hold genuine-channels for NVIDIA AI Enterprise software subscriptions. Add it to your RFQ and we quote node-aligned licensing along with the hardware.