As a U.S.-based, vertically integrated global manufacturer with domestic SMT capabilities, Hyve Solutions engineers and delivers custom AI compute, storage, and networking solutions for hyperscalers. We rapidly design, build, and deploy tailored rack-ready solutions optimized for any environment worldwide. Our liquid cooling expertise includes direct-to-chip and rack-level integration for high-density AI workloads, backed by comprehensive global service and support for the world’s largest data center customers.
Hyve’s vertically integrated AI Infrastructure portfolio includes both NVIDIA® and AMD based liquid and air cooled solutions, customized network switch platforms and features our storage solutions e.g., Hercules and Hyve JBOF.
Hyve Hercules
Ultra-high Density Mass Storage Server
5U form factor with 112 x 3.5” SAS/SATA3 HDDs
RVI optimized design for stable performance
Single Intel® Xeon® 6, 6700-series CPU
16x RDIMM 6400MT/s DDR5, 64GB, 2DPC
2x 12V 3600W/ 2x 12V CRPS 3200W Titanium Power Supply Unit
Scalable, Ultra Low Latency Hyve JBOF
Disaggregated, High-Performance Mass Storage Optimized for AI Workloads
- 2U DC-MHS 19” form factor in a rack-ready design
- 24x U.2/U.3 NVMe SSD drives
- PCle direct connect or NVMe oF (DPU) options
- Astera Labs® / Broadcom® PCIe Gen6 switch
- Hot-swappable
- TPM 2.0 module for security
- 1 x NVIDIA BlueField® -3 DPU
Hyve Mt Jade Custom AI Network Switch
Ultra-high Performance & Low-latency Network Switch for
Next-Gen Data Centers
- 64 x 800 Gb/s OSFP Ports
- Switching Capacity 51.2 Tbps
- 2RU fixed form factor, air-cooled
- Designed for AI/ML clusters, cloud computing, large-scale enterprise
- Supports ultra-high performance, low latency and high reliability
- Advanced adaptive routing, dynamic load balancing, and support end-to-end congestion control
- Support for Clos and non-Clos topologies such as Dragonfly
- Open Network Install Environment (ONIE) software installer
- Compatible with Open Source NOS including SONiC
Inference Platform with NVIDIA® RTX PRO™
8x NVIDIA RTX PRO™ 6000/4500 Blackwell
4U MGX form factor with Astera Lab® Gen6 switch or ConnectX-8 SuperNIC switch (optional)
1 x NVIDIA BlueField®-3 DPU
Hyve-designed 2S MB (Intel® and AMD® compatible)
32x RDIMM 6400MT/s DDR5, 64GB, 2DPC
1x E1.S PCIe Gen4 x2, 9.5mm Boot Drive
16x E1.S PCIe Gen5 x4, 9.5/15mm Hot-swappable Data Storage Bays
3+1x CRPS/M-CRPS, up to 3.6KW Titanium Power Supply Unit
Up to 5x HHHL cards supported
MIL-SPEC Certified Ruggedized Edge
Platform Ruggedized Edge Compute Platform for Extreme Conditions
- Ideal for processing Edge AI workloads, latency sensitive loT applications
- Mobile Edge Computing and 5G at the edge
- 2U, 19” MlL-SPEC optimized chassis with ruggedized enclosure
- Compliant with MIL-STD-810 for temperature, humidity, shock & vibration, and MlL-STD-461 for EMI/EMC
- Customizable and optimized with 56 CPU cores and 1024GB of DDR5
- 3 NVIDIA® L4 Tensor Core GPUs and 1S Intel® Xeon® 6 CPU
Hyve Orion™ Featuring NVIDIA® HGX™ Platform
GPU Powerhouse Ideal for AI Training, Inference, Data Analytics & Edge Computing
- JBOG and Monolithic versions
- Compatible with NVIDIA B200 and B300
- Air and Liquid Cooling Optimized
- Intel® / AMD Head Node supported
- Broadcom® PCIe Gen5 Solution
- 8x U.2 Gen4 x4 hot-swappable, 8x non-swappable within system (Optional)
TO LEARN MORE, Contact Sales@HyveSolutions.com
All trademarks, logos and brand names are the property of their respective owners. All company, product and service names used are for identification purposes only.