HP ZGX Fury G1n GB300 AI Workstation with NVIDIA Blackwell Ultra
PRODUCT DESCRIPTION
HP ZGX Fury G1n GB300 AI Workstation with NVIDIA Blackwell Ultra, 748GB Coherent Memory, 72-Core Grace CPU & Dual 400G ConnectX-8 Networking - Local Warranty
HP ZGX Fury G1n — Next-Generation Enterprise AI Workstation Powered by NVIDIA GB300
The HP ZGX Fury G1n is a next-generation AI supercomputer built around the NVIDIA Grace Blackwell Ultra (GB300) Superchip, delivering up to 20 petaFLOPS of FP4 AI compute for training, fine-tuning and deploying LLMs, AI agents, generative AI and scientific simulations — entirely within your own infrastructure, with complete data sovereignty.
748 GB Coherent Unified Memory for Large-Scale AI Models
252 GB of HBM3e GPU memory at 7.1 TB/s and 496 GB of LPDDR5X CPU memory unite over NVLink-C2C (900 GB/s) into a single 748 GB coherent memory space — supporting AI models up to 1 trillion parameters and accelerating RAG, foundation-model fine-tuning, digital twins, engineering simulation and cybersecurity analytics.
72-Core Grace CPU with Dual 400G ConnectX-8 Networking
The 72-core NVIDIA Grace Arm Neoverse V2 CPU works coherently with the Blackwell Ultra GPU, while dual NVIDIA ConnectX-8 400 GbE QSFP ports (up to 800 Gb/s) deliver ultra-low-latency clustering for distributed AI training and private AI clusters — datacentre-class capability without a specialised liquid-cooled rack.
Three Factory Configurations — Choose Your Graphics & Storage
The ZGX Fury G1n is available in three factory-built SKUs. All three share the identical GB300 Superchip, 748 GB coherent memory, 72-core Grace CPU and dual 400G ConnectX-8 networking, and all ship with 3/3/3 warranty — they differ only in the professional graphics card and the NVMe storage configuration. Select your configuration from the dropdown above.
Available Configurations — 3 SKUs (Singapore List Price)
| SKU 1 — E6ZS0PT#AB4 SGD $168,838 |
Graphics: NVIDIA RTX PRO 2000 Blackwell Storage: 4 × 2TB NVMe SSD (8TB total) Warranty: 3/3/3 |
| SKU 2 — E6ZS5PT#AB4 SGD $172,647 |
Graphics: NVIDIA RTX PRO 4000 Blackwell Storage: 2 × 2TB + 2 × 4TB NVMe SSD (12TB total) Warranty: 3/3/3 |
| SKU 3 — Part No. TBA SGD $158,683 |
Graphics: No discrete graphics card Storage: 4 × 2TB NVMe SED — Self-Encrypting Drives (8TB total) Warranty: 3/3/3 |
Which configuration? Choose SKU 1 for a balanced build with professional display graphics for local visualisation and CAD/DCC work. Choose SKU 2 when you need more on-board dataset capacity and a stronger visualisation GPU. Choose SKU 3 for headless, rack-adjacent or remote-accessed deployments, or where self-encrypting drives are mandated by security or compliance policy.
Prices shown are Singapore list price. For volume, project or education pricing, GST invoicing and lead times, contact SourceIT at sales@sourceit.com.sg.
Available in Singapore Through SourceIT
Enterprise procurement with GST invoice and local warranty support. For desktop-scale AI, see the NVIDIA DGX Spark; for the alternative GB300 station, see the MSI XpertStation WS300.
HP ZGX Fury G1n Datasheet — Technical Specifications
General
| Model | HP ZGX Fury G1n (NVIDIA GB300 Platform) |
| Product Type | Enterprise Desktop AI Supercomputer / AI Workstation |
| Chassis | Black, Professional Enterprise Workstation Design |
| Warranty | 3/3/3 HP Enterprise Workstation Warranty — Local Support via SourceIT |
Processor — NVIDIA GB300 Grace Blackwell Ultra Superchip
| GPU | 1 × NVIDIA Blackwell Ultra |
| CPU | 1 × NVIDIA Grace — 72-Core Arm Neoverse V2 |
| Interconnect | NVLink-C2C, 900 GB/s |
| AI Performance | Up to 20 PFLOPS FP4 Tensor Core |
| Model Support | AI Models Up to 1 Trillion Parameters |
| GPU Partitioning | Multi-Instance GPU (MIG) — Up to 7 Instances |
Memory
| GPU Memory | 252 GB HBM3e, 7.1 TB/s |
| CPU Memory | 496 GB LPDDR5X, 396 GB/s |
| Total Coherent Memory | 748 GB Unified |
Professional Graphics (Configuration Dependent)
| SKU 1 — E6ZS0PT#AB4 | NVIDIA RTX PRO 2000 Blackwell |
| SKU 2 — E6ZS5PT#AB4 | NVIDIA RTX PRO 4000 Blackwell |
| SKU 3 | No Discrete Graphics Card |
Storage (Configuration Dependent)
| SKU 1 — E6ZS0PT#AB4 | 4 × 2TB NVMe SSD |
| SKU 2 — E6ZS5PT#AB4 | 2 × 2TB + 2 × 4TB NVMe SSD |
| SKU 3 | 4 × 2TB NVMe SED (Self-Encrypting Drives) |
| Drive Platform | M.2 Gen 5 Platform Slots |
Networking
| SuperNIC | NVIDIA ConnectX-8 — Up to 800 Gb/s Ethernet |
| High-Speed Ports | 2 × QSFP (400 Gb/s Each) |
| Standard Ethernet | 10 GbE + 1 GbE Management |
Software & Power
| Operating System | Ubuntu with NVIDIA AI Developer Tools (DGX Software Stack) |
| AI Stack | CUDA, NVIDIA AI Enterprise, TensorRT, NVIDIA NIM, PyTorch/TensorFlow |
| Power | HP Enterprise High-Efficiency Power Supply (~1,600 W Platform) |
| Cooling | Datacentre-Class Component Cooling — No Liquid-Cooled Rack Required |
| Management | Secure Remote Management, Continuous 24/7 Operation |
Ordering Information — AI Supercomputers at SourceIT
| HP ZGX Fury G1n GB300 AI Workstation | This Product — 3 Configurations Available, from SGD $158,683 |
| MSI XpertStation WS300 (DGX Station GB300) | In Stock |
| NVIDIA DGX Spark 4TB (Desktop, 1 PFLOP) | 940-54242-0007-000 — In Stock |
| ASUS Ascent GX10 (GB10 Platform, from 1TB) | In Stock |
Datasheet & Downloads
Frequently Asked Questions — HP ZGX Fury G1n
What is the difference between the three ZGX Fury G1n SKUs?
Only the graphics card and the storage. SKU 1 (E6ZS0PT#AB4) pairs an NVIDIA RTX PRO 2000 Blackwell with 4 × 2TB NVMe. SKU 2 (E6ZS5PT#AB4) steps up to an RTX PRO 4000 Blackwell with 2 × 2TB + 2 × 4TB NVMe. SKU 3 ships with no discrete graphics card and 4 × 2TB self-encrypting drives for security-sensitive or headless deployments. The GB300 Superchip, 748 GB coherent memory, 72-core Grace CPU, ConnectX-8 networking and 3/3/3 warranty are identical across all three.
Why would I choose the No-Graphics SKU?
If the system is accessed remotely or sits in a shared compute room, a display GPU adds cost without value — and the 4 × 2TB SED drives give you hardware-level encryption at rest, which is often a requirement for regulated, government and financial-sector workloads in Singapore.
How does the ZGX Fury G1n compare to the NVIDIA DGX Spark?
Different class entirely: the ZGX Fury delivers up to 20 PFLOPS FP4 and 748 GB coherent memory for models up to 1 trillion parameters, versus the DGX Spark's 1 PFLOP and 128 GB (200B parameters). Choose the Spark for individual developers; the ZGX Fury for teams and production-scale AI.
HP ZGX Fury G1n vs MSI XpertStation WS300 — which should I choose?
Both are built on the NVIDIA GB300 platform with identical core compute. They differ in chassis engineering, storage configuration, management tooling and vendor warranty terms — the WS300 ships with 1.92 TB RAID 1 as standard, while the ZGX Fury is offered in the three factory configurations above. Talk to us about which fits your deployment.
Can multiple team members share one system?
Yes — Multi-Instance GPU (MIG) partitions the Blackwell Ultra GPU into up to 7 isolated instances for concurrent users and workloads.
Does it need a server room?
No liquid-cooled rack is required — datacentre-class cooling is built into the workstation chassis for continuous 24/7 operation. Plan a dedicated power circuit as for any high-end workstation.
How do I purchase in Singapore?
SourceIT supplies the HP ZGX Fury G1n with GST invoice, local warranty support and enterprise procurement assistance — contact us for configuration options, lead times and volume pricing.
