HP ZGX Nano G1n NVIDIA GB10 AI Station - 128GB / 4TB
PRODUCT DESCRIPTION
HP ZGX Nano G1n AI Station NVIDIA GB10 Grace Blackwell Personal AI Supercomputer, 128GB Unified Memory, 4TB Self-Encrypting NVMe SSD, ConnectX-7 200GbE, Wi-Fi 7, NVIDIA DGX OS, HP ZGX Toolkit (DD8J0PA) - 3 Years Onsite Local Warranty
HP ZGX Nano G1n — NVIDIA DGX Spark AI Performance with HP Z Workstation Tooling
The HP ZGX Nano G1n AI Station is HP's build of the NVIDIA DGX Spark platform: an NVIDIA GB10 Grace Blackwell Superchip pairing a 20-core Arm CPU (10 Cortex-X925 + 10 Cortex-A725) with a Blackwell GPU and 128 GB of unified LPDDR5x memory, delivering up to 1,000 TOPS of FP4 AI compute in a 150 × 150 × 51 mm desktop that weighs 1.25 kg. It develops, fine-tunes and runs AI models of up to 200 billion parameters locally, and two units link over ConnectX-7 to reach 405 billion.
What sets the ZGX Nano apart is HP's software: the free HP ZGX Toolkit VS Code extension sets up Python, Podman, Ollama, JupyterLab, PyTorch, CUDA and MLflow environments from a laptop, and HP Remote System Controller gives BIOS-level remote access. SourceIT supplies this 4 TB configuration (DD8J0PA) in Singapore with 3-year onsite warranty, a full GST invoice and enterprise procurement assistance.
128 GB Unified Memory at 273 GB/s
The GB10's 128 GB of LPDDR5x-8533 is a single coherent 16-channel memory pool shared by CPU and GPU at 273 GB/s, so large language models, vision-language models and RAG pipelines load without the VRAM ceiling of a discrete-GPU workstation. Fifth-generation Tensor Cores with FP4 precision and fourth-generation RT Cores accelerate inference, fine-tuning and rendering.
4 TB Self-Encrypting NVMe, ConnectX-7 200 GbE, Wi-Fi 7
A 4 TB PCIe NVMe TLC M.2 SSD with SED Opal self-encryption holds models, datasets and containers. Networking pairs an NVIDIA ConnectX-7 200 GbE controller with two QSFP ports for clustering a second unit, with a Realtek 10 GbE RJ45 port and a MediaTek Wi-Fi 7 / Bluetooth 5.4 module. Three USB Type-C 20 Gbps ports with DisplayPort 1.4a Alt Mode and an HDMI 2.1a port drive the displays.
HP ZGX Toolkit, DGX OS and TPM 2.0 Security
The ZGX Nano ships with NVIDIA DGX OS 7 (Ubuntu 24.04) plus NVIDIA Connect, CUDA 13, the DGX Dashboard and NVIDIA AI Workbench preinstalled. HP's ZGX Toolkit extension for Visual Studio Code provisions the AI development stack from a Windows 11 or Ubuntu client, while an integrated TPM 2.0 (convertible to FIPS 140-2 certified mode) and UEFI 2.8 firmware secure the platform for enterprise deployment.
240 W USB-C Power, 1.25 kg, 3-Year Onsite Warranty
A 240 W external USB-C power adapter (89 % efficiency, active PFC) feeds the system; the chassis measures 15 × 15 × 5.1 cm without feet and weighs from 1.25 kg. HP rates it for 5–30 °C operation. SourceIT supplies the DD8J0PA configuration with a 3-year onsite local warranty.
HP ZGX Nano G1n AI Station Datasheet — Technical Specifications
General
| Product Name | HP ZGX Nano G1n AI Station |
| Part Number (Singapore) | DD8J0PA (4 TB configuration) |
| Category | Personal AI supercomputer / AI developer desktop (NVIDIA DGX Spark platform) |
| Form Factor | Mini desktop |
| Local Warranty (Singapore) | 3 Years Onsite Local Warranty; procurement and GST invoice by SourceIT |
Processor & GPU
| Compute Platform | NVIDIA DGX Spark architecture — NVIDIA GB10 Grace Blackwell Superchip |
| CPU | 20-core Arm — 10 × Cortex-X925 performance cores + 10 × Cortex-A725 efficiency cores |
| GPU | NVIDIA Blackwell architecture GPU — fifth-generation Tensor Cores, fourth-generation RT Cores, 1 × NVENC, 1 × NVDEC |
| AI Performance | Up to 1 PFLOP (1,000 TOPS) FP4 with sparsity |
| Model Support | AI models up to 200 billion parameters on one unit; up to 405 billion parameters with two units linked over ConnectX-7 |
Memory
| Unified Memory | 128 GB LPDDR5x coherent unified system memory, 256-bit |
| Memory Bandwidth | 273 GB/s |
| Memory Speed | LPDDR5x-8533 MT/s, 16 channels, soldered — not upgradeable |
Storage
| SSD | 4 TB PCIe NVMe TLC M.2 2242 SED Opal self-encrypting SSD |
| Slot | 1 × M.2 PCIe Gen5 x4 |
Networking
| High-Speed Networking | NVIDIA ConnectX-7 200 GbE — 2 × QSFP 200 Gbps |
| Ethernet | 1 × RJ45 10 GbE (Realtek RTL8127-CG) |
| Wireless | MediaTek Wi-Fi 7 MT7925 (2×2) + Bluetooth 5.4 |
I/O Ports
| USB | 3 × USB Type-C 20 Gbps with DisplayPort 1.4a Alt Mode; 1 × USB Type-C power connector |
| Display / Audio | 1 × HDMI 2.1a with HDMI audio output |
Security
| TPM | Integrated TPM 2.0, convertible to FIPS 140-2 certified mode |
| Firmware | UEFI 2.8 |
Operating System & Software
| Operating System | NVIDIA DGX OS 7 (Ubuntu 24.04) — Windows not supported |
| Preinstalled | NVIDIA Connect, CUDA 13, NVIDIA RTX Driver 580, DGX Dashboard, NVIDIA AI Workbench |
| HP Software | HP ZGX Toolkit (VS Code extension); HP Remote System Controller |
Power, Physical & Environmental
| Power Adapter | 240 W external USB Type-C, 89 % efficiency, active PFC (HP ZGX 240 W USB-C AC adapter CM5A0AA) |
| Dimensions (W × D × H) | 15 × 15 × 5.1 cm without feet; 5.45 cm with feet |
| Weight | Starting at 1.25 kg |
| Operating Temperature | 5 – 30 °C |
Compatibility & Software Support
NVIDIA DGX OS and the NVIDIA AI Software Stack
Ships with NVIDIA DGX OS — an Ubuntu Linux build with the NVIDIA AI software stack preinstalled: CUDA, cuDNN, TensorRT and TensorRT-LLM, NVIDIA NeMo, NIM microservices, Triton Inference Server, RAPIDS and NGC containers, alongside PyTorch, TensorFlow, JAX, vLLM, Ollama, Hugging Face Transformers and ONNX Runtime. NVIDIA AI Workbench and the DGX Dashboard simplify environment setup, and Docker or Podman containers keep projects isolated. Because it is the same software foundation as the DGX Station and DGX data-centre systems, models developed here move to larger NVIDIA infrastructure without code changes.
Arm64 Architecture — Plan Your Software Accordingly
The GB10's Grace CPU is Arm-based, so the operating system, drivers and container images are arm64 (aarch64) builds. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64, but proprietary x86-only applications will not run natively — check any commercial software you rely on before ordering. Windows is not supported on this platform.
Clustering Two Units
The two QSFP 200 GbE ports on the NVIDIA ConnectX-7 SuperNIC let two GB10 systems be linked with a QSFP112 direct-attach cable (sold separately) to work with models up to 405 billion parameters. The 10 GbE RJ45 port covers everyday site networking, and Wi-Fi 7 keeps the system connected on a desk without cabling.
HP ZGX Toolkit and HP Z Ecosystem
The HP ZGX Toolkit is a free Visual Studio Code extension that connects to the ZGX Nano from a Windows 11 or Ubuntu 24.04 client and installs Python, Podman, Ollama, Miniforge, JupyterLab, PyTorch, CUDA, MLflow, Streamlit, Gradio and OpenWebUI in guided steps. HP Remote System Controller provides BIOS-level remote access. When a team outgrows 128 GB, the HP ZGX Fury AI Station G1n (NVIDIA DGX Station GB300) offers 748 GB and 20 PFLOPS with the same HP Z tooling and support model.
Package Contents & Ordering Information
What Ships With the System
- HP ZGX Nano G1n AI Station (DD8J0PA) with 4 TB self-encrypting NVMe SSD
- HP ZGX 240 W USB-C AC adapter and power cord
- NVIDIA DGX OS 7 preinstalled
- 3-year onsite local warranty
Keyboard and mouse are not included. The HP ZGX 0.4 m QSFP112 Ethernet DAC cable (C20MQAA) for linking two units is sold separately.
NVIDIA DGX Spark / GB10 Systems at SourceIT Singapore
The HP ZGX Nano G1n 4 TB is in stock at SourceIT Singapore with GST invoice and 3-year onsite warranty.
- NVIDIA DGX Spark 4TB — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- ASUS Ascent GX10 — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- Lenovo ThinkStation PGX — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- Dell Pro Max with GB10 — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- GIGABYTE AI TOP ATOM — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- MSI EdgeXpert MS-C931 — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- Acer Altos BrainSphere GB10 F1 — NVIDIA GB10 Grace Blackwell platform, same 128 GB unified memory and 1 PFLOP FP4 core compute
- MSI XpertStation WS300 (DGX Station GB300) — step up to 748 GB coherent memory and 20 PFLOPS for trillion-parameter models
Browse the full NVIDIA DGX Spark & DGX Station collection. For volume, project, research-grant or education pricing, GST invoicing and current lead times, email sales@sourceit.com.sg.
Datasheet & Downloads
Official HP documentation for the ZGX Nano G1n AI Station:
Frequently Asked Questions — HP ZGX Nano G1n AI Station
What is the HP ZGX Nano G1n?
HP's build of the NVIDIA DGX Spark platform: a 15 × 15 cm desktop with the NVIDIA GB10 Grace Blackwell Superchip, 128 GB unified memory and up to 1,000 TOPS FP4 for developing and running AI models of up to 200 billion parameters locally, with HP's ZGX Toolkit for environment setup. This configuration (DD8J0PA) has a 4 TB self-encrypting NVMe SSD.
What is the HP ZGX Toolkit?
A free Visual Studio Code extension from HP that connects to the ZGX Nano from your laptop and installs the AI development stack — Python, Podman, Ollama, JupyterLab, PyTorch, CUDA, MLflow, Streamlit, Gradio and OpenWebUI — in guided steps.
How does it compare to the NVIDIA DGX Station GB300 systems?
Different class: this is a desktop-scale GB10 system with 128 GB unified memory and 1 PFLOP FP4 for models up to about 200 billion parameters, aimed at individual developers, researchers and small teams. The DGX Station GB300 systems (MSI XpertStation WS300, HP ZGX Fury G1n, ASUS ET900N G3, GIGABYTE W775, Supermicro Super AI Station) carry 748 GB and up to 20 PFLOPS for trillion-parameter models and team-scale production AI.
Can I link two units together?
Yes. Two units connect over their ConnectX-7 200 GbE QSFP ports with a QSFP112 direct-attach cable (sold separately), scaling to models of up to 405 billion parameters.
What operating system does it run?
NVIDIA DGX OS, an Ubuntu Linux build with the NVIDIA AI software stack preinstalled. Windows is not supported.
Does it run Windows software or x86 applications?
No. The Grace CPU is Arm64, so only arm64 Linux software and containers run natively. Mainstream AI frameworks and NVIDIA NGC containers are available for arm64.
How does the ZGX Nano compare to the HP ZGX Fury G1n?
The Nano is the desktop-scale GB10 system (128 GB, 1,000 TOPS) for individual developers; the ZGX Fury G1n is the NVIDIA DGX Station GB300 system (748 GB, 20 PFLOPS) for teams and trillion-parameter models.
What warranty does it come with?
A 3-year onsite local warranty in Singapore, supplied through SourceIT.
How do I purchase the HP ZGX Nano G1n in Singapore?
SourceIT is an authorised IT reseller in Singapore and supplies the ZGX Nano with GST invoice, onsite warranty and enterprise procurement assistance — contact sales@sourceit.com.sg for volume pricing and lead times.
