HP ZGX Nano G1n NVIDIA GB10 AI Station - 128GB / 4TB
PRODUCT DESCRIPTION
HP ZGX Nano G1n AI Station - 128GB / 4TB (DD8J0PA) - 3 Years Onsite Local Warranty
NVIDIA GB10 Grace Blackwell Superchip on Your Desk
The HP ZGX Nano G1n AI Station is a personal AI supercomputer built around the NVIDIA GB10 Grace Blackwell Superchip: a 20-core Arm CPU (10 Cortex-X925 + 10 Cortex-A725) fused with a Blackwell-architecture GPU delivering up to 1000 TOPS of FP4 AI performance. It lets AI developers, data scientists and enterprise teams in Singapore prototype, fine-tune and run large language models locally instead of renting cloud GPUs.
128 GB Unified Memory, 4 TB Self-Encrypting NVMe
128 GB of LPDDR5x-8533 coherent unified system memory (273 GB/s) is shared between CPU and GPU, so models of up to 200 billion parameters fit on a single unit. This DD8J0PA configuration ships with a 4 TB PCIe Gen4 NVMe TLC M.2 SSD with OPAL self-encryption (SED), giving ample room for datasets, checkpoints and model libraries while keeping data at rest protected.
Scale to Two Units with ConnectX-7 200GbE
Dual QSFP 200GbE ports driven by the NVIDIA ConnectX-7 controller let two ZGX Nano units be linked with a QSFP DAC cable to work on models of up to 405 billion parameters. A 10GbE RJ-45 port, Wi-Fi 7 and Bluetooth 5.4, three 20Gbps USB-C ports and HDMI 2.1a complete the connectivity on a 150 x 150 x 51 mm chassis that weighs from 1.25 kg.
NVIDIA DGX OS and the HP ZGX Toolkit
The unit comes preloaded with NVIDIA DGX OS 7 (Ubuntu 24.04 based) with CUDA 13, the NVIDIA RTX driver and DGX Dashboard, and HP's ZGX Toolkit adds MLflow, Ollama and Visual Studio Code integration for instant model discovery, local training with experiment tracking, and export to deploy anywhere.
Buying for an AI lab, research team or multi-unit cluster? Email sales@sourceit.com.sg for volume pricing, QSFP cabling and delivery lead time.
HP ZGX Nano G1n AI Station Datasheet — Technical Specifications
General
| Product Name | HP ZGX Nano G1n AI Station |
| Part Number | DD8J0PA |
| Product Family | HP Z AI Stations (ZGX series), NVIDIA GB10 Grace Blackwell platform |
| Form Factor | Mini desktop AI station |
| Local Warranty (Singapore) | 3-year onsite local warranty, serviced in Singapore. Supplied by SourceIT with a GST invoice |
Processor and AI Performance
| Superchip | NVIDIA GB10 Grace Blackwell Superchip |
| CPU | 20-core Arm: 10 Cortex-X925 performance cores + 10 Cortex-A725 efficiency cores |
| GPU | Integrated NVIDIA Blackwell architecture GPU with 5th generation Tensor Cores and 4th generation RT Cores |
| AI Performance | Up to 1000 TOPS (FP4) |
| Model Capacity | Up to 200-billion-parameter models on one unit; up to 405-billion-parameter models with two units linked over ConnectX-7 |
Memory and Storage
| Memory | 128 GB LPDDR5x-8533 coherent unified system memory, 16 channels, soldered down (not upgradeable) |
| Memory Bandwidth | 273 GB/s |
| Storage | 4 TB PCIe Gen4 x4 NVMe TLC M.2 2242 SSD, SED OPAL self-encrypting |
| Storage Performance | Up to 7000 MB/s sequential read, 6000 MB/s sequential write |
Networking and Ports
| High-Speed Networking | NVIDIA ConnectX-7 200GbE controller, 2 x QSFP ports (200 Gbps signalling) for two-unit clustering |
| Ethernet | 1 x RJ-45 10GbE |
| Wireless | MediaTek MT7925 Wi-Fi 7 (802.11be, 2x2) and Bluetooth 5.4 |
| Rear Ports | 1 x USB Type-C power connector; 3 x USB 3.2 Type-C 20Gbps (DisplayPort 1.4a Alt Mode); 1 x HDMI 2.1a; 1 x RJ-45 10GbE; 2 x QSFP 200GbE |
| Audio | Audio over HDMI |
Software and Security
| Operating System | NVIDIA DGX OS 7 (Ubuntu 24.04 based). Microsoft Windows is not supported |
| Preinstalled Software | CUDA 13, NVIDIA RTX driver, DGX Dashboard, HP ZGX Toolkit (MLflow, Ollama, Visual Studio Code integration) |
| Security | Integrated TPM 2.0 (convertible to FIPS 140-2 certified mode); SED OPAL storage encryption |
Power and Physical
| Power Adapter | 240 W external USB Type-C power adapter, 89% efficiency, active PFC |
| Dimensions (W x D x H) | 15 x 15 x 5.1 cm (without feet); 15 x 15 x 5.45 cm (with feet) |
| Weight | Starting at 1.25 kg |
| Colour | Black |
| Ecolabels | EPEAT Gold (IEEE 1680.1); RoHS compliant |
Compatibility & Software Support
Operating System and Frameworks
NVIDIA DGX OS 7 is a Linux (Ubuntu 24.04) environment preloaded with the NVIDIA AI software stack, including CUDA 13, the NVIDIA RTX driver and DGX Dashboard, with Ollama and MLflow added by the HP ZGX Toolkit. It does not run Windows; connect from a Windows 11 or Ubuntu 24.04 client over the network, or attach a monitor and keyboard directly.
HP ZGX Toolkit
The HP ZGX Toolkit discovers the ZGX Nano on your network, syncs projects, tracks experiments in MLflow, serves models locally with Ollama and exports them for deployment on servers or the cloud, all from Visual Studio Code.
Two-Unit Clustering
Link two ZGX Nano units back-to-back through the QSFP 200GbE ports with a compatible QSFP112 DAC cable (sold separately, see HP ZGX 0.4m QSFP112 Ethernet DAC Cable) to pool 256 GB of unified memory for models of up to 405 billion parameters.
Displays and Peripherals
HDMI 2.1a drives a monitor directly, and each of the three USB-C 20Gbps ports supports DisplayPort 1.4a Alt Mode for additional screens, USB-C docks and storage.
Typical Use
Local LLM fine-tuning and inference, generative AI prototyping, computer vision and robotics development, data science, edge AI and sovereign or air-gapped AI deployments where data must stay on-premises.
Package Contents & Ordering Information
In the Box
HP ZGX Nano G1n AI Station (DD8J0PA), 240 W USB Type-C power adapter, power cord and documentation.
NVIDIA GB10 Personal AI Supercomputers at SourceIT
| Model | Configuration |
| HP ZGX Nano G1n AI Station (DD8J0PA) (this listing) | NVIDIA GB10, 128GB unified memory, 4TB SED NVMe, DGX OS, 3-year onsite warranty |
| NVIDIA DGX Spark (940-54242-0007-000) | NVIDIA GB10, 128GB unified memory, 4TB NVMe — NVIDIA Founders Edition |
| Dell Pro Max with GB10 (4TB) | NVIDIA GB10, 128GB unified memory, 4TB NVMe |
| Dell Pro Max with GB10 (2TB) | NVIDIA GB10, 128GB unified memory, 2TB NVMe |
| GIGABYTE AI TOP ATOM (ATAGB10-9000) | NVIDIA GB10, 128GB unified memory, 4TB PCIe 5.0 NVMe |
| Acer Altos BrainSphere GB10 F1 (DT.L1BSM.001) | NVIDIA GB10, 128GB unified memory, 4TB NVMe |
| MSI EdgeXpert MS-C931 (9S6-C9311-61S) | NVIDIA GB10, 128GB unified memory, 4TB NVMe |
| HP ZGX Fury G1n (NVIDIA GB300) | Step-up deskside NVIDIA DGX Station class system on the GB300 Grace Blackwell Ultra Superchip |
Accessories
HP ZGX 0.4m QSFP112 Ethernet DAC Cable (C20MQAA) for linking two ZGX Nano units.
Datasheet & Downloads
| HP product page | HP ZGX Nano AI Station on hp.com |
| HP QuickSpecs | HP ZGX Nano G1n AI Station QuickSpecs |
| HP datasheet | HP ZGX Nano G1n AI Station datasheet (PDF) |
| HP specifications | HP ZGX Nano G1n AI Station — full specifications |
Need a PDF sent directly? Email sales@sourceit.com.sg quoting DD8J0PA.
Frequently Asked Questions — HP ZGX Nano G1n AI Station
What is the configuration of DD8J0PA?
NVIDIA GB10 Grace Blackwell Superchip (20-core Arm CPU + Blackwell GPU), 128 GB LPDDR5x unified memory, 4 TB PCIe Gen4 NVMe SED SSD, ConnectX-7 200GbE, 10GbE, Wi-Fi 7, NVIDIA DGX OS 7.
Is the HP ZGX Nano the same as NVIDIA DGX Spark?
Both are built on the NVIDIA GB10 Grace Blackwell platform with 128 GB unified memory and run NVIDIA DGX OS. The HP ZGX Nano G1n adds the HP ZGX Toolkit, HP chassis design and HP warranty and support in Singapore.
How large a model can it run?
Up to 200 billion parameters on a single unit, or up to 405 billion parameters with two units linked over the QSFP 200GbE ports.
Can the memory be upgraded?
No. The 128 GB LPDDR5x unified memory is soldered to the board. Storage is an M.2 2242 SSD.
Does it run Windows?
No. It ships with NVIDIA DGX OS 7 (Ubuntu 24.04 based) and does not support Microsoft Windows. You can access it from a Windows 11 or Ubuntu client over the network.
What warranty do I get?
3-year onsite local warranty, serviced in Singapore. Supplied by SourceIT with a GST invoice.
