AI development is moving closer to the desktop. PNY NVIDIA DGX Spark Founders Edition combines NVIDIA Grace Blackwell architecture, unified memory and the NVIDIA AI software environment for local model development.
PNY supplies two regional configurations: DGXSPARK-FOUNEDIT-EU for the EU power-cable version and DGXSPARK-FOUNEDIT-UK for the UK version. Their core computing specifications are the same, but the regional part number matters for accurate procurement, inventory and deployment. This guide compares both items and explains how to order the correct configuration.
What Is NVIDIA DGX Spark?
NVIDIA DGX Spark is a compact personal AI computer created for developers, data scientists, researchers and students who need substantial AI performance without installing a rack server. At its heart is the NVIDIA GB10 Grace Blackwell Superchip. This platform combines a Blackwell-generation GPU with fifth-generation Tensor Cores and a 20-core Arm CPU containing 10 Cortex-X925 and 10 Cortex-A725 cores.
The system delivers up to 1 PFLOP of theoretical FP4 AI performance when using the sparsity feature. This makes it suitable for local model prototyping, inference, data preparation, experimentation and selected fine-tuning workflows. Instead of waiting for access to a centralized cluster, an AI professional can develop and test a workflow locally before moving it to a data center or cloud environment.
Why 128 GB of Unified Memory Matters
One of the most important DGX Spark features is its 128 GB of coherent LPDDR5x unified system memory. Conventional workstations normally divide memory between the CPU and GPU. Data may need to be copied between separate pools, creating limits when a model is larger than the available GPU memory. DGX Spark provides a shared memory architecture that allows the CPU and GPU to work with the same pool.
The 256-bit memory interface offers 273 GB/s bandwidth. According to the product datasheet, one system can support models containing up to 200 billion parameters. Two DGX Spark systems can be connected through NVIDIA ConnectX networking for inference with models containing up to 405 billion parameters. Actual performance and capacity depend on model architecture, precision, software configuration and workload.
Storage, Networking and Connectivity
Each PNY DGX Spark configuration includes a 4 TB NVMe M.2 drive with self-encryption. The local storage provides room for model files, containers, datasets, checkpoints and development tools. Networking includes a ConnectX-7 Smart NIC and one 10 GbE RJ-45 connection, supporting fast transfers and direct communication between compatible systems.
Additional connectivity includes Wi-Fi 7, Bluetooth 5.3 LE, four USB Type-C ports and HDMI 2.1a with multichannel audio. The system measures 150 x 150 x 50.5 mm and weighs approximately 1.2 kg, making it practical for offices, laboratories, classrooms and development spaces.
NVIDIA DGX OS and the AI Software Stack
Hardware is only part of the DGX Spark experience. The system runs NVIDIA DGX OS with Ubuntu Linux and is preconfigured for the NVIDIA AI software ecosystem. Developers can use familiar frameworks and tools such as CUDA, PyTorch, Jupyter and Ollama, along with access to NVIDIA NIM, NVIDIA Blueprints and containerized workflows.
DGXSPARK-FOUNEDIT-EU: EU Cable Configuration
DGXSPARK-FOUNEDIT-EU is the official PNY part number for the EU regional configuration. It is the correct item for customers whose deployment requires the supplied EU-compatible power cable. The computer, memory, storage, networking and software capabilities remain those of the PNY NVIDIA DGX Spark Founders Edition.
Organizations should place the complete part number on purchase orders and quotation requests. Using only a generic name such as “DGX Spark” can create uncertainty during sourcing, especially when stock for several regional packages is available. View the PNY NVIDIA DGX Spark EU cable product page for specifications and quotation assistance.
DGXSPARK-FOUNEDIT-UK: UK Cable Configuration
DGXSPARK-FOUNEDIT-UK is the corresponding PNY part number for the UK regional configuration. It should be selected when the delivered package must include the UK-compatible power cable. This distinction is particularly important for resellers, system integrators and organizations distributing equipment across several offices or countries.
The UK model has the same Grace Blackwell processing platform, 128 GB unified memory, 4 TB NVMe storage and professional networking as the EU version. The regional package and part number are the primary purchasing differences. View the PNY NVIDIA DGX Spark UK cable product page to request availability and project pricing.
EU and UK Models at a Glance
| Item | EU Configuration | UK Configuration |
|---|---|---|
| PNY part number | DGXSPARK-FOUNEDIT-EU | DGXSPARK-FOUNEDIT-UK |
| Regional cable | EU-compatible power cable | UK-compatible power cable |
| Architecture | NVIDIA Grace Blackwell | NVIDIA Grace Blackwell |
| Unified memory | 128 GB LPDDR5x | 128 GB LPDDR5x |
| Storage | 4 TB NVMe M.2 | 4 TB NVMe M.2 |
| Operating system | NVIDIA DGX OS | NVIDIA DGX OS |
Before ordering, verify the required cable standard, destination country and local electrical requirements. The regional code should match the quotation, purchase order, invoice and delivered package. Correct identification is especially valuable for large deployments because it reduces manual corrections and helps every location receive the intended configuration.
Who Should Consider PNY NVIDIA DGX Spark?
DGX Spark can serve a broad range of professional AI users. Data scientists can explore models locally without consuming shared cluster capacity for every experiment. Software developers can build AI-enabled applications in an environment aligned with NVIDIA infrastructure. Research teams can evaluate new methods, while universities can provide students with practical access to modern accelerated computing.
Procurement Checklist
- Confirm whether the destination requires the EU or UK cable configuration.
- Use the complete PNY part number on the request for quotation.
- Verify current stock, lead time and warranty terms.
- Plan network connectivity, storage capacity and workspace requirements.
- Review model size, precision and framework compatibility for the intended workload.
- Discuss multi-system networking requirements before deployment.
Why Source DGX Spark Through ICTECH Distribution?
ICTECH Distribution supports businesses, resellers, universities, research organizations and system integrators purchasing professional IT infrastructure. Customers can request project quotations for either PNY regional configuration and receive assistance identifying the correct part number for their deployment.
Conclusion
PNY NVIDIA DGX Spark brings Grace Blackwell AI computing, 128 GB unified memory, 4 TB of storage and the NVIDIA software ecosystem to a compact desktop platform. The EU and UK items deliver the same core AI capabilities, but buyers must select the correct regional power-cable configuration.
Choose DGXSPARK-FOUNEDIT-EU for the EU cable package and DGXSPARK-FOUNEDIT-UK for the UK cable package. Contact ICTECH Distribution with the exact part number to confirm availability, wholesale pricing and delivery requirements.
Frequently Asked Questions
Q1. What is the difference between DGXSPARK-FOUNEDIT-EU and DGXSPARK-FOUNEDIT-UK?
The main difference is the regional power-cable configuration and its associated PNY part number. Their principal computing specifications are the same.
Q2. How much unified memory does PNY NVIDIA DGX Spark include?
Both configurations include 128 GB of coherent LPDDR5x unified system memory with a 256-bit interface and 273 GB/s bandwidth.
Q3. Can DGX Spark run large AI models locally?
Yes. The datasheet states support for models containing up to 200 billion parameters on one system, depending on the model, precision and software configuration.
Q4. Can two DGX Spark systems be connected?
Yes. NVIDIA ConnectX networking can link two systems for compatible workloads, including inference with models containing up to 405 billion parameters as described in the datasheet.
Q5. Where can I request a quotation?
ICTECH Distribution provides quotation and project support for both PNY DGX Spark regional part numbers in the UAE and international markets.
