News

Personal AI Blackwell Solutions - NVIDIA DGX Spark & DGX Station

April 4, 2025
4 min read
EXX-Blog-Personal-AI-Blackwell-solutions-NVIDIA-DGX-Spark-Station-1.jpg

NVIDIA Spark - AI the Size of a Mac

Imagine a Grace CPU and a Blackwell GPU condensed into a small form factor Mac Mini size workstation. The DGX Spark is the world’s smallest AI supercomputer for users to push the boundaries of generative and physical AI with massive performance and capabilities in a portable form factor.

NVIDIA DGX Spark

The DGX Spark features an NVIDIA GB10 Grace Blackwell Superchip capable of over 1,000 TOPS of AI compute with cut-down Grace and Blackwell processors. It also features a ConnectX-7 Smart NIC for clustering two DGX Sparks together for double the performance. The DGX Spark can effectively finetune and iterate workflows that can be delivered straight to DGX Cloud or to existing DGX infrastructure. NVIDIA Spark will be available through NVIDIA for pre-order, and board partners will develop their chassis. Exxact may support DGX Spark in the future.

NVIDIA Grace CPU

20 ARM Cores (10 Cortex-X925 + 10 Cortex A725)

CPU Memory

128GB Unified LPDDR5X at 273 GB/s

NVIDIA GPU

NVIDIA Blackwell Ultra

Networking

1x 10Gb Ethernet

Networking NIC

NVIDIA ConnectX-7 Smart NIC at 200GbE

Net. Bandwidth

Up to 800Gb/s

Storage

Up to 4TB NVMe SSD

Connectivity

WiFi 7 & Bluetooth 5.3

Output

1x HDMI 2.1a

Power Consumption

170W

NVIDIA DGX Station B300 - Grace Blackwell Workstation

Imagine a 1U Grace Blackwell system that features a full-size Grace CPU and Blackwell GPU, and now put that in a workstation form factor. The DGX Station combines cutting-edge hardware with NVIDIA's comprehensive AI software suite, creating the ultimate desktop platform for teams pushing the boundaries of AI development.

NVIDIA DGX Station GB300 motherboard

The first desktop system to be built with the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, DGX Station features a massive 784GB of coherent memory space and NVIDIA ConnectX-8 SuperNIC to accelerate large-scale training and inferencing workloads. The rest of the specifications, such as storage and power supply, will be determined by the board partners like Asus and Supermicro.

NVIDIA Grace CPU

72 ARM Neoverse V2 Cores

CPU Memory

Up to 496GB LPDDR5X at 396 GB/s

NVIDIA GPU

NVIDIA Blackwell Ultra

GPU Memory

Up to 288GB HBM3e at 8TB/s

NVLink C2C

Up to 900GB/s

Networking

NVIDIA ConnectX-8 SuperNIC

Net. Bandwidth

Up to 800Gb/s

Conclusions

DGX Spark and DGX Station bring the power of the Grace Blackwell architecture, previously only available in the data center, to the desktop. Research organizations, AI developers, and data scientists can now access powerful AI computing resources right at their desks. These compact yet powerful systems enable rapid prototyping, model fine-tuning, and AI development without the need for constant data center access.

The DGX Spark offers an incredibly compact solution for those needing portable AI compute power, while the DGX Station provides workstation-class performance for more demanding workflows. Both systems represent a significant step forward in democratizing access to AI computing resources, allowing organizations of all sizes to leverage AI in a form factor that makes sense to them.

As an authorized NVIDIA Partner Network DGX provider, Exxact Corporation looks forward to supporting these innovative platforms, helping customers leverage these powerful tools for their AI development needs. Contact us today for more information.

Accelerate AI Training an NVIDIA DGX

Training AI models on massive datasets can be accelerated exponentially with the right system. It's not just a high-performance computer, but a tool to propel and accelerate your research. Deploy multiple NVIDIA DGX nodes for increased scalability. DGX H200 is available for quote today. B200 is available soon!

Get a Quote Today
EXX-Blog-Personal-AI-Blackwell-solutions-NVIDIA-DGX-Spark-Station-1.jpg
News

Personal AI Blackwell Solutions - NVIDIA DGX Spark & DGX Station

April 4, 20254 min read

NVIDIA Spark - AI the Size of a Mac

Imagine a Grace CPU and a Blackwell GPU condensed into a small form factor Mac Mini size workstation. The DGX Spark is the world’s smallest AI supercomputer for users to push the boundaries of generative and physical AI with massive performance and capabilities in a portable form factor.

The DGX Spark features an NVIDIA GB10 Grace Blackwell Superchip capable of over 1,000 TOPS of AI compute with cut-down Grace and Blackwell processors. It also features a ConnectX-7 Smart NIC for clustering two DGX Sparks together for double the performance. The DGX Spark can effectively finetune and iterate workflows that can be delivered straight to DGX Cloud or to existing DGX infrastructure. NVIDIA Spark will be available through NVIDIA for pre-order, and board partners will develop their chassis. Exxact may support DGX Spark in the future.

NVIDIA Grace CPU

20 ARM Cores (10 Cortex-X925 + 10 Cortex A725)

CPU Memory

128GB Unified LPDDR5X at 273 GB/s

NVIDIA GPU

NVIDIA Blackwell Ultra

Networking

1x 10Gb Ethernet

Networking NIC

NVIDIA ConnectX-7 Smart NIC at 200GbE

Net. Bandwidth

Up to 800Gb/s

Storage

Up to 4TB NVMe SSD

Connectivity

WiFi 7 & Bluetooth 5.3

Output

1x HDMI 2.1a

Power Consumption

170W

NVIDIA DGX Station B300 - Grace Blackwell Workstation

Imagine a 1U Grace Blackwell system that features a full-size Grace CPU and Blackwell GPU, and now put that in a workstation form factor. The DGX Station combines cutting-edge hardware with NVIDIA's comprehensive AI software suite, creating the ultimate desktop platform for teams pushing the boundaries of AI development.

The first desktop system to be built with the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, DGX Station features a massive 784GB of coherent memory space and NVIDIA ConnectX-8 SuperNIC to accelerate large-scale training and inferencing workloads. The rest of the specifications, such as storage and power supply, will be determined by the board partners like Asus and Supermicro.

NVIDIA Grace CPU

72 ARM Neoverse V2 Cores

CPU Memory

Up to 496GB LPDDR5X at 396 GB/s

NVIDIA GPU

NVIDIA Blackwell Ultra

GPU Memory

Up to 288GB HBM3e at 8TB/s

NVLink C2C

Up to 900GB/s

Networking

NVIDIA ConnectX-8 SuperNIC

Net. Bandwidth

Up to 800Gb/s

Conclusions

DGX Spark and DGX Station bring the power of the Grace Blackwell architecture, previously only available in the data center, to the desktop. Research organizations, AI developers, and data scientists can now access powerful AI computing resources right at their desks. These compact yet powerful systems enable rapid prototyping, model fine-tuning, and AI development without the need for constant data center access.

The DGX Spark offers an incredibly compact solution for those needing portable AI compute power, while the DGX Station provides workstation-class performance for more demanding workflows. Both systems represent a significant step forward in democratizing access to AI computing resources, allowing organizations of all sizes to leverage AI in a form factor that makes sense to them.

As an authorized NVIDIA Partner Network DGX provider, Exxact Corporation looks forward to supporting these innovative platforms, helping customers leverage these powerful tools for their AI development needs. Contact us today for more information.

Accelerate AI Training an NVIDIA DGX

Training AI models on massive datasets can be accelerated exponentially with the right system. It's not just a high-performance computer, but a tool to propel and accelerate your research. Deploy multiple NVIDIA DGX nodes for increased scalability. DGX H200 is available for quote today. B200 is available soon!

Get a Quote Today