The NVIDIA DGX Spark and DGX Station are both personal AI supercomputers designed to bring high-performance AI computing to the desktop, but they cater to different needs and offer distinct performance profiles.
DGX Spark
The DGX Spark is marketed as the world's smallest AI supercomputer and is designed for a broad range of users, including researchers, data scientists, and students. It features the NVIDIA GB10 Grace Blackwell Superchip, which includes a powerful Blackwell GPU with fifth-generation Tensor Cores and FP4 support. This configuration allows the DGX Spark to achieve up to 1,000 trillion operations per second (TOPS) for AI workloads, making it suitable for prototyping, fine-tuning, and deploying AI models locally or on cloud infrastructure[1][2][3].
The DGX Spark comes with 128GB of unified LPDDR5x memory and offers storage options of up to 4TB NVMe SSD. It also supports advanced networking capabilities, such as the ConnectX-7, allowing users to connect multiple Sparks for larger AI projects[3]. The system is compact, runs on a standard electrical socket, and is priced at $3,000, making it more accessible to a wider audience[2][7].
DGX Station
In contrast, the DGX Station is a more powerful desktop system designed for demanding AI workloads, targeting professional users and enterprises. It is powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, which includes an NVIDIA Blackwell Ultra GPU with the latest Tensor Cores and FP4 precision. This setup provides data center-level performance on a desktop, making it ideal for large-scale AI model training and inference[1][6].
The DGX Station boasts an impressive 784GB of coherent memory space, significantly more than the DGX Spark, allowing it to handle much larger and more complex AI models. It also features the NVIDIA ConnectX-8 SuperNIC, which supports networking speeds of up to 800Gb/s. This enables high-speed connectivity between multiple DGX Stations for massive AI workloads and accelerates data transfers[1][5].
While the DGX Spark is available for purchase at $3,000, the DGX Station is expected to be released later in 2025 through manufacturing partners like ASUS, Dell, and HP, with a price likely to be significantly higher, potentially in the range of $10,000 to $50,000[9].
Performance Comparison
In terms of raw performance, the DGX Station is designed to handle more demanding AI tasks due to its larger memory capacity and more powerful GPU. The DGX Spark, while powerful for its size and price, is better suited for smaller-scale AI projects and prototyping. Both systems leverage NVIDIA's Grace Blackwell architecture and advanced interconnect technologies to optimize performance for AI workloads, but the DGX Station's capabilities are more aligned with enterprise-level requirements.
Overall, the DGX Spark offers a balance of performance and affordability, making AI computing more accessible to a broader audience, while the DGX Station is positioned as a high-end solution for professionals and organizations requiring extreme AI processing power.
Citations:
[1] https://nvidianews.nvidia.com/news/nvidia-announces-dgx-spark-and-dgx-station-personal-ai-computers
[2] https://www.theverge.com/news/631957/nvidia-dgx-spark-station-grace-blackwell-ai-supercomputers-gtc
[3] https://www.pcmag.com/news/what-is-nvidias-dgx-station-a-new-specialized-desktop-line-for-ai-work
[4] https://www.nvidia.com/en-eu/project-digits/
[5] https://www.ainvest.com/news/nvidia-unveils-dgx-spark-dgx-station-revolutionizing-personal-ai-computing-2503/
[6] https://opentools.ai/news/nvidia-unleashes-the-future-with-personal-ai-supercomputers
[7] https://www.reddit.com/r/LocalLLaMA/comments/1jedy17/nvidia_digits_specs_released_and_renamed_to_dgx/
[8] https://www.nvidia.com/en-us/products/workstations/dgx-station/
[9] https://www.youtube.com/watch?v=krBh0Von-2A
[10] https://www.youtube.com/watch?v=csIhxri1JT4
[11] https://www.turtlesai.com/en/pages-2527/nvidia-introduces-dgx-spark-fka-digits-and-dgx-station