NVIDIA Launches DGX Spark And DGX Station: Supercomputer AI Directly On Your Work Table

JAKARTA In Computex 2025, NVIDIA introduces two new personal computing devices specifically designed for artificial intelligence (AI): DGX Spark and DGX Station needs. Both systems offer the computing power of a supercomputer class but in the form of desktop devices that can be used directly in private workspace. Carrying the latest technology from the Grace Blackwell chip, NVIDIA brings a new revolution in the world of AI and high performance computing.

DGX Spark and DGX Station are not just ordinary personal computers. These two devices are aimed at developers, researchers, and AI laboratories that require extreme computational performance to handle heavy tasks such as large-scale AI model training, complex simulations, and the development of a generating AI system. What distinguishes from traditional solutions is that all of this can now be done without the need for a large data center or a dedicated server space.

DGX Spark is an ideal choice for users who have limited space and budget but need qualified AI power. This device is powered by NVIDIA GB10 Grace Blackwell Superchip which is able to provide performance of up to 1 petaflop, an extraordinary number that could previously only be found in large-scale cloud computing systems.

Armed with a compact design and an integrated memory capacity of 128GB, DGX Spark can handle large dataset processing and complex neural networks without a hitch. In addition, the system is also fully integrated with NVIDIA ecosystems such as DGX Cloud, allowing users to build and test AI models locally and then scale to the cloud easily without having to change their workflows.

Meanwhile, the DGX Station comes as a more resilient version and is aimed at users who need maximum performance in their work environment. Using the latest generation GB300 chip, the DGX Station is capable of achieving computational performance of up to 20 petaflops, making it one of the most powerful AI desktop systems in the world.

The device is also equipped with 784GB integrated memory and connectivity support of up to 800Gb/s, enabling high-speed data transfer for large-scale AI training and inference needs. One of the main advantages of DGX Station is the Multi-Instance GPU (MIG) technology, which allows this system to be divided into several independent computer ministries' to run many tasks at once efficiently.

Both DGX Spark and DGX Stations come with complete software packages from NVIDIA, including DGX OS, NIM micro services, as well as various AI tools and libraries such as PyTorch and Jupyter that have been installed previously. This makes both devices ready to be used directly since they were first turned on "without complicated configurations, really plug and play systems for AI needs.

NVIDIA also announced that a number of well-known manufacturers such as Dell, HP, Acer, MSI, and GIGABYTE will start offering this device to the market from July 2025. This collaboration is believed to accelerate the adoption of advanced AI technology by professionals in various sectors, ranging from scientific research to the development of commercial technology.

With the launch of DGX Spark and DGX Station, NVIDIA once again proves its position as a leader in the global AI industry. Now, AI's supercomputer no longer only belongs to a large laboratory or a giant company but can be accessed directly by anyone who needs it, right on their desk.