Nvidia has introduced two powerful personal AI supercomputers at its GTC 2025 event. Named DGX Spark and DGX Station, these systems are powered by Nvidia's Grace Blackwell chip platform, designed to bring AI capabilities right to your desktop.

Key features of Nvidia DGX Spark and DGX Station

The DGX Spark is known as the world’s smallest AI supercomputer. It comes with the GB10 Grace Blackwell Superchip, delivering an impressive speed of up to 1,000 trillion operations per second. Its small size makes it perfect for researchers, developers, and students to create and improve AI models without needing a massive setup.

On the other hand, the DGX Station is a more advanced system. It features the GB300 Grace Blackwell Ultra Desktop Superchip and a massive 784GB of memory. This setup makes it powerful enough for large-scale AI training and inferencing. It also includes the ConnectX-8 SuperNIC, allowing super-fast network speeds of up to 800Gb/s. Both systems are designed to switch easily between local and cloud-based AI tasks, making them flexible and convenient for developers and data scientists.

Availability and pricing

The DGX Spark is available for reservation starting today. Meanwhile, the DGX Station is expected to launch later this year through popular brands like Asus, Dell, HP, and Lenovo. The DGX Spark’s base model is estimated to start at around $3,000, but Nvidia hasn’t shared official prices for both systems yet.

During his keynote, Nvidia CEO Jensen Huang highlighted the importance of these new systems, stating, "AI has transformed every layer of the computing stack. It stands to reason a new class of computers would emerge—designed for AI-native developers and to run AI-native applications."

You might also be interested in: Nvidia loses $600 billion as China's DeepSeek AI disrupts tech market