Nvidia has officially announced the DGX Spark, a compact AI supercomputer that delivers data center-level performance in a device small enough to fit in a backpack. Previously known under the codename Project Digits, the DGX Spark measures less than two inches tall and six inches wide, making it one of the most portable high-performance computing systems on the market. Despite its miniature size, the device boasts performance capabilities that surpass many full-sized desktops and laptops by delivering up to one petaFLOP of processing power.
Powering the DGX Spark is Nvidia’s cutting-edge GB10 Grace Blackwell Superchip, which delivers up to 1,000 trillion operations per second. The system includes 128 gigabytes of unified memory, optimized for handling advanced AI workloads. Nvidia is also bundling the device with its full AI software suite, which includes pretrained models, libraries, and essential tools that support seamless development and deployment of machine learning applications.
With a starting price of $3,999, the DGX Spark is positioned to bring high-performance AI capabilities to a broader market. Reservations are now open, and major manufacturers such as ASUS, Dell, and Lenovo are partnering with Nvidia to develop branded versions tailored to different user needs. The compact design and powerful performance make it suitable for researchers, engineers, and developers working beyond traditional office environments.
Nvidia also introduced the DGX Station, a larger and more robust AI workstation designed for enterprise environments. Built with the GB300 Grace Blackwell Ultra Superchip, it features 748 gigabytes of coherent memory and Nvidia’s ConnectX-8 SuperNIC, allowing for network transfer speeds up to 800 gigabits per second. The DGX Station is ideal for high-demand workflows such as AI training, simulations, and collaborative development.
Together, the DGX Spark and DGX Station represent Nvidia’s commitment to making AI supercomputing more accessible and portable. These innovations empower users from students to seasoned professionals, enabling advanced computing without the need for sprawling server racks. With this release, Nvidia continues to redefine the limits of what’s possible with compact AI technology.
