Nvidia CEO Jensen Huang officially debuted two high-performance “personal AI supercomputers,” the DGX Spark and DGX Station, during a keynote presentation this Tuesday. Designed to empower developers and enterprises, these machines enable the prototyping, fine-tuning, and deployment of AI models directly at the edge.
The Future of AI Computing
Positioning the new hardware as the definitive technology for the current era, Huang emphasized that these systems represent the new standard for enterprise computing. “This is the computer of the age of AI,” Huang stated. “This is what computers should look like and what they will run in the future.”
Technical Specifications and Performance
The hardware lineup targets a wide range of computational needs:
- DGX Spark: Previously known as Project Digits, this unit is powered by the GB10 Grace Blackwell Superchip, delivering a massive 1,000 trillion operations per second of AI computing power.
- DGX Station: This workstation-class machine features the Nvidia GB300 Grace Blackwell Ultra Desktop Superchip, paired with 784GB of memory for demanding workloads.
Market Availability and Ecosystem
Nvidia is moving quickly to integrate these systems into the enterprise ecosystem. The DGX Spark is available for purchase immediately. The DGX Station is slated for release later this year, with manufacturing support provided by industry leaders including Asus, Boxx, Dell, HP, and Lenovo.
Empowering the Age of AI Agents
According to Huang, the shift toward ubiquitous AI agents necessitates a fundamental change in infrastructure. By providing dedicated, powerful hardware for local environments, Nvidia aims to streamline how businesses build and manage the next generation of intelligent software. “AI agents will be everywhere,” Huang noted. “How they run, what enterprises run, and how we run it will be fundamentally different. And so we need a new line of computers.”