Nvidia's Tiny $3,000 Computer Steals the Show at CES

At the Consumer Electronics Show (CES) 2025, Nvidia unveiled its latest innovation, the Project DIGITS AI supercomputer, stealing the spotlight with its compact size and impressive performance. Priced at just $3,000, this tiny computer packs a powerful punch, offering 1 PFLOPS of FP4 AI compute performance. But is it worth the hype, and how does it compare to other AI supercomputers in its price range?



Nvidia's Project DIGITS is powered by the new GB10 Grace Blackwell Superchip, which delivers up to 1 petaflop of AI performance at FP4 precision. The Superchip is a system-on-a-chip (SoC) based on the Nvidia Grace Blackwell architecture, featuring an Nvidia Blackwell GPU with latest-generation CUDA cores and fifth-generation Tensor Cores, connected via NVLink-C2C chip-to-chip interconnect to a high-performance Nvidia Grace CPU with 20 power-efficient cores built with the Arm architecture. MediaTek collaborated on the design of the GB10 Superchip, contributing to its best-in-class power efficiency, performance, and connectivity.



The Project DIGITS AI supercomputer features 128GB of unified, coherent memory and up to 4TB of NVMe storage, allowing developers to run up to 200-billion-parameter large language models. Additionally, two Project DIGITS AI supercomputers can be linked to run up to 405-billion-parameter models using Nvidia ConnectX networking.

While Nvidia's Project DIGITS offers impressive performance, its price-to-performance ratio has been called into question by industry experts. Raja Koduri, a graphics chip design icon, estimated that the upcoming GeForce RTX 5070, priced around $600, has FP16 performance close to Project DIGITS. Tiny Corp, a startup known for its abrasive approach to AI acceleration, dismissed the $3,000 price point, advising customers to "Just buy a gaming PC."




Despite the skepticism, Nvidia's Project DIGITS has the potential to revolutionize AI research and development by providing access to powerful AI computing performance on a desktop system. With its extensive library of Nvidia AI software, including software development kits, orchestration tools, frameworks, and models available in the Nvidia NGC catalog and on the Nvidia Developer portal, Project DIGITS empowers developers to prototype, fine-tune, and run large AI models on their own desktop system before seamlessly deploying them on cloud or data center infrastructure.



In conclusion, Nvidia's Project DIGITS AI supercomputer has generated buzz at CES 2025 with its compact size and impressive performance. While its price-to-performance ratio has been called into question, the potential applications and use cases for Project DIGITS in AI research and development are vast. As Nvidia continues to push the boundaries of AI computing, Project DIGITS may just be the catalyst that democratizes AI supercomputing and accelerates innovation in the field.