NVIDIA’s Jensen Huang Hand-Delivers the “World’s Smallest Supercomputer,” the DGX Spark, to Elon Musk
In a moment that feels both futuristic and nostalgic, NVIDIA CEO Jensen Huang has personally delivered the DGX Spark, dubbed the world’s smallest supercomputer, to Elon Musk, marking another milestone in the company’s pursuit of accessible AI computing. The delivery took place just as SpaceX’s 11th Starship test was underway, symbolically connecting two of the world’s most ambitious technological endeavors.
This gesture echoes a familiar scene from years past when Huang delivered one of the first DGX-1 units to Musk during his time at OpenAI. The DGX Spark continues that tradition, embodying NVIDIA’s mission to make AI supercomputing power accessible to developers, researchers, and innovators worldwide.
Originally unveiled at CES 2025, the DGX Spark will officially hit retail shelves on October 15, 2025, with orders available through NVIDIA’s official store and partner vendors including Acer, ASUS, Dell Technologies, GIGABYTE, HP, Lenovo, and MSI.
DGX Spark Key Specifications:
NVIDIA GB10 Grace Blackwell Superchip — delivers up to 1 petaflop of AI performance at FP4 precision.
128GB unified CPU–GPU memory — enables local prototyping, fine-tuning, and inference without relying on the cloud.
NVIDIA ConnectX networking for clustering and NVLink-C2C interconnect offering 5x PCIe bandwidth.
NVMe storage for ultra-fast throughput and HDMI out for visual output.
While retail availability was initially targeted for July, the release was delayed to October due to production complexities involving the custom NVIDIA and MediaTek GB10 SoC. Despite the setback, the device is poised to redefine local AI computing—packing supercomputer-grade performance into a palm-sized chassis.
With a starting price of $3,999, the DGX Spark isn’t for the casual user, but it represents a powerful step toward democratizing AI development by allowing creators and researchers to train, test, and deploy AI models without relying solely on cloud infrastructure.
What do you think of NVIDIA’s DGX Spark? Would you invest in a personal supercomputer for AI experimentation, or stick with cloud-based solutions?
