TLDR: Nvidia CEO Jensen Huang personally delivered the company’s new DGX Spark AI supercomputer to Elon Musk at SpaceX and then to Sam Altman at OpenAI. This compact, shoebox-sized device offers a petaflop of AI performance, enabling local training and inference of large AI models, and signifies a major leap in portable AI computing.
In a move that underscores the intense pace of innovation and collaboration within the artificial intelligence industry, Nvidia CEO Jensen Huang personally hand-delivered the company’s latest AI supercomputer, the DGX Spark, to two of the most influential figures in tech: Elon Musk and Sam Altman. This gesture, combining symbolism with cutting-edge technology, marks a significant moment in the evolution of portable AI computing.
Huang’s first stop was SpaceX’s Starbase facility in Texas, where he presented the DGX Spark to Elon Musk. The delivery occurred just before SpaceX’s 11th Starship test flight, creating a striking visual contrast. Standing beside the colossal rocket, Huang humorously remarked, “Delivering the smallest supercomputer next to the biggest rocket.” A video shared by Musk on X showed the two leaders discussing Nvidia’s historical ties with OpenAI over lunch in the SpaceX cafeteria. Huang reminisced about personally delivering the first DGX-1 to OpenAI years ago, highlighting how the new Spark continues the mission of bringing data-center-level power directly to innovators.
Following his visit to Texas, Huang traveled to San Francisco to meet with Sam Altman at OpenAI’s headquarters. This reunion echoed Nvidia’s 2016 collaboration with OpenAI, where the DGX-1 played a crucial role in foundational AI research. OpenAI co-founder Greg Brockman expressed his appreciation on X, stating, “Thanks Jensen for the hand delivery of DGX Spark. Best delivery service ever. Amazing to see so much compute (1 petaflop!) in such a tiny form factor.”
The DGX Spark itself is a marvel of engineering. Described as no larger than a shoebox and weighing approximately 1.2 kilograms, it packs a formidable punch. This compact AI supercomputer delivers up to one petaflop of AI performance and features 128GB of unified memory. It is capable of running inference on AI models with up to 200 billion parameters and fine-tuning models of up to 70 billion parameters locally. Its advanced hardware includes the Nvidia GB10 Grace Blackwell Superchip, Nvidia ConnectX networking, NVLink-C2C for rapid data movement, NVMe storage, and HDMI output for direct visualization.
Also Read:
- Nvidia Forges Key Alliances to Propel AI Agents and 6G Networks to the Edge
- Arm and Meta Forge Deeper Alliance for Next-Generation AI Efficiency
Nvidia’s introduction of the DGX Spark signifies a strategic shift towards expanding the AI ecosystem beyond traditional data centers. By providing such immense computing power in a portable form factor, the DGX Spark empowers developers, researchers, and artists to create and refine AI innovations, such as chatbots and generative vision tools, without solely relying on cloud infrastructure. This enables faster iteration, privacy-friendly experimentation, and pushes the boundaries of AI computing into everyday development environments.
