The US military acquires its inaugural Nvidia GB300 AI supercomputer.
The Nvidia chip at the heart of America's export dispute with China has found a new location: a classroom in Monterey. The Naval Postgraduate School has established a DGX GB300 AI supercomputer, the first of its kind in the US military, as reported by Nextgov.
This machine operates on Nvidia’s Blackwell Ultra generation, which the US prohibits from being sold to Chinese companies. This particular unit was donated by Nvidia to the school's foundation. It is equipped with 36 Grace CPUs and 72 Blackwell Ultra GPUs, configured to perform tasks in parallel.
The institution plans to utilize it for training students in areas such as weather and ocean modeling, operations research, cybersecurity, and resilience planning. It is not a data center, as clarified by the school, but rather a single integrated system that integrates with existing power and water resources. Companies like Vertiv, DDN, and VAST Data assisted in its setup.
Nvidia presented the donation as a patriotic gesture and also as a demonstration of its military-oriented hardware. CEO Jensen Huang stated that the school's personnel could “train on it, build with it, and lead America’s fighting force.”
Focusing on understanding limits rather than just power
A more intriguing aspect is the mission. The institution aims to have officers who don't just rely on the technology without question. “The issue isn’t whether these technologies will be available,” said Captain Michael Owen, a Navy vice provost. It is “whether future military leaders grasp them sufficiently to assess their restrictions, apply them ethically, and use them responsibly.”
The supercomputer builds upon a research agreement between Nvidia and the school made in December 2024. This agreement also led to the establishment of an AI Tech Center in 2025, equipped with Nvidia’s Blackwell chips, Jetson devices, and its Omniverse simulation platform.
The symbolism is unmistakable. A chip deemed too sensitive for sale to Beijing is now being used as an educational tool for American officers. The company that produces it donated it at no cost. The lesson the Navy seeks is not about firepower, but rather about judgment: discernment of when even the most advanced AI in the world should not be wholly trusted.
Other articles
Cambridge's planet-scale AI model operates on AMD rather than Nvidia.
Cambridge's TESSERA transforms satellite data into reusable AI embeddings for every 10 meters of the Earth's surface, reducing the need for labeled data by 30 times. It operates on AMD hardware instead of Nvidia.
Tesla sold 25% more vehicles but experienced a 57% decrease in operating profit. The cause is attributed to spending on AI.
In Q2, Tesla's revenue reached $28.24 billion, marking a 26% year-over-year increase, with 480,126 deliveries. However, operating income plummeted by 57% to $398 million. Research and development expenditures surged by 49% to $2.37 billion, focusing on AI and robotics.
Samsung ignited the competitive foldable market, but I’m more enthusiastic about an impending unique competitor.
The speculated Mix Fold 5 from Xiaomi may combine wide dimensions similar to the Galaxy Z Fold 8 with its own premium chipset, a 200MP camera, and a significantly larger 6,000mAh battery.
The postponement of Google's Gemini highlights a more significant issue: employee morale.
Insiders informed Axios that low morale, a disputed Pentagon agreement, and a talent drain are hindering Google's AI models, leading to negative cash flow. Google refutes these claims.
The Framework Desktop combines Ryzen AI Max, supports up to 192GB of RAM, and incorporates local AI into a compact PC.
Framework has showcased a new Desktop model equipped with 192GB of RAM and a powerful new processor, designed for running large AI models from home. Here’s what we’ve learned up to this point.
A retina chip designed to restore vision is now available for purchase in Europe.
Science Corp's Prima retina implant has regained partial vision and has received EU approval, marking it as the first brain-implant device from the US to be available to regular patients.
The US military acquires its inaugural Nvidia GB300 AI supercomputer.
Nvidia has contributed a DGX GB300 supercomputer to the Naval Postgraduate School, marking a military first, to educate officers on the applications and limitations of AI.