In Monterey, California, the Naval Postgraduate School (NPS) is advancing the U.S. Navy’s AI capabilities through a strategic partnership with NVIDIA. The collaboration centers on a new NVIDIA DGX GB300 system, which will supercharge NPS’s ability to train and deploy AI models for real-world military applications. The system, supported by NVIDIA Mission Control software, will serve over 1,500 in-resident students, 600 faculty members, and thousands of external partners, with plans to expand infrastructure in the future. The DGX GB300 is a cornerstone of NPS’s new NVIDIA AI Technology Center, enabling cutting-edge research in mission planning, autonomous systems, disaster response, and environmental modeling. Retired Col. Randolph Pugh, NPS AI Task Force lead and AI Portfolio director, highlighted the system’s role in developing NPS’s own generative AI model—NPS GPT—providing secure, on-premises access to large language models while ensuring data privacy. NPS has long partnered with MITRE, a nonprofit that supports federal agencies through federally funded research and development centers. MITRE recently shared its Advanced Simulation for Planning and Enhanced Navigation (ASPEN) framework, built on the NVIDIA Omniverse platform. ASPEN uses NVIDIA Isaac Sim to simulate complex underwater environments, enabling realistic training for unmanned underwater vehicles (UUVs) in conditions like strong currents, low visibility, and acoustic interference. ASPEN 3 Network integrates real-world data from MITRE’s BlueTech Lab, an indoor maritime test facility, allowing for hardware-in-the-loop validation of autonomous systems. The simulation combines physics-based models with sensor data from sonar, inertial measurement units, lidar, cameras, and environmental factors such as wind, temperature, salinity, and water depth. The computational backbone for these simulations includes NVIDIA Jetson AGX modules at the edge, processing data in real time. MITRE uses an NVIDIA DGX SuperPOD to train large language and weather forecasting models within its secure Federal AI Sandbox— a dedicated environment for federal agencies to prototype and deploy advanced AI. NPS is leveraging the same three-computer approach from NVIDIA—DGX for training, Jetson for edge inference, and Omniverse for simulation—to build generative physical AI. This includes using NVIDIA fVDB, an open-source GPU-accelerated framework, to process massive 3D datasets and generate high-fidelity digital twins of ocean floors, water columns, atmospheres, and even space environments. These digital twins allow NPS to detect anomalies—such as satellite collisions or debris fields—and predict their trajectories, critical for safeguarding naval and space assets. “If a spacecraft loses maneuverability or an object crashes, we need to predict where every fragment goes,” said Pugh. AI is now embedded across all NPS curricula, empowering military students to conduct applied research that directly supports the Navy and Joint Force. Projects often emerge from hackathons or independent studies, with faculty guiding students to develop solutions that can be funded and operationalized by fleet commands. NVIDIA also supports NPS through the NVIDIA Deep Learning Institute, providing instructor toolkits and resources that enhance the institution’s defense-focused graduate education. “The access to DLI materials has been critical in preparing tomorrow’s leaders in AI,” said Pugh. With this infrastructure and partnership, NPS is not only training future naval leaders but also advancing AI innovation at the forefront of national defense.