For federal agencies and industry partners, NVIDIA’s GTC DC conference felt less like a standard tech gathering and more like a strategic planning session for the next era of artificial intelligence (AI). Sessions touched every part of the federal AI landscape: supercomputing, infrastructure, mission delivery, and policy, providing a clear picture of where government capabilities are headed.
In case you missed it, here are six essential takeaways for federal leaders.
- The United States is turning to AI supercomputers to accelerate science
NVIDIA CEO Jensen Huang announced a partnership with the Department of Energy to build seven new AI supercomputers to “advance our nation’s science.”
At Argonne National Laboratory, the flagship Solstice system will deliver 100,000 NVIDIA Blackwell GPUs, with an additional Equinox system adding another 10,000 GPUs in 2026. More systems – Tara, Minerva, and Janus – are also planned for Argonne. Los Alamos National Laboratory will deploy Mission and Vision systems built on the NVIDIA Vera Rubin platform and Quantum-X800 InfiniBand networking.
“These systems will be a powerhouse for scientific and technological innovation,” said DOE Secretary Chris Wright. Los Alamos Director Thom Mason added that the partnership will drive discoveries that strengthen the resilience of critical infrastructure.
Read more: NVIDIA Partnering With DOE to Build 7 AI Supercomputers
- More funding is needed to sustain America’s technological edge
Sen. Todd Young emphasized the importance of expanding federal research investments. Young, chair of the National Security Commission on Emerging Biotechnology, said the United States should double federal research expenditures to remain globally competitive.
Young called for shifting investments toward strategically significant technologies – particularly AI and biotechnology – with clear commercial and national security value.
“If an investment has an obvious spillover into the national security realm, as so much of AI does, then it should get some special dispensation,” he said.
Young also pushed for rewarding institutions that demonstrate efficient outcomes and for preserving the balance between public and private research.
Read more: Sen. Young Champions New Vision for Strategic Research Spending
- AI factories are emerging as the next modernization backbone
Across multiple sessions, one theme appeared repeatedly: AI factories – full-stack environments where every layer, from the silicon to the software, is optimized to transform raw data into real-time insights.
“We’re putting more demand on institutional frameworks than ever before,” said Ken Patchett, vice president of data center infrastructure at Lambda, which builds gigawatt-scale AI factories. He described them as the critical backbone for modern AI, built around dense compute, power, and cooling.
Together AI’s Mahadev Konar noted rising enterprise and sovereign demand, while ReflectionAI’s Ioannis Antonoglou stressed the importance of open infrastructure that allows nations and organizations to maintain control over their own models and data.
For federal missions, the AI factory model aligns with sovereignty, compliance, and mission-specific deployment requirements – elements increasingly emphasized across national AI policy.
Read more: AI Factories Power Next Wave of Modernization, Sovereignty, and Compliance
- AI factories accelerate government deployments
To make AI factories more accessible to federal agencies, NVIDIA introduced an AI Factory for Government reference design, developed with a coalition of software and services partners to support secure, scalable AI platforms in regulated environments.
Supermicro announced new systems aligned to that design, including a Super AI Station based on NVIDIA GB300 and new rack-scale NVIDIA GB200 NVL4 HPC solutions. The company also announced a compact, high-density 2OU NVIDIA HGX B300 8-GPU server and the NVIDIA Vera Rubin NVL144 and NVIDIA Rubin CPX platforms, which will launch in 2026.
Supermicro senior storage architect Randy Kreiser emphasized that Supermicro’s “data center building block” approach enables pre-built, validated AI systems that simplify secure deployment. Supermicro can pre-build and validate complete AI systems – for example, assembling a SuperPod in its San Jose factory, testing it under liquid cooling, and letting customers “touch it – either remotely or on site,” Kreiser said.
The momentum behind AI factory-aligned architectures mirrors federal priorities for sovereign, secure, and scalable AI infrastructure, including OMB guidance on AI procurement and the National Science Foundation’s National AI Research Resource Pilot.
Read more: Building AI Factories for the Federal Government
- Leidos outlines seven dimensions of trustworthy AI
Leidos presented its Framework for AI Resilience and Security, built around seven dimensions: fairness, assurance, security, resilience, explainability, accuracy, and adaptability. The overarching goal, said Corey Hendricks, vice president and chief engineer for the Commercial and International Sector, is “trustworthy automation at mission scale.”
Leidos supports AI systems across Transportation Security Administration and Federal Aviation Administration environments that protect more than 2 million travelers a day. Tools such as ProSight and ProVision apply deep learning to improve threat detection, reduce false alarms, and validate cargo manifests – enhancing both safety and passenger experience.
Read more: NVIDIA GTC: Leidos Highlights Trustworthy AI Systems for DHS, DOT
- Northrop Grumman showcases mission AI from space to seabed
Northrop Grumman’s Travis Garriss highlighted how AI-driven modernization plays out in extreme mission environments, “from the harshest environments in space to the bottom of the ocean.”
The company is using NVIDIA’s Isaac platform to enable agentic systems that can retrieve and analyze data and act on it autonomously – which is critical for missions where human access is impossible.
“Imagine … the James Webb Telescope,” said Garriss, vice president and chief information and digital officer for Northrop Grumman. “You don’t exactly pull up to that and type in changes when there’s an issue. There [is] no hands-on keyboard.”
Northrop Grumman is also using NVIDIA’s Omniverse platform to support large-scale digital twins, improving performance across spacecraft, aircraft, microelectronics, and underwater systems.
Read more: Northrop Grumman Speeds AI Innovation Through NVIDIA Partnership
A new phase of federal AI is here
Together, the GTC DC sessions illustrated a clear transition: Federal AI is moving from experimentation to operational deployment.
Key themes emerged:
- AI factories are becoming the core modernization architecture
- Frontier-scale supercomputing is accelerating scientific and mission innovation
- Strategic investments are essential to maintaining U.S. leadership
- Mission partners across industry are delivering trusted, resilient AI systems for federal environments
For more insights and session replays, explore NVIDIA GTC DC on demand.