Governments everywhere are racing to harness the power of AI — but legacy infrastructure isn’t built for the velocity, complexity or trust that mission-critical action now demands. Massive data streams, cyber threats and urgent operations require a new blueprint for creating AI factories purpose-built for the public sector’s unique standards and scale.
At NVIDIA GTC Washington, D.C., NVIDIA today unveiled the NVIDIA AI Factory for Government reference design, in collaboration with software and services leaders, to enable federal agencies and regulated industries to build and deploy new AI platforms and intelligent agents.
The reference design provides guidance for deploying full-stack AI infrastructure using the latest NVIDIA AI Enterprise software — now designed to meet stringent security standards for FedRAMP-authorized clouds and high-assurance environments.
Industry Leaders Tap NVIDIA AI Factory for Government
NVIDIA AI Factory for Government runs on recommended hardware configurations from NVIDIA Enterprise Reference Architectures, based on the NVIDIA Blackwell architecture, including NVIDIA RTX PRO Servers and NVIDIA HGX B200 systems. It also includes NVIDIA Spectrum-X Ethernet, the NVIDIA BlueField platform, NVIDIA-Certified Storage, the latest NVIDIA AI Enterprise software and NVIDIA Nemotron open models.
Building on the NVIDIA Enterprise AI Factory validated design announced at COMPUTEX, the AI Factory for Government reference design features NVIDIA AI Enterprise software, which is now built to meet rigorous security standards. NVIDIA AI Enterprise includes new capabilities like advanced code scanning, vulnerability management and continuous monitoring to ensure the highest levels of security. This enables AI workloads to remain configured, regularly updated and ready for mission-critical deployments across federal and high-assurance environments.
NVIDIA is working with leading technology platform providers to operationalize AI with the latest NVIDIA AI Enterprise software, helping serve the needs of the public sector and other highly regulated industries.
Palantir and NVIDIA are building a first-of-its-kind integrated technology stack for operational AI that integrates Palantir Ontology — the core of the Palantir AI Platform (AIP) — with NVIDIA data processing and route optimization libraries, open models and accelerated computing to speed up AI deployments for enterprises and government.
AIP will integrate NVIDIA AI Enterprise software, NVIDIA Nemotron and other AI capabilities included in the AI Factory for Government reference design to power domain-specific intelligence and AI agents for the complex operating environments of highly regulated industries.
To strengthen cybersecurity in AI factories, CrowdStrike will expand its Agentic Security Platform to support the NVIDIA AI Factory for Government reference design, enabling organizations to build and deploy AI agents in federal and high-assurance environments.
CrowdStrike will also integrate NVIDIA Nemotron open models, NVIDIA NeMo Data Designer and the NVIDIA NeMo Agent Toolkit through Charlotte AI AgentWorks to deliver autonomous, continuously learning AI agents for real-time threat detection and response across cloud, data center and edge environments.
ServiceNow is integrating the latest NVIDIA AI Enterprise software in the ServiceNow AI Platform for U.S. federal customers — suitable for FedRAMP-authorized clouds and high-assurance, on-premises environments — to boost productivity and drive savings for public sector agencies. ServiceNow also announced today its Apriel 2.0 model — the latest in its Apriel Nemotron model family, engineered to deliver frontier-level AI reasoning and multimodal capabilities to enterprises in a faster, smaller, more cost-efficient footprint.
Astris AI, a Lockheed Martin company, is integrating the latest NVIDIA AI Enterprise software into its Astris AI Factory to enable accelerated, secure AI deployments for classified and mission-critical environments, in alignment with strict federal security standards.
The Astris AI Factory has been used internally at Lockheed Martin for more than five years. Now commercially available, Astris AI will use the latest NVIDIA AI Enterprise software to develop and deploy internal AI agents in industries that rely on trust and precision.
“Success in complex missions depends on AI that’s secure and reliable,” said Jim Taiclet, chairman, president and CEO of Lockheed Martin. “By working with Astris AI and using the latest NVIDIA AI Enterprise tools, we’re speeding up how we develop and deliver AI systems that improve precision and performance in critical operations.”
Northrop Grumman is deploying an AI factory powered by NVIDIA RTX PRO Servers, Spectrum-X Ethernet networking and government-ready NVIDIA AI Enterprise software for deploying secure enterprise AI services, enabling development of advanced AI capabilities and boosting productivity and operational efficiency across its workforce of nearly 100,000 employees.
Partner Ecosystem Readies AI Factories to Accelerate AI Agents
NVIDIA partners are helping enable the world’s organizations to rapidly build and deploy AI agents and applications across federal and enterprise organizations with the NVIDIA AI Factory for Government reference design.
Developers deploying AI agents on their AI factory infrastructure can tap into partner platforms from Dataiku, DataRobot, Domino Data Lab and H20.ai to build, orchestrate, operationalize and scale agentic and predictive AI workflows. In addition, the reference design supports vector databases from Elastic and EnterpriseDB to help agents store, search and retrieve data.
Enterprises can secure their AI factories with software from observability and security partners including Dynatrace, Fiddler, JFrog, Trend Micro and Weights & Biases. Customers can also secure their agentic AI applications by adding NVIDIA Confidential Computing to their hybrid and on-premises AI factories through data security platform partners such as Fortanix.
Software infrastructure and deployment partners including Canonical, Mirantis, Nutanix, Red Hat, Spectro Cloud and Broadcom can help enterprises seamlessly scale and manage AI agent workloads across complex, high-assurance, on-premises environments.
Cloud providers including CoreWeave and Oracle Cloud Infrastructure are supporting the AI Factory for Government reference design so their government customers can build and deploy government-ready AI factories in secure cloud environments.
Leading server manufacturers including Cisco, Dell Technologies, HPE, Lenovo and Supermicro will offer full-stack AI factory offerings using the NVIDIA AI Factory for Government reference design to further accelerate AI deployments for the public sector and highly regulated industries.
Watch the GTC Washington, D.C., keynote from NVIDIA founder and CEO Jensen Huang and explore sessions.
See notice regarding software product information.