WASHINGTON – The U.S. Department of Defense has formalized agreements with eight artificial intelligence companies to deploy advanced AI capabilities across classified operational networks, signaling a systemic shift toward the integration of private-sector machine learning into national security infrastructure.
The move integrates a cross-section of the AI industry, from semiconductor hardware and cloud computing to generative models and satellite communications, to automate decision-making processes in military environments. This strategic alignment concentrates critical defense capabilities within a small group of dominant technology firms, deepening the Pentagon’s long-running reliance on commercial technology in programs governed by frameworks such as the DoD acquisition and modernization system.
Corporate Integration in Classified Networks
The agreements involve a combination of hyperscale cloud providers, hardware manufacturers, and specialized AI developers. The participating companies include:
- Amazon Web Services (AWS)
- Microsoft
- Nvidia
- OpenAI
- Oracle
- SpaceX
- Reflection
The inclusion of the “Big Three” cloud providers-AWS, Microsoft, and Google-leverages existing government cloud frameworks designed to handle sensitive data, including impact-level environments certified for classified workloads. These firms provide the scalable compute power required to run large-scale AI models on secure servers that must comply with federal cybersecurity and procurement rules.
Nvidia provides the essential GPU architecture and hardware acceleration necessary for training and deploying the neural networks used in these operational tools. Oracle adds database management and secure cloud infrastructure, while SpaceX provides the satellite connectivity and telemetry data required for real-time AI analysis in remote theaters, extending classified connectivity to ships, aircraft, and forward operating bases.
Collectively, the agreements move AI from pilot projects into the backbone of command-and-control, intelligence, logistics, and battlefield management systems. They also raise questions about vendor concentration and long-term lock-in, as a small circle of technology giants becomes embedded in the classified workflows that underpin U.S. defense planning.
“These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force and will strengthen our warfighters’ ability to maintain decision superiority across all domains of warfare,” the U.S. Department of Defense said in a statement.
The deployment of OpenAI’s capabilities suggests a move toward integrating large language models (LLMs) into classified workflows, potentially for intelligence synthesis, operational planning support, and automated document analysis. Pentagon officials have previously framed such tools as “decision aids,” stressing that they are intended to augment, rather than replace, human commanders – a distinction that will test emerging policy guardrails on responsible AI use in defense settings.
Governance, Oversight, and AI Doctrine
The multi-company arrangements slot into the department’s broader AI strategy, including internal directives on algorithmic accountability and the designation of responsible AI principles for combat and non-combat applications. Under these policies, systems deployed on classified networks are expected to undergo technical testing, red-teaming, and legal review to meet both operational requirements and the laws of armed conflict.
At the same time, the scale of the new integrations is likely to sharpen scrutiny from Congress, which oversees major technology contracts through defense appropriations and authorization processes, and from allies that share data with U.S. systems. Lawmakers and regulators are expected to press for clarity on how proprietary AI models will be audited, how data will be protected from commercial reuse, and how quickly humans can override or shut down automated recommendations in crisis scenarios.
Maritime AI and Specialized Detection
Parallel to the broader departmental agreements, the U.S. Navy has contracted with San Francisco-based Domino Data Lab to implement AI-driven mine detection in the Strait of Hormuz, a chokepoint critical to global energy shipments and long viewed as a potential flashpoint for maritime escalation.
The contract is valued at up to 99.7 million U.S. dollars. Domino Data Lab specializes in enterprise AI platforms that allow data scientists to orchestrate and scale machine learning models within secure, governed environments. In this case, the platform is being deployed under Navy program oversight and within rules of engagement that constrain how autonomous systems may identify and prosecute potential targets.
The Navy is utilizing this software to process data from multiple sensors, enabling unmanned underwater systems to identify and categorize new types of naval mines. By automating the identification process, the Navy reduces the time required to detect threats in high-traffic maritime corridors and can update detection models as adversaries adapt their mine designs.
The deployment focuses on enhancing the speed of detection and the accuracy of unmanned systems, reducing the reliance on human analysis for initial threat identification in contested waters while preserving human control over final decisions to neutralize or avoid suspected mines. Officials describe the effort as a test case for integrating commercially developed AI into highly sensitive missions, where errors carry both operational and diplomatic consequences.
The Domino Data Lab contract is currently active, with software integration proceeding for unmanned underwater systems and additional evaluation phases expected as the Navy measures performance against live data in the region.
