The Rapid Evolution of Scaled Software Solutions for National Defense

The urgency of the national defense mission is increasing the pressure to rapidly scale technology in the Department of Defense (DoD) and other security agencies. A January 2025 Defense Innovation Board study on scaling nontraditional defense innovation framed the challenge starkly: “We must act swiftly to ensure the DoD leads in global innovation and competition over AI and autonomous systems – and is a trendsetter for their responsible use in modern warfare. The importance of these tasks cannot be understated; our very democracy and way of life are at stake.” Alongside this study, the DIB released another on scaling unmanned weapons systems.

Software lies at the heart of both artificial intelligence (AI) and autonomous systems, and the DoD has made great strides in keeping pace with the speed of software advancement. However, the need for these two crucial technologies to both advance technically and permeate the defense domain threatens to outstrip current paradigms for software acquisition, operation, and testing.

The SEI’s 2006 report on ultra-large-scale systems and 2021 study on the future of software engineering research and development foresaw this moment. It is no longer theoretical: AI is being integrated into broad societal systems, and unmanned weapons systems are turning the tide on present-day battlefields.

But we do not meet this moment empty-handed. The DoD and the defense industrial base can double down on the proven software development accelerators of Agile methodologies and continuous deployment of capability. The SEI has been, and will continue to be, instrumental to their evolution. More than ever, software system developers and operators need these methods in order to move at speed and with discipline to deliver capability to the warfighter. The SEI continues to improve them with new tools like the Polar framework for DevSecOps data and a study of the state of DevSecOps within the DoD. Our recent Capability-Based Software Cost Estimation is a novel method, aligned to Agile and DevSecOps, built for flexibility and speed.

SEI work already transitioned to the DoD is bearing fruit. The SEI’s Architecture Tradeoff Analysis Method has hastened the analysis of software architecture and design tradeoffs. The faster a development organization moves, the more it will need to manage its technical debt, another long-standing area of SEI expertise. We have also worked with DoD acquisition programs to improve their processes, as we did in the development of 2020’s Software Acquisition Pathway, which a 2025 memo from the secretary of defense mandated as the preferred software pathway for all DoD components.

Where point solutions once satisfied end users, national defense now calls for large-scale solutions, such as development of the Universal Command and Control (UC2) program led by the SEI in 2022. SEI and Carnegie Mellon University researchers are working to accelerate assurance of large-scale systems through compositional techniques and automatic integration of verification results into certification claims. As software systems expand, so too do their attack surfaces. Vulnerability risk management has stretched to address the entire software supply chain, as has the SEI’s Acquisition Security Framework. The scale of AI vulnerabilities will continue to rise, and the SEI’s AI Security Incident Response Team (AISIRT) stands ready to lead the communal response. The federal workforce too must scale to overcome security threats, which is why the SEI develops automated malware reverse engineering tools and the DoD’s first certification for malware reverse engineers.

The old software paradigm said you could make a system either quickly or at scale. Now we must do both, and in every warfighting domain: land, air, sea, cyberspace, and space. The SEI and other defense federally funded research and development centers have been laying the groundwork, but much work, innovation, and creativity lie ahead.

Photo: NASA