In 2024, the United States made significant progress in the field of artificial intelligence (AI), including military planning and intelligence, especially with the breakthrough achieved in the integration of AI in the Joint All Domain Command and Control (CJADC2) system. These advancements surpass what unmanned drones, killer robots, or any standalone advanced weapon systems can achieve, helping the military maintain an edge throughout modern warfare.
This year, significant expansions have been seen in both the functionality and the number of personnel involved in the application of artificial intelligence in the military domain. The progress made in utilizing AI in the military field has turned the impossible into possible, comparable to the revolutionary impact unmanned systems had on the nature of warfare.
Despite appearing seemingly inconspicuous, these applications are crucial, as the ultimate lethal capability often does not necessarily involve direct killing weapons. The unification of Germany by Prussia 200 years ago was largely attributed to the outstanding leadership of its general staff. In World War II, Nazi Germany used radio communication to coordinate mobile warfare, while their opponent, France, was still using wired telephones and signal flags, giving Germany the opportunity to crush its enemy with a small number of armored units. Today, the Pentagon is striving to build an unprecedented intelligent system in just 16 months to replace its vast bureaucratic administrative structure to meet critical needs in global warfare. The Pentagon may have long realized that victory does not belong to those with the largest missiles, but to those who possess the most effective capability to manage extremely complex war systems.
Today, traditional methods involving documents, voice, and any other form of communication are no longer sufficient for grasping and managing rapidly changing events in the five domains of land, sea, air, space, and cyberspace. The need for rapid data, information, and command sharing among these five “domains” has compelled the U.S. Department of Defense to develop a Joint All Domain Command and Control system driven by artificial intelligence, abbreviated as CJADC-2 (Combined Joint All Domain Command & Control-2). Although CJADC-2 is still a new functional capability in development, many are still amazed by its complexity and doubt its actual use in combat situations. However, over the past year, its initial functionalities have transitioned from experimental to real-world operational.
In February of this year, the Deputy Secretary of Defense, Kathleen Hicks, announced the initial version of CJADC-2, which, in software industry jargon, is dubbed as the “minimum viable capability”.
The initial version represents the basic capability of integrating software applications, data integration, and cross-domain combat concepts designed to provide decision superiority to warfighters. In other words, it can be applied across all combat domains, enabling warfighters to have a “God’s eye view” and possess the advantage to defeat any global adversary when threatened or necessary. The focus of this version is on sharing messages among the eleven four-star combatant commands of the U.S. military, and further expansion can commence immediately.
Hicks stated that the minimum viable capability of CJADC-2 has now been achieved and is ready for deployment. She acknowledged the challenges involved but emphasized that, through the efforts of many teams, they have successfully delivered on time and on target. The first iteration has proven to be low-latency and reliable, as she described the software’s capabilities as “beautiful” in terms of what it can accomplish for weapon hardware.
She explained that investing in AI greatly enhances the speed, quality, and accuracy of commander decisions, providing them with a decisive advantage in deterring conflicts and winning battles.
One of the key functionalities that the military is most concerned about regarding CJADC-2 is firepower control. The U.S. military boasts some of the most advanced weapon systems, from the Air Force’s stealth fighter jets and bombers, to the Navy’s missile submarines and destroyers, and the Army’s new precision strike missiles, among others. Traditionally, determining which weapon to use against which target requires a significant amount of intelligence, time, personnel, and analysis. Therefore, adapting to rapidly changing battlefield conditions with the increasing intelligence provided by drones, satellites, and other sensors has made the intelligence acquisition, identification, classification, decision-making, and action process increasingly complex and burdensome for humans.
Much of CJADC-2’s work is focused on the “Joint Fires Network” (JFN). It will automatically provide decision recommendations on “who should shoot whom” to assist human commanders in making decisions and directing forces to complete missions. The prototype of the Joint Fires Network left a profound impression during the “Shield of the Pacific” wargame in June this year, with its “1.0 version” expected to be operational early next year.
The minimum viable capability of CJADC-2 and the Joint Fires Network take a top-down approach in viewing warfare from a global perspective, which will more or less change the U.S. military’s bottom-up and creative operational models. Therefore, the work of CJADC-2 touches on tactical aspects, requiring frontline units to directly share data rather than waiting for orders from higher command, necessitating the establishment of data connections between CJADC-2 and frontline units or individual soldiers. This has proven to be technically complex, even considered “impossible”, especially as the military deploys increasingly diverse types of complex sensors.
Recently, the Pentagon announced signing a three-year contract worth $100 million with the industrial company Anduril, aimed at extensive deployment of the company’s Lattice Mesh software suite in CJADC-2. This allows the processing unit in CJADC-2 to understand feedback from hundreds of different sensor systems without waiting for central processing to organize data. This enables CJADC-2 to establish command and control relationships between small units, mobile armored vehicles, tanks, artillery, and even individual soldiers.
One ongoing challenge in the development of the CJADC-2 program is that it does not face a single issue, but rather dozens of different applications from different contractors for various purposes. This has forced the Pentagon to depart from the traditional approach of creating large-scale projects operated by a single primary contractor, instead leading multiple main contractors within a single project. Due to the different suppliers often using incompatible standards and protocols, sharing data has become difficult or impossible. Therefore, the Pentagon hopes for CJADC-2 to function more like a Lego toy, where any component can be interconnected, replaced, or reassembled.
The Pentagon’s AI office introduced a new method in May that designed software systems and constructed contracts for these systems, called Project Maven, an AI intelligence analysis system that decouples different elements of projects, allowing units built by different contractors to work together seamlessly. This can almost be seen as a milestone in the development process of CJADC-2, where the integration of AI systems has made many previously thought “impossible” tasks possible and directly led to CJADC-2 being deployed in active combat exercises.
As part of the broader effort to mature CJADC-2, the Pentagon plans to demonstrate a new security framework during multinational exercises in 2025.
Recently, a spokesperson for the Joint Staff announced that the Pentagon plans to implement a new collaborative mission environment online to support the 2025 sea mission led by the UK, aiming to adopt a zero-trust and data-centric security functionality based on a joint framework. The spokesperson referred to it as multiple security and collaborative data services shared between global partners and hosted users.
Zero trust is a network security framework that assumes adversaries have already infiltrated IT networks, requiring continuous monitoring and verification as users and their devices move within the network.
During the 2025 sea mission, the U.S., UK, Canada, along with 13 other partner countries from Europe and Asia, will apply the zero trust and data-centric security capabilities tested during the 2024 “Olympus Project” under the Indo-Pacific mission network and collaborative partner environment.
Jim Knight, the head of the UK Olympus Project, stated that security and information sharing pose a contradiction. Security personnel come in and want to take some restrictive action. Zero trust and data-centric security are not only security measures but can also help achieve information sharing.
The key challenges hindering the CJADC-2 program’s development are being resolved through the evolution and involvement of artificial intelligence, and the partial capabilities of CJADC-2 are beginning to enter the practical validation phase, setting the stage for rapid development. The U.S. military’s culture has long prided itself on inspiring individual capabilities, and in the information age, networks may empower individuals with greater potential. CJADC-2 is the most effective tool to unleash this vast potential and will accelerate the widening gap between the U.S. and its adversaries in comprehensive battlefield capabilities.