The operational record of the past three years has settled a long-running debate in defense circles: the electromagnetic spectrum is not a supporting domain in modern warfare — it is the decisive terrain. In Ukraine, Russian forces learned that unencrypted radio traffic and predictable GPS signatures made armored formations legible to drone operators within minutes of radio-on. In the Red Sea, Houthi targeting relied on commercial maritime AIS data and basic radar returns to cue UAV and cruise missile attacks against shipping corridors the Navy had treated as secure. In the Iran-Israel conflict, electronic deception and radar spoofing were employed operationally in ways that exposed fundamental gaps in legacy air defense signal processing. Across each of these theaters, the side that understood, controlled, and exploited the electromagnetic spectrum faster than its adversary achieved disproportionate tactical effect. The United States military has absorbed that lesson. The institutional response is now visible in budget lines, acquisition restructures, and field exercises — and it is centered on a shift from hardware-defined to software-defined, AI-enabled electronic warfare.
The Army's transformation is particularly concrete. In January 2026, General David Hodne signed a new Electromagnetic Spectrum Operations concept of operations that acknowledged a structural problem the service had deferred for years: EMSO capabilities were fragmented across different warfighting functions, built on hardware that could not be updated at software speed, and organized in a way that prevented meaningful integration of AI and machine learning for real-time decision support. The memo was followed by action. On January 27, CACI International was awarded a five-year task order valued at up to $250 million to provide integration and sustainment support for the Army's advanced EW and spectrum-related technologies under the Capability Program Executive for Intelligence and Spectrum Warfare. In February, the Army issued a broad request to industry for electromagnetic spectrum solutions under a potential indefinite-delivery, indefinite-quantity contract structure — signaling that the service intends to compete EW capabilities at commercial speed rather than through traditional program-of-record timescales. Prototype development for new modular EW systems is scheduled to begin in Q2 2026. The Army is also adding eighteen new electronic warfare company units across its divisions — the organic force structure required to operationalize a capability that has historically been concentrated at higher echelons.
From Static Signatures to Adaptive Sensing
The distinction between legacy EW and cognitive EW is not primarily a question of capability breadth — it is a question of how quickly a system can characterize a new threat and generate an effective response. Legacy EW systems operate from a library of known emitter signatures. They perform well against adversaries whose systems were collected, characterized, and encoded into that library. They fail against novel waveforms, frequency-hopping radios, spread-spectrum signals, and adaptive radar modes that were not anticipated at the time the threat library was built. That failure mode is not a software update problem — it is an architectural problem. Systems that require human analysts to identify a new signal, characterize it, generate a response waveform, and push an update through a logistics chain are operating on a timescale that is orders of magnitude slower than the adversary's ability to change their emission profile.
Cognitive EW solves this at the architecture level. AI and machine learning models trained on broad signal data can classify novel emitters in real time, identify anomalies that pattern-match to known threat categories even when the specific waveform is new, and generate candidate response waveforms without waiting for a human analyst to complete a characterization cycle. The Spectrum Situational Awareness System — one of the Army's new capability investments — extends this further by giving commanders real-time visualization of their own unit's electromagnetic signature, enabling signature management at the formation level rather than treating EW purely as an offensive capability. The BEAST+ system, deployed with 3rd Infantry Division units during Combined Resolve 25-02, demonstrated this integrated approach: detecting signals across the electromagnetic spectrum to identify enemy locations and potential jamming activity while providing line-of-sight awareness in support of ground maneuver. The Modular Electro-Magnetic Spectrum System, with $9.1 million requested in FY2026, carries this logic further — a platform-agnostic EW kit designed to provide both force protection and electromagnetic technical effects from any host vehicle. The modular architecture matters as much as the capability: systems that can be reconfigured and updated without physical hardware replacement are the only class of EW capable of staying current with an adversary who changes their electronic order of battle continuously.
The Edge Imperative in Contested Spectrum Environments
There is an architectural requirement that runs through every AI-enabled EW program the Army is now funding, and it maps directly to the problem that autonomous platform developers have confronted in parallel: the AI inference has to happen at the edge, on the platform, within the size weight and power constraints of a deployed system that has no guaranteed connectivity to cloud infrastructure. This is not a design preference — it is a physical reality imposed by the operational environment. An EW system operating in a degraded, denied, and intermittent communications environment cannot route its signal classification decisions through a cloud-hosted model and back. The latency is prohibitive. The connectivity is unreliable. The adversary's jamming may specifically target the communications link that would provide access to off-platform AI resources. An EW system that cannot sense, classify, and respond autonomously when disconnected from its network is not an EW system in any operationally meaningful sense — it is a sensor that requires a favorable electromagnetic environment to operate in an electromagnetically contested one.
The Army Research Laboratory's FREEDOM program — Fundamental Research for Electronic Warfare in Multi-Domain Operations — is funding the foundational science required to solve this at the hardware and algorithm level: RF sensing architectures, AI models optimized for embedded inference, and closed-loop EW techniques that can sustain autonomous operation across the contested spectrum. The commercial technology base is moving in the same direction. Software-defined radio platforms with sufficient processing headroom to run onboard ML inference are now available at unit costs that support high-volume fielding. The gap is not the hardware — it is the AI models trained specifically for the defense EW problem space, the integration frameworks that allow those models to operate within MOSA-compliant architectures, and the evaluation infrastructure required to validate autonomous EW behavior before operational deployment. A broader assessment of AI-enabled EW capabilities across Brigade Combat Teams is planned for June 2026, with full deployment targeted by 2028. That timeline is defined by the software and integration work remaining — which is where the real competition for spectrum superiority is now being decided.
