On May 11, 2026, the Missile Defense Agency disclosed Project Maverick, an effort to demonstrate an interim counter-hypersonic intercept by fiscal year 2027 using offboard, multi-phenomenology sensor data fused through a tactical battle management system that cues an existing test interceptor against a hypersonic glide vehicle flying up the U.S. East Coast. MDA Director Lt. Gen. Heath Collins framed the demonstration as an opportunity to prove out capabilities across the kill chain ahead of the Glide Phase Interceptor's expected operational delivery in the early 2030s. The agency is seeking roughly $460 million in FY27 for Maverick, the Low-Cost Interceptor effort, and related advanced research, with budget documents folding Maverick under the broader Low-Cost Defeat initiative. The headline number is modest by missile-defense standards. The architectural commitment buried inside it is not.
The substantive content of Maverick is not the test vehicle. It is the explicit decision to design the engagement around engage-on-remote — a fire control concept in which the shooter does not need its own organic track on the target. Detection and tracking are produced by external sensors, fused by an off-platform battle manager, and handed forward as fire-control-quality data to a weapon that commits on someone else's picture. For hypersonic threats this is not a stylistic choice. Ground radars cannot see a 60-kilometer-altitude glide vehicle until terminal entry; ship and ground interceptors that wait for organic track will be inside the threat's last-mile maneuver envelope before they cue. The engagement window from initial space- or air-based detection to last-launch-opportunity collapses to tens of seconds. Closing that window requires that fusion, identification, and engagement assignment happen automatically, against tracks of varying quality from disparate phenomenologies, and that the launching unit treats the fused track as authoritative.
What Engage-on-Remote Actually Demands
Engage-on-remote sounds like a connectivity problem. It is not. The transport layer — Link 16, IBCS network, future low-latency space relays — is the easy part. The hard part is the fusion logic and the trust model. Multi-phenomenology means the inputs are heterogeneous: low-Earth-orbit infrared track from the Hypersonic and Ballistic Tracking Space Sensor layer; ground- or sea-based S-band and X-band radar returns at different angles, geometries, and revisit rates; potentially airborne infrared search-and-track contributions. Each sensor has its own noise floor, its own bias model, its own track confidence semantics. Fusing them into a single track that an interceptor's guidance computer will accept as truth requires real-time covariance handling, sensor weighting, and disagreement resolution that is, at the operational tempo Maverick is targeting, an AI-mediated decision. The launching battery commander is not going to inspect a sensor-fusion residuals plot at second 47 of a 75-second engagement.
That is where the verification problem enters. A tactical battle management system that resolves multi-sensor disagreements and recommends — or in extremis, commits — a hypersonic intercept is making a high-consequence decision under irreducible time pressure. Doctrine and policy require a human operator in the engagement loop, but the human's role compresses to authorization of a computed solution rather than construction of one. The assurance question becomes: how does the operator know the fusion was right? How does the post-engagement review board know whether a missed intercept was sensor noise, fusion error, interceptor failure, or threat maneuver outside the modeled envelope? Without instrumented, auditable, multi-model verification of the fusion and engagement-recommendation chain, the United States will field a counter-hypersonic capability whose successes and failures are equally opaque. That is not a sustainable operating posture for a strategic-stakes weapon.
Why the Interim Matters More Than the Endpoint
The Glide Phase Interceptor program — accelerated in April 2026 with a $475 million contract modification that pushed Northrop Grumman's award above $1.3 billion — is on a trajectory to deliver initial capability by the December 2029 statutory milestone and full operational capability by December 2032. GPI is the long-term answer: a purpose-built mid-course interceptor that engages hypersonic glide vehicles before terminal entry. Maverick exists because the threat does not wait for 2032. China has demonstrated the DF-17 and DF-27 and is fielding the CJ-1000 land-launched hypersonic scramjet; Russia continues to operate the Kinzhal and Avangard. The interim capability matters not because it will be elegant — it will not be — but because it forces the United States to operationalize the sensor-fusion and engage-on-remote architecture under realistic constraints in 2027 rather than 2032. The institutional learning required to operate a multi-sensor, multi-service, AI-enabled fire control chain at hypersonic tempos is the gating factor. Hardware can be procured on contract schedules; doctrine, tactics, and operator trust in automated fusion cannot.
For the defense industrial base, the signal in Maverick is that battle management software, sensor fusion middleware, and verification tooling have moved from program-of-record afterthoughts to the central enabler of counter-hypersonic intercept. The downstream procurement implications are concrete: open-architecture battle managers that can ingest new sensor feeds without recompetition, fusion algorithms with documented performance envelopes across phenomenologies, and assurance frameworks that allow a combatant commander to certify the chain end-to-end. Programs that deliver interceptors without delivering trustworthy fire control evidence will fail the operational test that Maverick is, in effect, structuring the entire counter-hypersonic enterprise to pass.
