At the Space and Missile Defense Symposium on August 13, Lt. Gen. Joseph Jarrard, deputy commander of NORTHCOM, gave a one-word answer to whether he could defend against a drone swarm launched from inside the United States: "no". He did not cite a shortage of interceptors, lasers, or microwave weapons, but a lack of detection: "We don't have the sensors, and depending on where that swarm is going to attack, it depends on whether we have any sensors at all".
This issue on lack of sensors inverts the existing counter-drone conversation that has run for nearly a decade. Countless industry demo videos, budget lines, and press coverage all lead with the effector, the system that visibly brings a drone down, which, as I have previously discussed, is biased and in some cases blatantly false. Lt. Gen. Jarrard is correct to put the sensor first because every C-UAS engagement has to begin with a track. An interceptor without cueing doesn't know where to look, and a jammer that can't see an RF signal doesn't know what to counter against. The Pentagon has requested roughly $21 billion for drones and counter-drone capability, with much of the counter-drone portion funding EW, kinetic interceptors, lasers, and high-power microwave systems. Yet, NORTHCOM's commander still identified seeing the threat as his chief C-UAS constraint.
Against a genuine UxS swarm, effector capacity is its own hard problem, since even a near-perfect track picture will fail if the defense cannot service forty targets before they arrive. DoW is not wasting money on interceptors and directed energy systems, but detection development needs equal priority simply by nature of sequencing. An effector without a track contributes little to the battle, while a track without an effector still buys warning time for everyone in the flight path. The Port Shuaiba drone attack showed that cost in March: no sensor detected the inbound drone, it arrived after an all-clear, and six soldiers died from a detection failure rather than an interceptor shortage.
The domestic version of this problem is much harder than the battlefield version because the threat hides inside ordinary civilian activity. An attacker can buy a COTS drone, vibe code some protocols for automated launch and waypoints, and launch it from a parking lot near the target, thus compressing the interval between detection and impact to minutes and burying the hostile track in routine drone traffic or environmental noise. NORTHCOM currently relies on a network of radars built to track aircraft and missiles, not the smaller Group 1 quads, and it is left borrowing law enforcement sensor networks it does not own to help close the gap.
The services are making real progress to work in this gap, with JIATF-401 issuing updated guidance in January extending installation commanders' defensive authority beyond the fence line. Exercises like Falcon Peak are working sensor integration across military, law enforcement, and infrastructure systems. The World Cup security operation detected and seized more than 600 drones across eleven cities this summer, proving that large-scale detection works against drones that emit and operators that law enforcement can locate. However, those two conditions are rapidly narrowing. A drone controlled through fiber-optic cable emits no RF signal, and AI-controlled drones are already navigating on their own without a call-back or heartbeat signal to a controller.
Beneath the sensor gap sits a legal one, as the authority permitting the military to detect and mitigate drones over domestic installations, 10 U.S.C. 130i, applies only to "covered facilities and assets" that the Secretary of War designates, tied to specific missions such as nuclear deterrence, missile defense, and national security space. The bipartisan COUNTER Act, introduced in December, proposes expanding the definition to all military bases with a secured perimeter, which means many bases lack coverage today. Portions of the underlying authority terminate on December 31, 2026 unless Congress acts. As it currently stands, a commander at a non-covered installation can face an unidentified drone with neither a sensor to track it nor any clear legal authority to engage it.
Ukraine's Operation Spiderweb used 117 FPV drones concealed in wooden containers on trucks to strike Russian airfields, damaging or destroying roughly a third of Russia's strategic missile-carrier fleet, an estimated $7 billion in losses from drones costing a few thousand dollars each. This same playbook could be used against an American installation: trucks park legally outside a base security fence; a launch produces no radio signature if the drones fly waypoints or use fiber optic spools, so the RF detection layer never trips. The first indication is visual, a gate guard or an aircrew seeing something small and fast, giving only seconds for any reaction before adversary drones strike parked aircraft or fuel storage depots. Attribution could take weeks, as it did at Langley in 2023, where no one ever publicly attributed the overflights at all. Every element of this sequence exists today, and the current C-UAS programs of record do not change the detection front end against non-emitting threats anytime soon.
Ukraine has fielded a solution that works for their airspace, cheaply, under harder conditions than ours. Its acoustic detection networks, Sky Fortress and Zvook, run more than 24,000 passive sensors built from commercial components at $400 to $1,000 per node, under $5 million for the entire network, feeding detections into a common operating picture within 12 seconds and cueing mobile gun teams; in one documented raid the network detected all 84 inbound drones and enabled the intercept of 80. Acoustic sensing is passive, so jammers cannot attack it, anti-radiation weapons have nothing to home on, and it detects fiber-controlled and autonomous drones because it never depended on their emissions. Lithuania is acquiring the system for deployment this year and Romania is testing it. Lt. Gen. Sean Gainey, then commanding Army Space and Missile Defense Command and previously the Pentagon's senior counter-drone officer, said in July 2024 that the US should integrate a low-cost acoustic network of this kind into its own air defenses. However, two years later no such network surrounds any American installation.
Part of the reason Lt. Gen. Gainey's recommendation went nowhere is that no one owns this mission. NORTHCOM commands homeland defense but owns almost none of the sensors or forces the mission requires. Each service funds its own installation defense out of operations money, one base at a time. The FAA controls the airspace any sensor network would monitor, and the state and local officers now earning counter-UAS certification answer to governors, not combatant commanders. A distributed sensing mesh across hundreds of installations is everyone's interface requirement with no program ownership, and capability gaps with that structure are there until someone assigns an owner and a budget line. We have seen this structure fail countless times before (like the Pentagon's Replicator Program).
The other primary reason for lack of adoption is a social one, and anyone who has watched the current Flock camera controversy is very familiar with it. Flock's license plate readers now sit in thousands of American communities, and the backlash is significant, with cities canceling contracts, state investigations into who accesses the data, and residents demanding the removal of cameras they never voted on. Those are cameras photographing license plates on public roads, and in some cases may be recording audio as well. A similar network of government microphones strung through American neighborhoods will read worse, no matter what the sensors actually process and however righteous or noble the purpose behind it may be.
The technical answer, that acoustic nodes classify rotor signatures against a threat library and discard everything that is not a drone track, will most certainly not survive discussion with Karen and Chad in a city council meeting. For most of the US, a Ukraine-style national mesh is a non-starter, and the architecture has to reflect that reality. Build perimeter-out on federal land and installations, extend to critical infrastructure where the owners consent, publish the data handling in plain language, and accept that American coverage will run thinner than Ukraine's, because Americans are not under nightly bombardment and will not trade privacy for protection against a threat most have never seen. A serious homeland architecture will pair a perimeter-out passive layer with radar tuned for small slow targets, and will tie military, law enforcement, and civilian sensors into one fused picture, positions link-agnostic effectors such as high-power microwave at fixed sites. It all must rest on a legal authority that covers every base and critical infrastructure with a perimeter.
The future divides on whether these decisions will happen before or after an incident. If we build these networks now, the numbers argue for themselves: the Air Force's Air Base Air Defense budget line alone quintupled to $836 million in FY26, while Ukraine built its entire national acoustic mesh for under $5 million, riding on existing infrastructure and giving the thousands of state and local officers now earning certification under the new counter-UAS authorities something to act on. If the decisions wait, the pattern repeats from recent history: an attack succeeds, authority and money surge in afterward, and the country builds the architecture reactively, facing intense political pressure, writing the rules in the months after a tragedy rather than the years before it. We know the threat will not wait and is not going to go away. AI-enabled autonomy is spreading to cheap mass-produced airframes, and quieter electric drones will eventually stress even acoustic detection, which argues for building the layered mesh while the current threat generation is still the one we need. Lt. Gen Jarrard accurately described the current situation this month... whether it describes 2028 depends on the decisions we make here and now.