What a $74 million drone defense assumes

This is a good report. It is short, unclassified, free, fact-checked, and an earlier draft was commented on by the author of the Congressional Research Service's report on Pentagon counter-drone programs. It also does something most counter-drone writing never does: it writes down a concrete architecture, with a count of every radar and interceptor, and prices it. That makes it the most useful public document on base drone defense I have read this year, and the reason it is useful is that every assumption is on the page. So let's read them.

What the $74 million buys

CBO built a benchmark: a square installation, 5 miles by 5 miles, about 16,000 acres, which it says would contain the important areas of most bases (command staff, family housing, barracks, key equipment). It then priced five options against that square. The most complete, Option 5, contains:

• 4 radars, one at each corner, about $10 million each. Radar is the longest-range sensor CBO considered, a little over 6 miles.
• 4 kinetic interceptor systems, colocated with the radars, about $4 million each, reaching about 9.5 miles. These are missiles or interceptor drones that physically hit the target.
• 12 radio-frequency (RF) signal detection and defeat systems around the perimeter, about $600,000 each. They detect a drone's control link out to about 2.5 miles and jam or spoof it out to about 1.5 miles.
• 16 handheld counter-drone devices, about $75,000 each, effective to about 550 yards. CBO describes these as a portable version of the RF detection and defeat system.
• 1 command-and-control system, about $1.25 million, to tie it together.

Procurement comes to $65.6 million and installation (power, towers, weather enclosures, data connections) to $8 million, for a total acquisition cost of $73.6 million. Spares plus management and support add $5 million a year. CBO then assumes each base is independent, so a hundred bases cost a hundred times as much: $7.36 billion to buy and $500 million a year to support. That is the number in the headlines.

The assumptions, in CBO's own words

The first is the one I would put on the cover. When CBO worked out how many of each system a base needs, it did so on paper, under perfect conditions:

When determining the number and types of C-sUAS systems that would be needed, the agency assumed ideal conditions for each system considered.

The report is candid elsewhere about what "ideal" leaves out. Its own table of detection limitations says radar returns from small drones may be mistaken for birds or swaying trees, and that not every drone talks to its operator over a radio link. It says ambient radio signals from broadcast stations and other transmitters shorten RF detection range, that hills and woods shorten sensor ranges generally, and that placing RF systems takes site surveys and ongoing monitoring of the local radio environment. None of that changes the equipment count, because the count was done for a flat, quiet, square base.

The second assumption is about people:

CBO's estimates do not include personnel costs to operate the systems because CBO expects that the services would use existing military forces.

CBO then prices the alternative in a footnote, and the number is bigger than the support budget it just excluded it from. A company of 66 soldiers, enough for three shifts of round-the-clock coverage, would cost about $6.6 million a year per site. The equipment's annual support is $5 million. If the troops are real, the people cost more than the upkeep.

The third is replacement:

Those procurement costs might have to be repeated every four or five years for the systems to remain effective, depending on how rapidly the threat evolves.

The headline $7.4 billion is a one-time purchase. CBO is explicit that in this field a purchase has a shelf life of roughly one presidential term.

The fourth is research. The Department of Defense spends over $300 million a year on counter-drone research, development, test and evaluation, and CBO leaves it out on the reasoning that DoD will fund it no matter how many bases get equipped. That is a defensible choice for a per-base cost, and it is also a reminder that the per-base cost is not the program cost.

And finally, scope. The report does not examine the legal authority of an installation commander to engage drones inside the United States, does not assess defending mobile forces, and says outright that comparing the cost of the drone with the cost of the defense is beyond its scope. Those are three of the questions people most often think a report like this is answering.

A rough ten-year picture, using only CBO's numbers

CBO does not publish a life-cycle figure, so here is the simplest one its own inputs allow. This is my arithmetic, not CBO's, and it is illustrative. Take one base over ten years, buy the Option 5 package twice (once now, once at year five), pay the $5 million a year in support, and add the $6.6 million a year company of soldiers in the case where the services cannot find existing troops:

• Equipment, bought twice: 2 × $73.6 million = $147.2 million
• Support, ten years: 10 × $5 million = $50 million
• Personnel, ten years, if added: 10 × $6.6 million = $66 million

That is about $263 million per base over a decade with added troops, or about $197 million without them, against a headline of $74 million. Neither figure includes the research budget or the site-specific extras CBO says an irregular or hilly base would need. I am not claiming this is the right life-cycle cost. I am pointing out that the published figure is the first year's purchase order, and that CBO itself says so if you read past the summary.

Where the capacity actually lives

This is the part I have not seen anyone point out, and it is what I would want a base commander to read first.

CBO's table of options lists the maximum number of simultaneous engagements for each one. Option 5 can engage 32 drones at once. Break that down by component and you get: 16 from the handheld devices, 12 from the perimeter RF systems, and 4 from the kinetic interceptors. Radar is a sensor and the command-and-control system coordinates, so neither adds an engagement.

So 28 of the 32 engagement slots come from radio-frequency defeat. And CBO is clear about the limit of that whole category:

RF signal defeat systems would be ineffective against sUASs that do not rely on a radio communication link.

The report also says, in its first chapter, that Ukraine and Russia can now fly some drones without detectable radio signals, either on autonomous waypoints or through a thin fiber-optic cable trailing the aircraft, and that these methods make drones harder for current systems to detect and influence. I have written about fiber-optic drones on this site several times, because they break the core assumption of electronic warfare: that the thing you want to stop is transmitting.

Put those two facts together and you get my reading of Option 5 against a drone with no radio link. Detection falls to the 4 radars, which CBO warns can confuse drones with birds. Defeat falls to the 4 kinetic interceptors. That is four simultaneous engagements, the same capacity CBO lists for Option 4, which is just the radars, the interceptors and the command system, at $59.2 million.

There is a cost story inside that as well. In the Option 5 package, the RF systems and handhelds cost $8.4 million of the $65.6 million procurement bill, about an eighth. The radars and interceptors cost $56 million. The cheap layers carry the capacity, and the expensive layers carry the resilience against drones that don't transmit. Against a radio-controlled drone, the design has depth. Against a fiber-optic one, it has four shots at a time.

To be fair to CBO, the report doesn't hide this. It states the RF limitation directly, calls for continued research against drones that don't use a radio, and adds a line about swarms that deserves to be quoted in full:

If the attack was large enough, even the systems in Option 5 could be overwhelmed.

My point is narrower. The capacity figure for the recommended option is dominated by one defeat mechanism, and the direction the threat is moving is the one that mechanism cannot follow.

Why this beat cares

My first career was counter-IED and electronic warfare. RF defeat is the trade I know best, and I believe in it. It is cheap, it scales, it is reusable, and against the drones most people can buy it works. CBO's pricing reflects that: an RF detect-and-defeat node costs about $600,000, against $10 million for one radar and $4 million for one interceptor system.

But every jammer I ever worked with had the same boundary written into its physics. It works on emitters. A threat that stops emitting doesn't get jammed, it gets ignored. On the road that was a command wire. Over a base it is a fiber-optic spool. The lesson from the road was never "jammers don't work." It was that you had to know which of your layers depended on the enemy's cooperation, and to count your capacity without them. That is the count I did above, and it is the one I would ask for before approving any layered design.

None of this means the Option 5 architecture is wrong. Layering is right, the report says so, and I agree. It means the "32 simultaneous engagements" figure should always be read together with the question of how many of them survive a drone with no radio.

What I could not confirm

How I read the report. The Congressional Budget Office website refuses automated requests, so I opened the report page and the 27-page PDF in a browser and extracted the full text there. I read all of it. It is the version reposted on 14 July 2026 with one correction, which changed the normal operating altitude listed for the largest drone class (Group 5) in Table 1-1 and doesn't affect anything above. I did not open the separate spreadsheet of underlying data.

The fiber-optic reading is mine. CBO does not run a scenario against drones without a radio link and does not publish engagement capacity by threat type. The 28-of-32 breakdown comes from adding up the components of Option 5 in CBO's own tables, and the claim that the RF and handheld slots would not work against such a drone rests on CBO's own statements that RF defeat is ineffective without a radio link and that the handheld is a portable RF system. A drone that still emits something, or a handheld that behaves differently from how CBO describes it, would change the count. I have also not assumed that a kinetic interceptor can reliably engage a fiber-optic drone. CBO does not address that, and neither do I.

The ten-year figure is illustrative. It uses only CBO's published inputs and simple multiplication. It assumes the full package is replaced once, which CBO describes as possible rather than certain, and it ignores inflation, research spending and site-specific costs.

Documents cited by CBO that I did not open: the December 2024 Department of Defense counter-unmanned-systems strategy fact sheet, the Congressional Research Service report R48477 that is CBO's source for the research budget, and the Inside Defense article behind the 135 vendor responses. Where I use those facts, I am relying on CBO's account of them.

Nothing here assesses any product or vendor, and nothing here describes how to defeat or evade any defensive system.

The signal

CBO's estimate is honest, and the headline built from it is accurate as far as it goes. What it measures is the purchase price of a layered defense for an idealised base, operated by troops who already exist, bought once. Its own text says the equipment may need replacing every four or five years, that adding soldiers would cost more per year than maintaining the equipment, and that its layers were sized for ideal conditions.

The most important number in the report isn't a price. It's the 32 in the simultaneous-engagements column, and the fact that 28 of those depend on the drone having a radio. When someone quotes you a drone defense by the number of targets it can handle at once, the right follow-up question is: how many of those still count against a drone that doesn't transmit?