From Exquisite & Expensive to Cheap, Smart & Mass-Produced.
CORE THESIS
The Inverted Cost Equation The cost equation of modern warfare has inverted. A $2 million precision missile is losing strategic relevance when a $500 AI-guided drone can achieve comparable tactical effects at scale — and when losing 500 such drones costs the same as losing one legacy asset. This is not a marginal efficiency improvement. It is a structural transformation in how states threaten, deter, and destroy. |
Introduction: The Price of War Has Changed
The battlefield mathematics that defined twentieth-century Western military dominance are being rewritten in the mud and smoke of eastern Ukraine — and the lesson is reverberating from the Pentagon to Silicon Valley boardrooms to the factories of Mariupol and Dayton, Ohio. For decades, Western defence procurement rested on a foundational assumption: superior technology, however expensive, would always defeat mass. The Abrams tank, the Javelin missile, the Patriot battery — all were built on the principle that exquisite capability justified exquisite cost.
That assumption is now under severe pressure. Since February 2022, the Russo-Ukrainian war has produced a relentless empirical test of cost-versus-effect in modern warfare, and the results are shaping procurement decisions across every major military on earth. At the tactical level, a first-person-view (FPV) drone costing between $400 and $500 in components — assembled in a garage, guided by a 20-year-old with a headset — has proven capable of destroying a $2–$4 million main battle tank. FPV drones now account for an estimated 65 percent of all Russian armoured vehicle losses in Ukraine.
The implications are vast. As the Center for Strategic and International Studies (CSIS) concluded in late 2025: "low-cost swarms will likely prove cost-effective and create new options in offensive and defensive campaigns." The Pentagon has heard that message. So has Anduril. So has the venture capital industry, which poured more than $3 billion into defence-tech startups in 2024 alone — a figure that may be dwarfed by what follows.
Source: CSIS, 'Calculating the Cost-Effectiveness of Russia's Drone Strikes,' December 2025; Fortune, June 2025
I. The $2M Missile vs. the $500 Drone
Unit Economics, Production Throughput, and Acceptable Loss Rates
The starkest expression of this transformation is the Javelin comparison. The FGM-148 Javelin anti-tank guided missile — once celebrated as the saviour of Ukrainian independence in the frantic weeks of March 2022 — costs approximately $78,000 per round, with its Command Launch Unit adding a further $200,000 to the system price. It has a maximum effective range of under three miles, requires clear line-of-sight to acquire a target, and demands months of specialist training to employ effectively.
Set against it, an FPV drone costs $400–$500 in off-the-shelf commercial components. It can strike beyond visual line-of-sight at ranges exceeding 15–20 kilometres. It is controlled by a three-person crew operating from a minivan. And critically — it can be lost without strategic consequence. As analysis published in early 2026 noted, FPV drones provide "similar top-attack capabilities" to the Javelin while operating "at orders of magnitude cheaper" per engagement.
"The Javelin system is built around two primary components: a disposable missile tube and a reusable Command Launch Unit. The FPV drone eliminates both. At $500 a shot, losing it is a rounding error."
— Strategic intelligence analysis, 19FortyFive, March 2026
The production math compounds the advantage. Ukraine produced approximately 1.7 million drones in 2024. Its government target for 2025 was 4.5 million FPV units — a monthly run-rate of over 500,000 systems. At $500 each, that represents a $2.25 billion annual production bill: less than a single Arleigh Burke-class destroyer. Russia, meanwhile, has reduced the unit cost of its Shahed-type strike drone from approximately $200,000 in 2022 to around $70,000 by 2025, according to Ukrainian defence intelligence — a trajectory that illustrates how adversary industrial learning curves will challenge Western cost assumptions.
The cost asymmetry in air defence is even more alarming. CSIS analysis found that each Shahed drone costs an estimated $20,000–$50,000; the Western Patriot interceptors deployed to defeat them cost upwards of $3 million per missile. Russia can sustain the exchange indefinitely. NATO cannot.
Source: CSIS, December 2025; Inside Unmanned Systems, January 2026; Kyiv Post, February 2025
II. Attritable vs. Exquisite
Defining the Spectrum — From $5K MARCBOT to $27.5M Valkyrie
The defence community has developed a vocabulary for this new cost spectrum. "Attritable" systems are platforms cheap enough that commanders can accept their loss in combat without strategic concern. "Exquisite" systems are the opposite: high-capability, low-volume assets whose loss carries operational and political consequence.
The spectrum today runs from roughly $5,000 for a MARCBOT EOD robot — the blocky, tracked devices American soldiers have used since Iraq to investigate roadside bombs — to the $27.5 million XQ-58A Valkyrie Collaborative Combat Aircraft (CCA), Kratos's stealthy autonomous wingman designed to fly alongside manned F-35s. Between those poles sits the most contested and commercially dynamic segment of the market: the sub-$10,000 tactical drone.
The Attrition Spectrum: Key Price Points (2025–2026) ▸ $400–500: Ukraine-type FPV tactical strike drone (commercial COTS components) ▸ $5,000: MARCBOT EOD reconnaissance robot ▸ $20,000–50,000: Russian Shahed-136 one-way attack drone (CSIS estimate) ▸ $78,000: Javelin ATGM per round (+ $200K CLU) ▸ $100,000–200,000: Coyote Block 2/3 interceptor (U.S. Army) ▸ $500,000: Anduril Roadrunner autonomous interceptor (est.) ▸ $27.5 million: Kratos XQ-58A Valkyrie CCA ▸ $2 million+: Long-range precision cruise missile |
The military utility of this spectrum was made concrete by Operation Spiderweb in June 2025, when Ukrainian forces used purpose-built FPV drones — not standard $500 units but systems specifically designed for the mission — to penetrate deep into Russian airspace and damage or destroy over 40 high-value aircraft, including Tu-95MS nuclear-capable bombers. The operation took 18 months to plan and executed a strategic-level effect at tactical-level cost.
The lesson for Pentagon planners, as AUVSI President and CEO Michael Robbins noted in December 2025, is that the Drone Dominance Program represents a moment for the Pentagon to deploy "smart, tangible incentives to help American industry deliver capability to our warfighters with speed." The question is whether the institutional machinery of U.S. defence procurement can match the pace at which the threat environment is evolving.
Source: DefenseScoop, December 2025; Ukraine's Arms Monitor, December 2025
III. Supply Chain Realities
COTS Components, China's Dominance, and America's Onshoring Push
The FPV revolution rests on a supply chain vulnerability that keeps U.S. procurement officials awake at night. The small drone ecosystem is built predominantly on Chinese-manufactured components: flight controllers, motors, electronic speed controllers, FPV cameras, and video transmission systems. DJI, the Shenzhen-based drone manufacturer, dominates the global commercial drone market with an estimated 70–80 percent market share, and its components — or close derivatives — find their way into the FPV systems deployed on the front lines of Ukraine by both sides of the conflict.
This is not a theoretical risk. In January 2025, the FCC announced immediate implementation of Section 1709 of the FY25 National Defense Authorization Act, which added foreign-manufactured drones and key components to the Covered List — effectively banning their use in U.S. government applications. The message was unmistakable: Washington regards Chinese supply chain dominance in small UAS as a national security threat, not merely a trade concern.
The Pentagon's response has two principal industrial tracks. The first is the Drone Dominance Program (DDP), unveiled by Defence Secretary Pete Hegseth in December 2025, which plans to purchase more than 200,000 industry-made drones by 2027, with 30,000 units to be delivered by July 2026. The second is Arsenal-1, Anduril's hyperscale manufacturing facility near Rickenbacker International Airport in Ohio, into which the company has already invested nearly $1 billion of its own capital, with planned annual capacity in the tens of thousands of autonomous systems and projections of over 4,000 direct jobs.
Source: DefenseScoop, December 2025; FNEX/Anduril Series G Filing, June 2025
The onshoring challenge is formidable. As research from the Congressional Research Service noted in its September 2025 analysis of the Replicator initiative, "persistent technical issues, including systems' glitches and problems integrating systems with existing command structures, were slowing down the initiative's progress, where some systems were 'unreliable, or were so expensive or slow to be manufactured they couldn't be bought in the quantity needed.'" The same report found that the target of fielding multiple thousands of autonomous systems by August 2025 resulted in only hundreds actually delivered.
The gap between ambition and execution reflects a deeper structural problem: America's defence industrial base was not designed for mass production of inexpensive, rapidly-iterating consumer-technology-derived systems. It was designed — and remains optimised — for low-volume production of complex, exquisite platforms.
IV. 'The Factory Is the Weapon'
Red Cat, Anduril, and the New Logic of Defence Manufacturing
The phrase that has come to define the new defence-industrial moment is deceptively simple: the factory is the weapon. It originated in Silicon Valley defence circles as a rebuttal to the legacy contractor model, in which the weapon system itself — the aircraft, the missile, the vehicle — is the unit of competition. In the new model, the ability to produce weapons at scale, at speed, and at cost is the competitive advantage. It is the difference between a military that can sustain a drone attrition campaign and one that cannot.
Anduril Industries — founded in 2017 by Oculus creator Palmer Luckey and Founders Fund partner Trae Stephens — has become the most visible embodiment of this logic. Having raised $2.5 billion in its Series G round in June 2025 at a $30.5 billion valuation, the company is now in advanced negotiations for a further round of approximately $4 billion that would value it at $60 billion — nearly double its mid-2025 mark and placing it alongside legacy contractors such as Lockheed Martin and Northrop Grumman in raw capitalisation terms.
"As we continue working on building a company that has the capacity to scale into the largest problems for the national security community, we thought it was really important to shore up the balance sheet and make sure we have the ability to deploy capital into manufacturing and production."
— Trae Stephens, Anduril Executive Chairman, Bloomberg Television, June 2025
What distinguishes Anduril from a traditional prime contractor is its self-funding model. Rather than relying on government R&D contracts, it invests private capital in developing systems — autonomy software, counter-drone platforms, the Roadrunner reusable interceptor — and sells finished products to the government. Revenue doubled in 2024 to $1 billion. Arsenal-1 is the physical expression of the strategy: a factory that produces autonomous systems the way a tech company ships code updates.
Red Cat Holdings (NASDAQ: RCAT) represents the smaller-company tier of the same transformation. Through its Teal Drones subsidiary, Red Cat won the U.S. Army Short Range Reconnaissance (SRR) Program of Record in November 2024 with its Black Widow drone — an NDAA-compliant, U.S.-built ISR platform manufactured in Salt Lake City, Utah. The company reported full-year 2025 revenue of $38–41 million, representing 153 percent year-over-year growth, driven by a $35 million Army contract expansion and growing international orders from Asia-Pacific allies. Following the FCC's December 2025 enforcement of NDAA Section 1709 — restricting Chinese drone components — Red Cat's stock rose more than 60 percent in early 2026, a market signal about who wins when supply-chain nationalism becomes procurement policy.
The Breaking Defense reporting team observed in December 2025 that the Pentagon's renamed Replicator initiative — now called the Defence Autonomous Working Group (DAWG) — is conducting wargames and expanding its focus to "larger, longer-ranged attack drones" for the Pacific theatre. Pentagon CTO Emil Michael, speaking at the Reagan Forum alongside Admiral Sam Paparo, framed a distinction the entire industry is navigating: Drone Dominance addresses smaller FPV-type systems for small-unit infantry use, while DAWG tackles the longer-range, larger platforms suited to the vast distances of a potential Indo-Pacific conflict.
Source: Breaking Defense, December 2025; DefenseScoop, December 2025; Red Cat Holdings preliminary revenue report, March 2026
V. Venture Capital and the Defence-Tech Boom
$3B+ in Global Defence VC (2024–2025) and the Anduril Trajectory
The capital markets have registered the shift. After decades in which Silicon Valley venture capital treated defence as toxic — or at minimum, off-brand — a structural re-engagement began around 2021 and accelerated sharply through 2024 and 2025. Crunchbase data shows $3 billion invested in defence-tech startups in 2024, up 11 percent from the prior year. Anduril's $2.5 billion Series G round alone — with its $1 billion lead cheque from Founders Fund described by the firm as the "largest check" it has ever written — crystallises how the sector has graduated from niche to core.
The investment thesis is straightforward. Geopolitical risk is elevated. NATO allies are rebuilding. The Pentagon has publicly committed to mass procurement of attritable autonomous systems. Software-defined platforms iterate faster than hardware-only competitors. And the Total Addressable Market is immense: global defence spending exceeded $2.2 trillion in 2024, with autonomous systems representing the fastest-growing segment.
Key Investment Milestones: Defence-Tech VC, 2024–2026 ▸ 2024: $3B+ invested in global defence-tech startups (Crunchbase) ▸ June 2025: Anduril Series G raises $2.5B at $30.5B valuation, led by Founders Fund ($1B — firm's largest ever cheque) ▸ 2024: Anduril revenue doubles to $1 billion ▸ March 2026: Anduril in advanced talks for ~$4B round at $60B valuation (Axios/Bloomberg) ▸ Q4 2025: Red Cat Holdings posts 1,842% year-over-year revenue growth ▸ 2025: Pentagon Drone Dominance Program targets 200,000 drones by 2027 at $1B+ projected spend |
The Founders Fund rationale — articulated by Trae Stephens and reflecting Peter Thiel's long-standing argument that Western democracies have systematically underinvested in hard-technology defence — is that defence-tech companies following Anduril's model can compound faster than legacy primes because they control their own R&D and are not dependent on government contract funding cycles.
There are sceptical voices. The rapid escalation in Anduril's valuation — from $13 billion in August 2024 to $30.5 billion in June 2025 to a reported $60 billion target in March 2026 — has prompted questions about whether private market pricing reflects genuine revenue multiples or strategic enthusiasm. Arsenal-1's production lines are not yet fully operational; Pentagon officials have set July 2026 as the expected date for the factory to "go hot." The venture investment is, at this stage, a bet on execution as much as proven performance.
VI. 2026 Updates: The Cautionary Tales
Replicator's Mixed Results and the RCV as a Cautionary Programme
The most instructive recent case studies for the economics of attrition cut in both directions: one illustrating what mass-production ambition looks like when it collides with institutional procurement realities; the other demonstrating how software costs can hollow out the hardware savings that justified a programme in the first place.
Replicator: Ambition Meets Acquisition
The Replicator initiative, unveiled by Deputy Defence Secretary Kathleen Hicks in August 2023, aimed to field "multiple thousands" of all-domain attritable autonomous systems within 18 months — a timeline that, as Bill Greenwalt testified to the House Armed Services Committee, "the Pentagon's acquisition system is simply not capable of acting on." The programme received a planned $1 billion across FY2024 and FY2025. What it delivered, according to the Congressional Research Service's September 2025 assessment, was hundreds — not thousands — of systems by the August 2025 target date.
The failure was not primarily technological. It was structural: procurement bureaucracy, software integration challenges, cost overruns on individual units, and the absence of a dedicated budget line all combined to slow delivery. The Switchblade 600 kamikaze drone — one of Replicator's headline platforms — costs far more than the FPV equivalents being produced in Ukrainian garages. The programme has since been rebranded as the Defence Autonomous Working Group under the Trump administration, with a pivot toward larger, longer-range systems for Pacific scenarios and a separate Drone Dominance track for small FPV-type units.
The RCV: When Software Costs Outrun Hardware Savings
The Army Robotic Combat Vehicle programme offers perhaps the starkest illustration of the second failure mode. After years of development across light, medium, and heavy variants, the Army cancelled the RCV programme in May 2025 as part of an 8 percent budget reduction exercise — despite a competition winner (Textron Systems's Ripsaw 3) having already been identified. The stated reason from Programme Executive Officer Major General Glenn Dean was blunt: "We don't want to downselect just to one vendor and pay almost $3 million per copy."
Source: Breaking Defense, May 2025 and August 2025; CRS Report on RCV Program, May 2025
The RCV's difficulties illustrate a pattern that is becoming depressingly familiar in ground autonomy: a hardware platform whose unit economics are manageable, but whose software integration — autonomous navigation in off-road environments, electronic warfare resilience, interoperability with command networks — generates costs that dwarf the physical chassis. The Army's own Robotic Technology Kernel (RTK) autonomy package drew criticism from the Senate Armed Services Committee, which noted that "multiple commercial products" offering "advanced ground autonomy" already existed and urged the Army to "reexamine its funding decisions." In August 2025, the Army announced a new competition under a new name, targeting a price cap of $650,000 per unit — less than a quarter of the RCV's projected cost.
The Cautionary Tale: RCV vs. the $500 Drone The RCV programme spent years and hundreds of millions of dollars attempting to field an unmanned ground combat vehicle at ~$3 million per unit, only to be cancelled. In the same period, Ukrainian manufacturers scaled production of $500 FPV drones to 1.7 million units per year. The lesson is not that ground robotics are impossible — it is that software-defined capability, rapid commercial iteration, and tolerance for acceptable loss rates are prerequisites for the new economics of attrition warfare. |
VII. The Procurement Official's Dilemma
There is a human dimension to this transformation that aggregate data obscures. Somewhere in the Pentagon's Programme Executive Office structure sits an official who, within a single fiscal year, has approved both a $2 million Javelin purchase order and a $500 drone procurement. The paperwork looks almost identical. The compliance requirements are similar. The time to approval may actually be longer for the cheaper system, because attritable drones do not fit neatly into the acquisition category frameworks built for legacy platforms.
That official knows, from reading the Ukraine lessons-learned briefs, that the drone will likely achieve comparable tactical effect to the missile in a high-intensity peer conflict. They also know that their system of record, their career progression, and their institutional relationships are built around the exquisite platform. The economic logic of attrition is clear; the institutional logic of transformation is not.
This is the central tension that the Factory-as-Weapon model must resolve — not in laboratories or strategy documents, but in procurement offices, budget cycles, and the quiet decisions made by officials whose names never appear in the trade press. The economics have changed. The question is whether the institution will.
STORY ANGLE Interview a procurement official who has approved both a $2M Javelin purchase and a $500 drone purchase in the same fiscal year. Ask them to walk through each approval process — the paperwork, the timeline, the stakeholders, the compliance requirements. The disparity between the bureaucratic burden for a cheap attritable system versus a legacy precision munition is itself a story about why the revolution in drone warfare is happening faster in Ukraine than in Ohio. |
Key Sources & Expert References
Center for Strategic and International Studies (CSIS): 'Calculating the Cost-Effectiveness of Russia's Drone Strikes,' December 2025 — csis.org
DefenseScoop: 'Pentagon Unveils Drone Dominance Program with Gauntlets,' December 2025 — defensescoop.com
Breaking Defense: 'It's Alive: Replicator Lives On as DAWG,' December 2025 — breakingdefense.com
Breaking Defense: 'Army to Cancel Robotic Combat Vehicle Award,' May 2025 — breakingdefense.com
Defense News: 'Pentagon Says $1 Billion Planned for Replicator,' March 2024 — defensenews.com
Inside Unmanned Systems: '2025 Proved the Case for Drone Defense,' January 2026 — insideunmannedsystems.com
AUVSI (Michael Robbins, President & CEO): Statement on Drone Dominance Program, December 2025 — auvsi.org
Congressional Research Service: Replicator Initiative Report (IF12611) and RCV Program Report (IF11876), 2025
Fortune/CNBC/Axios: Anduril Industries Series G coverage, June 2025; Axios on $60B round, March 2026
Red Cat Holdings (RCAT): Q3 FY2025 Earnings and Preliminary FY2025 Revenue Guidance, November 2025–March 2026
Leave a comment