Join War on the Rocks and gain access to content trusted by policymakers, military leaders, and strategic thinkers worldwide.
Anduril, the $30 billion defense startup, is on a roll. In May, it partnered with Meta to build virtual reality devices for the U.S. Army. In July, it won a $100 million contract to build the Army’s command-and-control software. On Aug. 5, 2025, it became the third supplier of rocket motors to the U.S. government. It is a leading contender to help build the “Golden Dome” missile-defense system. It has recently gone global, establishing offices in Seoul and Taipei.
The media is star-struck. The New Yorker’s Dexter Filkins, proclaiming that “the future of warfare is being invented in places like Ukraine,” endorsed the neo-defense tech dogma of building cheap, massed weapons to deter China. As evidence, he interviewed Anduril’s founder, Palmer Luckey, and its head of defense strategy, Christian Brose, author of The Kill Chain (the only book most Silicon Valley venture capitalists have ever read about warfare). A Wall Street Journal editorial went further: Nearly half the article’s text consists of quotes from Brose.
Anduril is a good company. So are many in the neo-defense tech ecosystem. But if suddenly, seemingly overnight, everyone shares nearly identical opinions about the future of warfare, some skepticism is in order.
First, be skeptical of claims that something cheaper is better. It’s possible, but it’s rare. Second, judge companies on incentive structures and business models, rather than altruistic or patriotic motives. The idea that “cheaper is better” derives from an over-indexing on the lessons of Ukraine, and committing to it fully will lock in business models that risk U.S. victory in the wars of the future.
The Prevailing Narrative
The five prime contractors — Lockheed Martin, Raytheon, Northrop Grumman, Boeing, and General Dynamics — have not enjoyed glowing press, and often for good reason. China has vastly outstripped the United States in its efficiency in producing warships, aircraft, missiles, and other weapons. Increasingly, China has approached the quality and sophistication of U.S. technology as well. While the primes have built extraordinary capabilities for the U.S. military, they have not kept pace with the times, notably in production rates and software. America’s industrial base is calcified.
A new array of companies promises to save America from these woes. These are the neo-defense tech companies, such as Anduril, Helsing, Saronic, Epirus, and Shield AI. For years, Silicon Valley turned up its nose at working for the U.S. government, partly for ideological reasons, and partly because of the byzantine government acquisition system. In the past five years, however, a backlash against the Valley’s reflexive disdain for national defense has grown. “Patriotic investing” — the promise that private capital can unlock innovation to serve American national interests — is hot.
Many of these neo-defense tech companies and their venture capital funders invoke the Russo–Ukrainian War to argue that “cheap mass” is essential in modern warfare. In this narrative, the Ukrainian nation has resisted the Russian onslaught by adopting drones and other asymmetric capabilities to reverse the “cost exchange ratio” in warfare, forcing Russia to use expensive items like aircraft, ships, and tanks against Ukraine’s inexpensive weapons. The U.S. Navy’s exchanges with the Houthis in the Red Sea also demonstrate this concept. In addition, the rapid depletion of NATO weapons stockpiles to support Ukraine has revealed the weakness of U.S. and allied manufacturing capacity.
According to this perspective, what is needed is an industrial renaissance to build cheap weapons in large quantities. Many neo-defense tech companies propose solutions predicated on this “cheap mass” concept, from ShieldAI’s aerial drones to the numerous companies hawking explosive surface torpedoes (the less fancy, but more accurate, term for most “small unmanned surface vessels”). An ancillary benefit of cheap mass, the patriotic investors and salespeople claim, is that neo-defense tech will create thousands of stable, good-paying American manufacturing jobs.
Looking Under the Hood
There are two problems with this narrative. First, cheaper isn’t always better. In fact, it isn’t even always cheaper.
Ukraine’s innovative employment of asymmetric capabilities has been backed by the exquisite air defense systems, intelligence support, and sophisticated missiles the United States and Europe have provided. While neo-defense tech company leaders, investors, and boosters frequently disparage aircraft carriers and F-35s (seemingly their only recurring examples of “bad” exquisite products, or the loss-exchange ratio), small, slow, short-range, non-stealthy drones are not a panacea. Indeed, U.S strikes on Iran needed stealthy B-2 bombers, submarine-launched cruise missiles, as well as fourth and fifth-generation aircraft to deal with enemy air defenses.
Moreover, a grinding artillery war of attrition on the European landmass between two militaries wrought with funding, personnel, and training problems, is not an appropriate analog for superpower war in the Indo-Pacific. Unmanned surface vessels, for instance, provided the Ukrainians with a strategic advantage only because they could be launched from shore to travel short distances, and because Russian incompetence in surface warfare and damage control provided easy targets. In the Indo-Pacific, large, manned warships will be needed to deploy unmanned surface vessels. And unmanned underwater vehicles can never compete with the range and firepower of America’s (expensive) nuclear-powered submarines. There is a real risk that the U.S. military over-indexes on lessons learned from the Ukrainian battlefield, to the detriment of deterring (and, if necessary, defeating) China.
Second, it is nearly impossible to get rich selling cheap hardware, even in large quantities. The patriotic motives in neo-defense tech press releases (deterrence, reindustrialization, etc.) are alluring, but they cannot change the basic economics. A company only makes money selling something cheap if it is consumed and constantly replaced. That requires a major war, and wagering the future of a company on a catastrophic world war is not a good business model. Traditional defense contractors keep production lines open in peacetime by selling abroad. But if the item being produced is cheap and readily constructed, then allies and partners can build their own equivalents — and they already are.
Short of a war, then, how will neo-defense tech profit from selling cheap hardware? Once a company has sold all the front-line widgets the soldiers need, and the military has stuffed all the warehouses with spares, what then?
This is where the new weapons differ from those of old, and it provides a clue to the true business model. The terms of the day are “software-defined” weapons and “collaborative autonomy” (e.g., drone swarms). The concept of cheap mass requires that weapons communicate with one another: A lot of small things, employed in tandem, can destroy a large, expensive enemy target. Drones, missiles, and unmanned surface vessels, to be effective at all, require a command-and-control system. This challenge increases as the number of weapons exceeds the number of human operators.
This future system-of-systems will require a single battle-management software, or at most a few software suites integrated with one another. For Anduril, that software is “Lattice”; for ShieldAI, it is “Hivemind”; for Helsing, it is “Altra”; for Saronic, it is “Echelon.” Even though theoretically, all of these software suites should be able integrate with one another — under the defense standardization program’s “modular open systems approach” — a “zoo of systems” is not scalable, nor especially profitable for individual companies. For many neo-defense tech companies, perhaps the true business model is to get the end-user hooked on cheap hardware, maybe even by selling it for minimal or no profit (which is easier for privately-held companies to do). After that, perhaps the company’s software becomes indispensable, providing revenue via permanent licensing and updates. The hardware may still provide revenue via maintenance, but the software adds another, more important, revenue layer. “Cheap,” therefore, isn’t necessarily cheap in the long term.
This does not mean the neo-defense tech offerings are bad or useless, or that it’s wrong to make money. But the business model is not “we’re going to give manufacturing jobs to American citizens” or even “we’re going to win wars.” Rather, it is: “We are going to license and update your software.”
Prudence Wins Wars, Too
If the hundreds of companies currently selling “cheap mass” are really selling software-as-a-service, this leads to a conundrum: they cannot all win this race. The military does not need dozens of competing command-and-control systems, nor can they all efficiently be knit together, even through modular open systems. The future is not going to be five primes replaced by 100 contractors or 1000 medium-sized businesses. It will probably just be six or seven primes.
It may be true that neo-defense technology companies are selling important products, but they are also selling a narrative. It is only fair to question a narrative written by those who stand to profit from it. One question is whether the weapons used in Ukraine today are really as useful to the United States as these newer companies say. More important, however, is whether buying those weapons locks the United States into a particular way of fighting. Being skeptical does not make one a Luddite. It makes one prudent, and prudence is no less a war-winning virtue than innovation.
Jonathan Panter is a Stanton nuclear security fellow at the Council on Foreign Relations, and a former U.S. Navy surface warfare officer. He holds a Ph.D. in political science from Columbia University.
Image: Midjourney
Correction: A previous version of this article mistakenly referred to Palmer Luckey as Anduril’s CEO.
