A few days ago, Palmer Luckey, the founder of Anduril Industries and a leading voice in the Silicon Valley defense-tech ecosystem, delivered a speech at National Taiwan University in Taipei.1 Framed as a call to action for Taiwan’s next generation of engineers and technologists, the speech urged students to apply their talents to national defense and help build a high-tech deterrent against the growing threat from China. But beyond its rhetorical appeal to patriotism and innovation, Luckey’s remarks recycled a familiar set of assumptions that have become gospel among defense-focused venture capital firms and companies spawning in Silicon Valley and making their way east to Washington to sell their products. These firms, flush with cash and influence, consistently misinterpret the nature of military power, how wars are actually won, and why military technology evolves the way it does. Luckey’s speech exemplifies the strategic naïveté that emerges when technologists mistake tactical disruption for strategic transformation.
Over the past year, nearly every major American media outlet has published a long-form article touching on the “drone and AI revolution” in warfare.2 From cable news segments to longform features in major newspapers, the narrative is remarkably consistent: low-cost drones and autonomous systems are transforming the modern battlefield and heralding a new era of war, one that the United States is aloof from.3
Dexter Filkins in The New Yorker writes, “A growing consensus of defense experts holds that the United States is dangerously unprepared for the conflicts it might face.”4 This techno-optimism (and fear) has also permeated the discourse of the foreign policy elite and government. A growing number of opinion pieces and essays in flagship publications like Foreign Policy, Foreign Affairs, and The National Interest have echoed and amplified this framing.5 An article in Foreign Affairs asserts that “the United States has largely missed this revolution in military technology” in reference to the rapid evolution and adoption of drone warfare.6 Secretary of Defense Pete Hegseth pledged in a memo that the DOD would be “unleashing the combined potential of American manufacturing and warfighter ingenuity” to increase production and integration of drones.7 Furthermore, the Army has stated it would like to have upwards of 90% of its aircraft be unmanned in the coming years. 8
The government and these outlets increasingly portray drones, machine learning, and algorithmic targeting as not just tactical tools but strategic game-changers—capable of offsetting traditional military disadvantages and reshaping how nations deter and fight wars. The result is a growing intellectual consensus, or at the least an evident enthusiasm, that elevates emerging technologies as the decisive factor in 21st-century conflict, often without sufficient attention to the broader institutional, logistical, and political realities that shape war’s outcomes.
However, all of these pieces and commentaries either disregard or fundamentally misunderstand why the United States allocates military spending the way it does, and more crucially, how its force structure reflects deep-rooted institutional, strategic, and political choices, rather than mere technological backwardness. The implication that Washington is simply asleep at the wheel, failing to invest in cheap drones and artificial intelligence because it lacks vision or agility, reveals a profound misreading of both military history and defense policy.
One of the most common critiques of U.S. defense spending, and a constant feature in these pieces on drones and AI, is that the military and policymakers prioritize quality over quantity, purchasing a small number of exquisite conventional weapons, such as jets and aircraft carriers, rather than investing in mass production. However, the dirty little secret is that the United States actually spends relatively little on acquiring new weapons in the first place. In 2024, only $172 billion, roughly 20% of the total defense budget, was allocated for new equipment.9 That figure is simply inadequate to modernize aging aircraft squadrons, recapitalize the Navy’s shrinking fleet, replenish critical munitions stockpiles, and invest seriously in emerging technologies like drones and autonomous systems.10
This shortfall is not just a budgeting issue; it’s also a political one. Defense spending is a political hand grenade, and the level of funding actually required to build the kind of mass many now envision for sustaining a prolonged high-intensity conflict is politically and fiscally daunting. In that vacuum, defense VC firms have found a willing audience. They promise to do more with less, to deliver high-tech capabilities at scale within existing budgetary constraints. It’s a compelling pitch: cheap, fast, and disruptive. But it’s also a convenient illusion. These firms thrive on the premise that innovation alone can overcome structural limits, when in reality, producing meaningful combat power at scale requires far more than clever engineering and lean startups. It requires strategic choices, industrial policy, and political will—none of which can be coded into a prototype.
Moreover, this techno-centric narrative overlooks a critical reality: that militaries do not win wars merely by acquiring new gadgets, but by developing the institutions, training regimens, logistical networks, and doctrinal concepts that enable those tools to be used effectively at scale. Technology does not exist in a vacuum; it must be embedded within a structure that can harness it to achieve strategic effects. The belief that the United States can leapfrog its adversaries through agile startups and disruptive prototypes reflects the same flawed logic that plagued earlier revolutions in military affairs. Drones and AI systems may be tactically sound, and in many cases, they are; however, they do not, by themselves, answer the questions of manpower, sustainment, strategic coherence, or political will. In elevating technology to the status of strategy, these commentators and entrepreneurs mistake innovation for “warfighting” and “disruption” for victory.
It’s also hard to take firms like Anduril and Palantir seriously when their most hyped contributions to defense innovation often amount to little more than rebranding existing technologies with a Silicon Valley gloss. Much of what they market as “disruptive” is, in fact, derivative—iterations of capabilities that have existed for years, now dressed up with buzzwords like “AI-enabled” or “autonomous.” Take, for instance, the Coyote Block 2; a proven component of the U.S. Army’s Low, Slow, Small Unmanned Aerial System Integrated Defeat System (LIDS). This system has been in service for some time, performing a straightforward yet essential mission: intercepting and destroying small drones.11 Anduril’s recent answer to this, the “Roadrunner,” is billed as a cutting-edge loitering interceptor.12 But in practice, it replicates the same core function with marginal enhancements, repackaged in a sleeker design and infused with branding language that flatters venture capital expectations more than it reflects operational novelty.
Changes in military technology occur for a wide range of reasons, including strategic ambitions, tactical innovations, institutional culture, and political ideologies.13 Yet to understand how and why military technologies develop correctly, it is equally important to distinguish between revolution and evolution in military change. These terms are often used interchangeably, but they point to fundamentally different processes.
A military revolution implies a systemic, discontinuous break with prior modes of warfare. It suggests that new technologies, or new ways of using them, trigger dramatic shifts not only in tactics or equipment, but in the structure of armies, the scale of conflict, and the political and social systems that support them.14 Revolutions in military affairs (RMAs) are said to alter the character of war so profoundly that they redefine the very conditions of strategic success. Historical examples often cited include the rise of gunpowder weapons and trace italienne fortifications in early modern Europe, the Napoleonic mass army, or the development of nuclear weapons and ballistic missile doctrine in the twentieth century.
By contrast, military evolution describes a slower, incremental process in which new technologies are adopted and adapted to enhance existing military structures rather than transform them. Evolutionary changes might improve range, speed, accuracy, or survivability, but they do so within an established paradigm of force employment. For example, improvements in tank armor, aircraft avionics, or battlefield communications systems may all enhance effectiveness, but they do not fundamentally alter the role those systems play within a combined arms framework. In this sense, evolutionary change often strengthens or augments existing capabilities rather than displacing them. This is what we are seeing with drones and AI in its current stage.
If there is one dependent variable that shapes the trajectory of a military revolution or evolution, it is constraint. Military innovation is rarely a product of pure technological opportunity or creative genius in isolation. Instead, it is most often driven by pressing limitations, such as shortages of manpower or material, logistical bottlenecks, battlefield attrition, or the inability to achieve desired strategic outcomes using existing means. Whether on the battlefield or behind the lines, in supply depots, factories, and war ministries, constraints compel adaptation.
Consider the German development of the Sturmgewehr 44 (StG 44) during World War II. German forces operating on the Eastern Front faced overwhelming Soviet numbers and increasingly close-range engagements. Traditional bolt-action rifles lacked sufficient fire rate, and submachine guns lacked range and versatility. The result was an intermediate cartridge rifle, explicitly designed to bridge this tactical gap —a compromise solution born of battlefield necessity rather than long-term doctrinal planning.15 Similarly, the British adoption of radar in the 1930s was less a product of technological foresight than of dire strategic constraint. Britain could not match the Luftwaffe bomber force plane-for-plane, so it built a defensive system, the Chain Home radar network, that allowed fewer fighters to be vectored more effectively.16
Going further back in history, we will find similar patterns. The American Civil War offers another example: the Union blockade and Confederate material shortages forced the South to experiment with innovations like the CSS Hunley, an early submarine. Far from a symbol of technological superiority, it was an act of desperation aimed at breaking a strategic chokehold.17 The same dynamic played out in World War I, when the stalemate of trench warfare forced the British to develop the tank—not out of a desire for armored warfare per se, but because neither massed infantry assaults nor artillery alone could break the deadlock of the Western Front.18
Modern examples follow the same pattern, but nonetheless, Palmer states:
Just a few weeks after Russian tanks rolled into Ukraine in 2022, I went to the front lines. I’m not a soldier. I’m not used to setting foot in a war zone. I was there to train Ukrainian soldiers on advanced military technology that I had developed. I wanted to understand the future of war, in practice.
And what I saw was remarkable.
Ukraine wielded technology in a devastating fashion against the enemy. In doing so, they’ve upended decades of conventional wisdom in how wars are fought and won. Today, victory on the battlefield does not depend on who has the shiniest, most capable weapons system. Success relies on the ability to apply new technologies in the largest numbers.
With drones costing just a few thousand dollars each, a handful of Ukrainian pilots remotely carpeted airstrips with explosives thousands of miles into Russia.19
This statement more or less skips the first 20 months of the war, which was a conventional slugfest dominated by Anti-Tank Guided Missiles (ATGMs), tanks, landmines, and artillery, lots of it. Both sides now lean so heavily on drones, not because they represent the peak of modern military innovation, but because both sides are critically short on traditional combined arms platforms. Artillery, armor, and manned aircraft are either depleted, vulnerable, or logistically constrained. Neither military currently possesses the operational depth or air superiority required to conduct large-scale, coordinated strike warfare from the air, let alone integrate it effectively with maneuver forces on the ground. In this context, drones are not revolutionary tools of a new era, but rather compensatory measures that fill the vacuum left by the erosion of conventional capabilities. Constraint, once again, drives adaptation.
Palmer also completely mischaracterizes the situation in Taiwan. He states
I want you to close your eyes, and imagine a scenario: In 2029, Xi Jinping orders the invasion of Taiwan. But after years of preparation, with support from strong allies and powerful partners, Taiwan is ready. Thousands of AI-powered drones spring toward the incoming Chinese fleet. Autonomous submarine systems and surface craft emerge from the sea to protect the island. Mass-producible missiles crowd the skies over Taiwan, stopping hundreds of Chinese fighter jets. The day is won. The military that the Chinese Communist Party spent decades to develop for that singular purpose—invading Taiwan—is destroyed for a generation, and Xi’s so-called “Chinese Dream” is dead with it.20
Luckey’s imagined scenario, where Taiwan defeats a Chinese invasion in 2029 outright with swarms of AI-powered drones (cruise missiles), autonomous submarines (torpedoes), and mass-producible missiles (more cruise missiles), is an excellent piece of storytelling. But as a strategic vision, it is hilariously flawed. It reflects a deep misunderstanding of how wars are actually fought and won, and worse, it risks misleading both policymakers and the public into believing that technological innovation alone can substitute for statecraft, alliances, and institutional military power.
First and foremost, the scenario overstates what technology can achieve in isolation. Autonomous drones and smart munitions may enhance Taiwan’s defenses at the margins, but they cannot replace the logistical infrastructure, trained personnel, and political coordination required to fight and sustain a major war. Building, integrating, and maintaining these systems at scale within four years is simply not realistic under current industrial and institutional constraints. Taiwan cannot outproduce or out-develop the People’s Liberation Army through startup culture or venture capital.
Moreover, the scenario treats the Chinese military as if it would passively walk into a technological ambush. In reality, the PLA is a dynamic and adaptive force that is rapidly developing its own drone systems, countermeasures, jammers, and cyber capabilities.21 A real invasion would likely be preceded by extensive cyber and electronic warfare, sabotage, and coercive pressure aimed at weakening Taiwan from within.22 Any defense plan that assumes a clean, one-sided technological victory ignores the complexity and brutality of high-intensity conflict.23
Perhaps most dangerously, Luckey’s narrative misrepresents the nature of deterrence. It assumes Xi Jinping would simply cancel his war plans after reviewing a war game where Taiwan wins. But deterrence is not based on enemy simulation models but rooted in credible alliances, demonstrated political will, and the threat of escalation beyond the battlefield.24 Taiwan’s most powerful deterrent is not a drone fleet; it is the likelihood that a Chinese invasion would trigger a devastating regional or global response. That deterrent is political, not technological.
Ultimately, this type of futurism does more harm than good. It distracts from the complex, slow, and unglamorous work of building readiness: investing in conventional defense, deepening alliance structures, training forces, and preparing the population for resilience. Drones and AI may be part of the equation, but they are not the solution. War is not an app. And Taiwan’s defense cannot rest on a fantasy of software-powered salvation.
If you’ve read so far, you’d probably assume the author hates drones and thinks they’re useless. On the contrary, drones are doing an immense amount of work on today’s battlefields. They are reshaping reconnaissance, improving target acquisition, enabling precision strikes at lower cost, and complicating the operational picture for even the most advanced militaries.25 Their ubiquity in Ukraine, the South Caucasus, and other recent conflicts is not in dispute; they have significantly altered the tempo and character of tactical engagements.
The fundamental question, however, is not whether drones are helpful; it’s how to integrate them effectively within a broader, combined-arms framework.26 In other words, how can drones enhance rather than substitute for traditional capabilities like armor, artillery, infantry, and airpower?27 How can they support maneuver rather than merely serve as tools of attritional harassment? How do they enable coordination across domains: land, air, sea, and cyber, rather than exist as isolated capabilities? A battalion commander in the Ukrainian Army stated to me a few weeks ago that even in the Summer of 2025, the single most decisive factor in his unit’s ability to hold and take ground is the number of infantry.28 Not drones, not technology, just dudes with a rifle in a dugout.29
Furthermore, he offered a pointed caution against drawing simplistic lessons from Ukraine’s changes in their war. “If we fought like Americans and had the resources of Americans, we would have been in Rostov a while ago,” he quipped, half-joking but entirely serious.30 If you believe we should build the exact army Ukraine has to fight the exact kind of war Ukraine is fighting, then you should also expect the exact outcome Ukraine is experiencing: high casualties, enormous material strain, and incremental territorial losses. Designing a force structure around another country’s wartime improvisations means inheriting not only their tactics but also their limitations and consequences. The heavy reliance on drones, loitering munitions, and decentralized tactics is not a doctrinal preference; rather, it is a response to necessity, shaped by limited airpower, strained logistics, and a fight for national survival.
Trying to universalize Ukraine’s battlefield innovations without understanding the strategic context risks mistaking improvisation for principle. What works for a mid-sized, embattled country fighting on home terrain with limited resources may not translate to a global military like the United States, which possesses expeditionary capabilities, alliance networks, and different political objectives. As the commander implied, emulating Ukraine’s tactics without its constraints would be like learning to race with a crutch. The real lesson is not to adopt Ukraine’s tools in a vacuum, but to appreciate how those tools are integrated within a broader system stretched to its limits.
Even on the Russian side, where Moscow has arguably come closer than any other state to fielding a drone “swarm” of Shahed’s in a sustained military campaign, the results have been far more conventional and far less transformative than the hype suggests.31 Russia has deployed hundreds of drones in coordinated strikes for months now, often alongside cruise missiles, in what is effectively a modernized version of a traditional strategic bombing campaign. These attacks have targeted mainly civilian infrastructure, energy grids, and urban centers, with the apparent goal of demoralizing the population and straining Ukraine’s air defenses.
Yet despite their volume and persistence, these drone strikes have done little, if anything delivering decisive results. They have not broken the Ukrainian will, nor have they significantly altered the war's trajectory. Instead, they echo the long history of strategic bombing efforts: operationally expensive, tactically disruptive, but strategically inconclusive.32 What has been labeled a “swarm” amounts, in practice, to a familiar concept with updated tools, a repackaged version of attritional air warfare, not a revolutionary new form of conflict. The reality is that saturation drone attacks, while visually striking and symbolically potent, have yet to produce the kind of battlefield dominance or strategic breakthrough their advocates claim.
Furthermore, the past few months have offered a vivid reminder of what high-end conventional weapons can still achieve when wielded with precision and coordination. The Israeli air campaign over Iran and the accompanying B-2 stealth bomber strikes on nuclear infrastructure showcased the enduring relevance of advanced manned aircraft, standoff munitions, and traditional airpower doctrine.33 Far from being eclipsed by drones or autonomous systems, these operations demonstrated how legacy platforms, when integrated with superior intelligence, planning, and logistics, can produce results at scale and over distance.
The Israeli strikes leveraged a mix of real-time targeting data, long-range precision munitions, and hardened platform survivability to penetrate what was supposed to be one of the most heavily defended airspaces in the world. Similarly, the B-2 Spirit bombers—stealth aircraft designed for deep-penetration strategic attacks—executed a complex transcontinental mission to neutralize hardened Iranian nuclear targets.34 These were not experiments or demonstrations; they were deliberate applications of conventional firepower to shape the strategic environment. In both cases, the message was clear: despite the allure of low-cost drones, which supplemented Israeli air superiority, there is still no substitute for the combination of stealth, speed, payload, and professionalized command and control when seeking to produce decisive outcomes against hardened or deeply buried targets.
The risk is not that drones are ineffective, but that their growing success in constrained environments leads to doctrinal laziness. Militaries may be tempted to treat drones as silver bullets, rather than embedding them into the complex web of logistics, communications, training, and strategy that underpins real operational effectiveness. Proper integration means using drones not just to strike or surveil, but to open opportunities for maneuver, preserve force survivability, and sustain momentum, all in coordination with human-led formations and traditional platforms. Drones alone can’t win wars. However, when appropriately nested within a combined arms approach, they can enhance other forms of power, making them more effective, precise, and survivable.
What we need, then, is not another breathless sermon about the drone revolution, but a sober recognition of both the promise and the limits of emerging technologies. Warfare is not transformed by gadgets alone, but by the institutions, doctrines, and strategic cultures that surround them. If the goal is to deter aggression and, if necessary, prevail in war, then the conversation must shift from innovation as spectacle to integration as practice. This means investing not only in platforms, but in people, logistics, alliances, and readiness. It means resisting the illusion that software can short-circuit strategy. And it means understanding that drones are not the future of war—they are simply one piece of a much older, more complicated, and more human story.
