I’m not here to whisper sweet nothings about killer robots while you sip your morning coffee. The war with AI isn’t coming—it’s already here, unfolding in real-time across muddy Ukrainian fields and Pentagon war rooms, and we’re all pretending it’s someone else’s problem.
The Numbers Don’t Lie (Even When We’d Prefer They Did)
The global military AI market reached $9.31 billion in 2024 and is projected to explode to $19.29 billion by 2030—that’s a 13% compound annual growth rate for those keeping score at home.
But here’s where it gets spicy: the Pentagon alone dropped $1.8 billion on AI systems in 2024, while China’s People’s Liberation Army is spending at least $1.6 billion annually on AI-enabled systems, though the actual figure likely sits “in the low billions” because, shocker, the most expensive AI projects are classified.
We’re witnessing the world’s most expensive game of technological chicken, except nobody’s swerving.
Ukraine: The World’s Most Brutal Beta Test
If you want to understand where warfare is headed, stop reading think tank papers and look at Ukraine. In 2024, Ukraine produced nearly 2 million drones, with 96% manufactured domestically, and began purchasing 10,000 AI-enhanced drones.
That’s just the appetiser. Ukrainian forces have increased their target engagement success rate from 10-20% to around 70-80% by removing the need for constant manual control through autonomous navigation.
Let that marinate for a second. Three-to-four-fold improvement in killing efficiency through algorithms.
In December 2024, Ukrainian forces executed the first fully unmanned operation near Lyptsi, deploying dozens of uncrewed ground vehicles and FPV drones with no infantry participation. Robot armies aren’t science fiction anymore—they’re Tuesday.
And here’s the kicker: most of Ukraine’s autonomous-final-approach algorithms are derived from free, internationally available open-source models.
They took publicly available AI, retrained it on their extensive combat data, and weaponized it. The democratisation of autonomous killing has arrived, courtesy of open-source software.
The Speed Problem (Or: Why Humans Are the Bottleneck)
Sweden flew the world’s first AI-piloted fighter trial on its Gripen E in 2024. In 2023, DARPA developed an AI system that autonomously piloted an F-16 and engaged in dogfighting scenarios with a human-piloted F-16.
The message is clear: human reaction time is obsolete. The tempo itself becomes a weapon when algorithms can coordinate strikes faster than neurons can fire.
The Replicator Initiative: America’s “Oh Shit” Moment
In August 2023, Deputy Secretary of Defense Kathleen Hicks announced the Replicator initiative to field “multiple thousands” of autonomous weapons systems within 18 to 24 months to counter China’s advantage in mass.
The Pentagon’s explicitly stating it needs to “counter the People’s Liberation Army’s mass with mass of our own.”
The program received about $500 million for fiscal year 2024 and a similar amount for 2025, bringing total lifecycle costs to approximately $1 billion. For context, that’s chump change in defense spending, which tells you this is being fast-tracked at breakneck speed.
What are they buying? Switchblade 600 loitering munitions that can be launched from land, ships, and aircraft, plus counter-drone capabilities.
The defense contractor Anduril’s CEO, Palmer Luckey, wasn’t subtle about it, describing himself as part of a “warrior class that is enthused and excited about enacting violence on others in pursuit of good aims.”
Comforting.
China’s Algorithmic Warfare Doctrine
By 2030, the People’s Liberation Army expects to field a range of ‘algorithmic warfare’ and ‘network-centric warfare’ capabilities operating at different levels of human-machine integration. They’re not dabbling—they’re restructuring their entire military philosophy around AI.
Popular Mechanics reported in June 2024 that China could have fully autonomous AI weapons on the battlefield within two years.
In October 2024, China demonstrated AI-powered combat systems at their Golden Dragon joint military drills in Cambodia—months before the U.S. Army sent an armed “robot dog” to the Middle East for testing.
China’s advantage? Unlike China, which can integrate AI military advances into its state-run defense sector rapidly and without much bureaucratic red tape, the United States relies on partnerships with private firms that must weigh financial and public relations concerns.
The Escalation Paradox
Here’s where it gets existentially uncomfortable: a 2024 Stanford and Georgia Tech paper tested five language models in military and diplomatic decision-making and found that all models took actions that escalated the fight, with models developing arms-race dynamics and in rare cases even deploying nuclear weapons.
One researcher at the Future of Life Institute stated that autonomous weapons should be viewed as “on par with the start of the nuclear era,” adding that viewing this through an arms race lens, which is what the Pentagon is doing, then… well, we all know how arms races end.
The Swarming Nightmare
The Pentagon is overseeing more than 685 AI-related projects, several of which are tied to major weapons systems. The Air Force wants an AI-enabled fleet of more than 1,000 unmanned warplanes, with the first ones scheduled to be deployed by 2028.
Now imagine those thousand aircraft coordinating autonomously. AI-driven swarms involve autonomous agents that would interact with and coordinate with each other, likely in ways not foreseen by humans and also likely indecipherable to humans in real-time.
We’re building systems we can’t fully understand, giving them lethal authority, and hoping they stay predictable in the chaos of combat. What could possibly go wrong?
The First Autonomous Kill (That We Know About)
In March 2020, an autonomous drone in Libya, the Turkish-made Kargu-2 quadcopter, killed a human being without any input from a human operator. This happened during conflict between Libyan government forces and a breakaway military faction.
That was five years ago. The technology has advanced exponentially since then. Russia deployed its new Klin loitering munition in June 2025, featuring autonomous targeting and an airburst warhead designed to neutralise small, fast-moving aerial threats.
Russia’s Artemis-10 FPV drone can lock onto targets at 500 meters and pursue them autonomously using machine vision, even if communications are lost due to electronic warfare.
These aren’t prototypes. They’re deployed. Right now.
The Uncomfortable Truth
A red-team exercise at the Munich Security Conference in March 2024, with 28 experts in the room, surfaced scenarios where rivals coordinate strikes on Western AI infrastructure. The simulation showed how advanced AI could enable authoritarian leaders to target opposition and dissidents more effectively.
What keeps military commanders awake? Coordinated hits on data centers, AI-fueled disinformation that mutates in real time, and autonomous swarms operating faster than humans can react.
So, When Did We Go to War with AI?
The question assumes a clear starting line. But Ukraine produced approximately 2 million drones in 2024, 96.2% of which were domestically manufactured, with AI primarily serving to augment existing capabilities.
President Zelensky announced in October 2024 that Ukraine’s defense industry had reached an annual production capacity of up to four million drones.
Russia is reportedly matching this production pace, with Vladimir Putin directing the Russian government to strengthen AI cooperation with China.
The war with AI didn’t start with a bang. It started with procurement contracts, classified budgets, and open-source code being repurposed for autonomous targeting. It started when militaries worldwide realised that whoever fields AI-enabled weapons first gains a potentially insurmountable advantage.
We’re not going to war with AI. We’re already there, watching algorithms learn to kill more efficiently with each engagement, while humans scramble to maintain the illusion of control.
Ukraine estimates it wants at least half the drones it buys in 2025 to have AI guidance—an increase from 0.5% to 50%. That’s not gradual adoption. That’s a technological sprint toward autonomous warfare.
The Uncomfortable Question Nobody’s Asking
If autonomous weapons can lower the political cost of using force by reducing casualties for the aggressor, doesn’t that make conflict more likely? When sending robots to die becomes cheaper than diplomacy, what incentive remains for peace?
The data suggests we’ve already answered that question. The global military AI market is doubling every six years. The Pentagon launched a $100 million AI program in December 2024 to accelerate deployment of frontier AI models, running four pilot programs in 90-day experiments.
Ninety days. That’s how quickly we’re moving from concept to battlefield deployment.
