I recently read a report about how AI is redefining the logic of modern warfare, and it was quite shocking.

In simple terms, it’s a global surveillance network made up of Palantir, Anduril, and top-tier large models that has completely rewritten traditional military decision-making processes. It’s no longer generals commanding thousands of troops in front of a map, but algorithms completing the entire kill chain—“sense-lock-on-execute”—within seconds.

What impressed me most was a system called “Dad, Where Are You.” It doesn’t track aircraft or missiles; it tracks when a person comes home. The logic is cold: launch an attack when the target reunites with family, which is easier than striking a military base. Behind this is the fact that AI has evolved from a supporting tool into a true decision-maker.

There’s another detail worth paying attention to—the role of the Claude model throughout the process. The Pentagon once tried to remove its safety guardrails so it could be integrated directly into fully autonomous lethal weapon systems. In the end, this task was handed over to OpenAI and Musk’s xAI. Behind it all is a fierce clash over AI ethics, as well as a reshuffling of power between Silicon Valley and Washington.

From a technical perspective, this operation demonstrates several astonishing breakthroughs. The Starshield satellite constellation bypassed electromagnetic jamming through 200 Gbps laser inter-satellite links. Drone swarms can carry out missions autonomously even when GPS and human operators are lost, and they can even seamlessly switch between different AI systems while in flight—like updating apps on a phone.

Even more worth thinking about is the capital logic behind it. Silicon Valley venture capital firms such as a16z completed a $15 billion funding round, shifting their bets from social software to hard-tech defense companies. Their playbook is: not producing an F-35 worth $100 million, but producing 10,000 autonomous drones worth $10,000 each. This “expendable” view of war has fundamentally changed how the military-industrial complex operates.

But behind this victory lie three paradoxes of the clocks. The military clock has been turned to extreme speed—from target confirmation to execution takes only a few seconds. The economic clock is accelerating consumption; the drones get cheaper, and mass consumption will rebound on the supply chain. The slowest is the political clock—algorithms can precisely kill a leader, but they cannot automatically win over public support.

That’s also the most unsettling part. When war becomes as efficient and low-casualty as clicking a screen, the political threshold for humans to start wars is lowered in a dangerous way. We’re entering an era of software-defined geopolitics, and human commanders may no longer have time to feel fear.