US President Donald Trump’s deepest challenge in talks with Chinese leader Xi Jinping this week isn’t to do with trade or Taiwan. It’s that China has mastered the innovation model that the US invented.

Since at least World War II, the US has capitalized on a security flywheel in which its dominance in commercial technology has driven new military capabilities. Then, automotive assembly lines pivoted almost overnight to producing tanks and bombers. In subsequent decades, the Pentagon itself funded speculative basic research that resulted in semiconductors, GPS, and the internet — all of which looped back to strengthen American military advantage.

But since the end of the Cold War, that flywheel has seized. The defense sector consolidated into a smaller cluster of giant contractors. The best engineering talent migrated to Silicon Valley. Procurement rules calcified around rules designed for cutting edge, but expensive and slow-cycle, weapons systems, hostile to the iteration speed that commercial technology demands.

Instead, it is now China that is carrying out that playbook.

DJI illustrates the logic. The company did not become the world’s dominant drone manufacturer by winning defense contracts; it was founded to build better camera platforms for photographers and filmmakers. Commercial pressure drove down costs and accelerated development in ways no defense procurement cycle could replicate. The military applications arrived later, essentially for free. China now supplies roughly 80 percent of global drone components — motors, batteries, flight controllers — embedding Beijing inside the supply chains of almost every combatant in every active conflict, including, awkwardly, the Pentagon’s own “Blue List” of approved domestic drone platforms.

BeiDou, a satellite positioning system guided by Chinese navigation chips, extended the same strategy into space. The Chinese government began building the system in the 1990s as a state-funded strategic priority, subsidizing satellite launches and ground infrastructure at a scale no commercial logic could have justified. Today, BeiDou operates more than 45 satellites and claims over a billion users worldwide, giving it a commercial user base that makes it impossible to sanction away. Among others, Iranian drones and missiles navigate using BeiDou.

AI now plays a key role in perpetuating China’s security flywheel, with Beijing’s choice to embrace open-source AI platforms accelerating its velocity. Within months of DeepSeek shocking Silicon Valley last year by matching frontier-lab performance at a fraction of the cost, China’s state arms manufacturer Norinco had integrated a DeepSeek-powered system into an autonomous combat vehicle. And Beihang University, China’s premier military aviation research institution, filed patents for DeepSeek-powered drone swarm decision-making. Researchers at Xi’an Technological University, meanwhile, said their DeepSeek-based planning system could assess 10,000 battlefield scenarios in 48 seconds — a task they said would occupy a human military planning team for 48 hours.

Paul Scharre at the Center for a New American Security argues in his book Four Battlegrounds that AI developed for civilian markets now forms the foundational substrate of military competition, and that the US defense establishment has moved too slowly to reckon with the implications. Christian Brose, an executive at the defense company Anduril, makes the companion case in The Kill Chain: A Pentagon acquisition system built for Cold War timelines and cost structures cannot compete with an adversary whose military capability grows directly from commercial market dominance.

Trump arrives in Beijing with a crowded agenda: the Strait of Hormuz, Taiwan arms sales, tariffs, deals for jets and soybeans,, and the case of jailed Hong Kong media tycoon Jimmy Lai.

What those discussions will almost certainly not address is a deeper, structural problem: A supply chain with the US military as its only customer produces exactly the dynamics that lost America this competition in the first place. Prices stay high because competition stays thin. Iteration slows because the acquisition cycle governs the pace.

Neither procurement reform nor a surge of venture capital enthusiasm for defense companies addresses the foundational issue. China has built a commercial flywheel that generates military capability at industrial scale by letting billions of civilian consumers bear the R&D and manufacturing costs. Washington is trying to replicate that outcome through the very government-directed procurement model that caused the original flywheel to stop spinning.

The US wrote the playbook, then largely set it aside for 30 years while its principal strategic rival studied, adapted, and applied it with greater discipline and consistency. The gap those three decades opened will not close across a negotiating table in Beijing — and may not close before the technologies involved have fundamentally reshaped the terms of the next conflict.