This summer’s public drills were not a cosplay of future war. They were a progress report, and a warning. State footage and government reporting of a July exercise in Inner Mongolia showed the PLA stringing together reconnaissance drones, loitering munitions, FPV kits, ground command nodes and high-energy laser interceptors into a single mission flow: sense, decide, mass, and strike. The optics were polished for domestic and international audiences. The operational message was blunt. China is practicing swarm-enabled saturation campaigns and the counter-systems needed to survive them.
The imagery and CCTV narration emphasized six mission phases: reconnaissance, AI planning, infiltration, aerial attack, elimination and anti-access. Public footage highlighted an OW-5 series truck-mounted laser shooting down a target drone at distance and a family of Norinco “Flying Dragon” loitering munitions and small tactical UAVs operating together in tasked waves. Those details matter because they show the PLA is not only experimenting with individual platforms. It is rehearsing an integrated, layered kill chain that pairs cheap attritable air vehicles with higher-end sensors and effects.
Do not confuse showmanship with finished capability. The footage is a deliberate mix of demonstrable systems and aspirational doctrine. Still, the ingredients on display are operationally meaningful: numbers, networking, automated planning, and a portfolio of counters to Western-style kinetic defenses. That combination is what military planners worry about: cheap sensors and shooters working at tempo to force defenders into costly, brittle responses. Independent analysts have been flagging this risk for years, and think tank work has framed the problem clearly: China has the industrial base and geography to scale drone fleets in ways that complicate U.S. and allied defenses in a Taiwan contingency or other high-end scenarios.
The invisible enabler beneath the choreography is algorithmic. Public reporting this year has put a name on one of the AI engines Beijing finds useful and fast: DeepSeek. U.S. officials and international reporting have alleged ties between that commercial AI ecosystem and PLA procurement, and raised concerns about how inexpensive, efficient reasoning models accelerate mission planning, automated target-finding and swarm coordination. If you can compress days of analysis into seconds and generate robust plan options that feed local controllers or autonomous agents, you compress the kill chain and increase the threat created by massed, low-cost unmanned systems.
What the PLA demonstrated this summer is therefore twofold. First, swarm logic is moving from lab to field in modular, repeatable forms: ISR nodes feed planners, planners feed tasking to swarms of loitering munitions, and effects are layered with countermeasures to blunt known defensive seams. Second, Beijing is simultaneously fielding counter-swarm technologies: directed-energy weapons, short-range interceptors and electronic attack systems. That is sensible operationally. A swarm-based offense and a layered counter-swarm defense are two sides of the same doctrinal coin. Public displays of lasers and integrated C2 are as much about messaging to domestic industry and foreign buyers as they are about testing actual kill ratios.
So where does that leave deterrence and defense? First, buyers of the “we will just shoot them down” narrative should be humble. Cost asymmetry is real. As other conflicts have shown, shooting expensive interceptors at cheap drones is an unsustainable arithmetic unless the defender changes the cost curve: hardening critical nodes, using layered electronic and directed-energy effects, and buying large quantities of affordable counter-UAS systems. Second, automation at the planning layer shifts tempo considerations. Faster decision loops favor whoever can close the sensor-to-shooter pipeline most reliably and at scale. That is why the DeepSeek reporting matters: algorithmic efficiency is a force multiplier when platforms are plentiful.
Policy and procurement implications are stark. Western militaries must stop imagining parity means matching one-for-one and instead accept three hard facts: attritable systems will proliferate; integrated autonomy will compress decision time; and industry scale in China gives Beijing options for mass. The right response is multipronged: invest in resilient physical protection for critical bases and nodes, accelerate affordable counter-UAS inventories, and double down on long-endurance, long-range manned and unmanned sensors that raise the cost of massed local swarms. Equally important is an honest debate about rules, limits and operational doctrine for autonomy; celebrating automation without governance risks creating new escalation dynamics.
Finally, if summer drills taught one practical lesson it is this: tomorrow’s battles will be decided by orchestration as much as by hardware. The PLA’s tests show an appetite for holistic systems thinking where numbers, software and industrial capacity are fused into doctrine. That is not an inevitability of Chinese victory. It is a challenge. The choice for U.S. and allied planners is clear: meet the orchestration problem with equal imagination, not just more sensors and interceptors. Or accept a future where decision tempo and massed autonomy rewrite the rules of deterrence in Asia and beyond.