Russia is now fielding long-range, fiber-optic–controlled drones while simultaneously accelerating development of fully autonomous, AI-driven attack drones that mark a dramatic shift in its battlefield strategy. Ukrainian officials confirm that Moscow is deploying jam-proof fiber-optic FPV platforms with a 50-kilometer range, forcing Ukrainian units to cover supply routes with netting to protect vehicles from incoming strikes.
At the same time, Russia is rolling out a new generation of AI-enabled systems powered by smuggled NVIDIA Jetson processors, which were supposed to be blocked by sanctions but are now appearing inside multiple Russian drones.
Ukrainian drone specialist Serhii “Flash” Beskrestnov reports that Russia has adapted Ukrainian innovations for its V2U autonomous strike drone, while other captured systems, like the Shahed MS001 and Tyuvik, show that Russia is rapidly building an arsenal of “digital predators” capable of finding and attacking targets without human control.
The V2U remains Russia’s most advanced autonomous platform. First deployed in Ukraine’s Sumy region in February 2025, the drone was being used 30–50 times per day by mid-May across multiple fronts. It navigates by using machine vision to compare live imagery with digital terrain maps stored on a solid-state drive, upgraded to 128 GB in later versions. A 14-megapixel camera, laser rangefinder, and NVIDIA Jetson Orin processor enable the drone to identify targets, conduct terrain analysis, and strike without GPS or human commands.
Russian forces update its onboard code almost weekly, training the AI directly through battlefield experience. The drone’s AI is powerful enough to fly up and down roads searching for targets, though its limited ability to distinguish targets has led to mistakes, including striking civilian infrastructure such as a public toilet instead of a vehicle.
To extend range, Russian forces deploy large “mother drones” that transport smaller V2Us deep into contested airspace before releasing them for individual strikes. Russia is also experimenting with coordinated swarms of seven or eight V2Us, each painted with distinct wing colors to visually identify one another.
According to Ukrainian analysts, these drones can maintain formation, coordinate attack order, and perform anti-air evasive maneuvers if a member of the group is shot down. This behavior resembles early machine-driven swarm intelligence and may be adapted for Russia’s long-range Shahed drones, creating fully autonomous loitering-munitions swarms capable of saturating Ukrainian defenses.
Captured drones reveal the scale of Russia’s dependence on foreign components. The V2U relies on an NVIDIA Jetson Orin for AI processing, Intel wireless adapters, Sony optical sensors, and numerous Chinese-made motors, drives, and batteries. The MS001, an upgraded Shahed variant, also carries a Jetson Orin paired with a thermal imager and digital modem, allowing it to recognize ground objects and strike moving targets rather than simply flying to preset coordinates.
A third autonomous drone, the Tyuvik, resembles a miniature Shahed and is now reportedly in mass production. Tyuvik can locate and strike moving armored vehicles despite using inexpensive commercial hardware, suggesting it also depends on smuggled Western or Chinese AI processors.
Hellbound and Down 