Ukraine Finds Nvidia Microcomputers in Russian Missiles and Drones

Ukrainian forensic examiners and intelligence analysts have discovered consumer-grade Nvidia Jetson microcomputers inside Russian military hardware, including fully autonomous strike drones tested in Zaporizhzhia and a newly deployed S-71M Monokhrom cruise missile. The findings, made public by Ukraine's Main Intelligence Directorate (HUR), highlight how commercial artificial intelligence hardware continues to reach Russian weapons systems despite international export controls and Nvidia's official withdrawal from the market. Analysis of the recovered drone guidance systems revealed that Russia tested self-targeting unmanned vehicles between May and July, delegating strike decisions entirely to the Nvidia Jetson Orin processor without requiring human confirmation.

Air defense officials reported that one of these autonomous drones killed three people during a strike on Zaporizhzhia on July 6.

Separately, HUR specialists identified an Nvidia Jetson Orin NX module featuring 16 gigabytes of memory inside the wreckage of an S-71M cruise missile, marking the most powerful Nvidia processor recovered from Russian weaponry to date. Manufacturing labels on the missile's board indicated that the module was produced in March 2025, long after supply restrictions were implemented.

The Jetson Orin NX features eight ARM Cortex-A78AE cores, 32 Tensor Cores, and dedicated Deep Learning Accelerators capable of performing up to 157 trillion AI operations per second to process targeting imagery. The component was cataloged among 35 foreign electronic parts identified in the S-71M missile system on the official War&Sanctions portal.

Ukraine's military intelligence agency warned that current sanctions and export monitoring remain insufficient to prevent commercial off-the-shelf microelectronics from entering Russian arms manufacturing pipelines. In response to the discoveries, Nvidia stated that while the Jetson line is designed for civilian robotics, commercial supply chains remain difficult to restrict entirely when consumer-grade components are resold through third-party networks.