Yandex is relocating three of its supercomputers following drone attacks on its data centers in October. The company stated that the equipment remained undamaged and will be integrated back into its infrastructure after the move to new locations is completed.
The affected systems are the Chervonenkis, Galushkin, and Lyapunov supercomputers. These machines operate as independent nodes rather than a single cluster, allowing them to be housed in different facilities. Previously, Chervonenkis and Lyapunov were located in Sasovo, Ryazan Oblast, while Galushkin was based in Vladimir.
The supercomputers are built using AMD Epyc processors and Nvidia A100 graphics accelerators. In the June 2026 Top500 ranking, Chervonenkis held the 101st position, Galushkin ranked 134th, and Lyapunov was 161st. Their performance in the Linpack test is 21.53, 16.02, and 12.81 petaflops, respectively.
Yandex utilizes these machines for training neural networks and supporting service performance, including translation tools. According to the company, the systems are part of a significant investment strategy, with automated processes allowing the hardware to switch between handling user requests and model training based on traffic load.
