Telecom Exchange launches rental of dedicated servers with GPU configurations for AI tasks

Telecom Exchange has introduced a service for renting dedicated physical servers hosted on partner infrastructure, providing an optimal solution for companies that are actively adopting AI but want to avoid significant costs. The GPU-based configurations enable resource-intensive AI tasks such as training neural networks, processing large data volumes, and high-performance computing. For training and fine-tuning large models, clusters of multiple servers can be rented. GPU clusters are also suitable for rendering, 3D graphics, and VDI for designers, as well as for Data Science specialists. It is possible to deploy Kubernetes clusters or other systems for managing distributed computing on multiple physical servers. The service utilizes RAG (Retrieval-Augmented Generation) technology to create systems that can interact with client knowledge bases and documents. There is no need to purchase expensive equipment; everything is available for rent, including internet links and local networks with bandwidth up to 400 Gbps, GPU cards, and hardware RAID controllers. More than 100 ready-made configurations are available in Tier III data centers with deployment from 2 minutes. The infrastructure capacity for AI tasks can be scaled without equipment replacement. Dina Fomicheva, head of corporate sales at Telecom Exchange, noted growing demand for AI computing power, as training models and inference require resources that are expensive to purchase and maintain independently. Through the partner program, they can offer servers with GPUs for rent while maintaining full control, isolation, and configuration flexibility for each client. The service costs from 50 thousand rubles per month.