Lenovo ThinkSystem ST650 V3 Server

Tower Server Lenovo ThinkSystem ST650 V3 for Virtualization, Databases, and Edge AI

Lenovo ThinkSystem ST650 V3 is a powerful, 2-socket enterprise-class server in a Tower (pedestal) form factor, convertible to a standard 19-inch rack (4U). The platform is designed for remote and branch offices (ROBO), large enterprises, and service providers. It targets demanding workloads: local virtualization, databases, edge computing (Edge AI), retail analytics, and Virtual Desktop Infrastructure (VDI). The server is built on a modern architecture supporting PCIe Gen 5.0, DDR5 memory, and high-density GPU accelerators.

By combining high compute density, support for up to four full-profile GPUs, and a flexible storage subsystem, the ST650 V3 addresses workloads traditionally reserved for rack platforms. At the same time, its Tower form factor allows deployment in office spaces and unattended facilities without a dedicated server rack.

Use Cases

The Lenovo ThinkSystem ST650 V3 is the optimal solution for organizations requiring:

• Local virtualization in remote and branch offices (ROBO, branch data centers)
• Medium and large-scale databases, including memory-intensive solutions
• Edge computing and local AI inference (Edge AI)
• Retail analytics, video analytics, and computer vision systems
• Virtual Desktop Infrastructure (VDI)
• Workloads requiring up to 4 GPUs in a single chassis without switching to a rack platform

Compute Subsystem

The server supports two 4th Gen Intel Xeon Scalable (Sapphire Rapids) or 5th Gen Intel Xeon Scalable (Emerald Rapids) processors with TDP up to 250 W per socket. Maximum core count reaches 120 per system, enabling efficient consolidation of virtual machines, containers, and databases on a single platform. Built-in Intel AMX, DSA, and QAT accelerators speed up inference, encryption, and data processing workloads.

Memory

The server provides 32 DIMM slots (16 per socket). It supports high-speed DDR5 ECC RDIMM memory at speeds up to 5600 MT/s. Maximum memory capacity is 4 TB using 128 GB modules. Memory Mirroring, Memory Sparing, and ADDDC modes are supported — critical for databases and virtualization workloads sensitive to data integrity and fault tolerance.

Graphics Accelerators (GPU)

The chassis architecture supports up to 4 full-profile, double-wide GPUs (e.g., NVIDIA H100/A100) or up to 8 compact single-wide accelerators (e.g., NVIDIA L4/T4). This makes the server a full-fledged AI station for local inference: video processing, image recognition, recommendation systems, generative models, and real-time analytics — without sending data to the cloud.

Storage Subsystem

The modular backplane design supports up to 32 × 2.5″ (SFF) or up to 16 × 3.5″ (LFF) drives. SAS/SATA (HDD/SSD) interfaces and high-speed AnyBay NVMe PCIe Gen5 are supported. The AnyBay configuration allows flexible mixing of NVMe, SAS, and SATA drives in the same bays, adapting the server to changing workload profiles without chassis replacement. For secure OS boot, an internal dual-drive M.2 module with hardware RAID 1 is used, separating the boot volume from the data array and improving fault tolerance.

Interfaces and Expansion

The motherboard provides up to 9 PCIe slots (up to 5 Gen5 and up to 4 Gen4 slots), plus a dedicated OCP 3.0 slot for network cards without consuming standard expansion slots. This configuration allows simultaneous installation of GPU accelerators, RAID controllers, high-speed network adapters (including 100/200GbE), and expansion cards without compromises.

Management, Reliability, and Security

• Management. Monitoring and configuration are handled by the embedded Lenovo XClarity Controller 2 (XCC2) service processor, providing remote access to the server regardless of OS state. The platform supports centralized management via Lenovo XClarity Administrator, as well as Redfish API and integration with existing monitoring systems.
• Power and Cooling. The server is equipped with two redundant hot-swap power supplies rated at 750 W, 1100 W, 1800 W, or 2400 W with Platinum/Titanium certification. Cooling is provided by four redundant hot-swappable fans (in dual-CPU configurations). Redundant power and cooling allow the server to remain operational if a component fails.
• Security. The integrated Lenovo ThinkShield suite provides hardware-level protection: TPM 2.0 crypto chip, chassis intrusion switch, Secure Boot, and firmware verification compliant with NIST SP 800-193. This makes the server suitable for workloads requiring corporate and regulatory cybersecurity compliance.

Why GHI SERVERS

• Direct supply from China. We work with CNY and RUB. Turnkey logistics of server equipment from China, including customs clearance and delivery to our Moscow warehouse.
• Custom configuration. We help select processors, memory capacity and type, drive backplanes (SFF/LFF/AnyBay), GPU accelerators, and network adapters for your task and budget.
• Moscow warehouse. Prompt replacement of defective units and warranty service without long downtime.
• 20+ years of expertise. Authorized Lenovo partner since 2005. Pre-sales engineering, configuration assistance, and post-sales support.

Service and Support

• 24/7 technical support
• Pre-sales engineering and configuration assistance
• OEM warranty and full provenance documentation
• Installation, setup, and deployment of server equipment
• Component replacement and warranty service