Summary
Lenovo announced new ThinkSystem servers powered by 6th Generation AMD EPYC processors, alongside storage, interconnect options and AI infrastructure services. The portfolio is designed for AI and high-performance computing workloads across data centres, hybrid environments and colocation facilities.
Lenovo announced a new portfolio of servers, storage, interconnect options and services for organisations running artificial intelligence and high-performance computing (HPC) workloads. The company said its new ThinkSystem servers use 6th Generation AMD EPYC SP7 processors, with Lenovo Neptune liquid cooling and expanded services intended to help customers deploy and operate AI systems across their own data centres, hybrid environments and colocation facilities.
The announcement brings together several parts of an AI or scientific-computing system. Processor performance matters, but so do memory bandwidth, moving data between components, storage throughput and removing heat. Lenovo’s additions address those parts alongside deployment and operating services.
Servers, cooling and storage target system bottlenecks
The dual-socket ThinkSystem SC755 is a Neptune liquid-cooled server. Lenovo says water can remove up to 100% of system heat and that the SC755 delivers up to 40% higher sustained performance than similar air-cooled systems. Those figures are based on Lenovo internal testing; the company says results vary with system configuration, workload, facility water temperature and data-centre conditions.
Lenovo positions the ThinkSystem SR755 for high-bandwidth work including engineering, financial modelling, electronic design automation, large-scale virtualisation and agentic AI infrastructure. The company describes it as a high-compute-density system with high memory bandwidth, and says it is engineered for lower power consumption and better economics than dual-socket configurations. That comparison is based on Lenovo internal analysis.
The announced storage systems are the ThinkSystem DE6800 and DE6500, designed for high-throughput, latency-sensitive block workloads. Lenovo says the DE6800 is aimed at applications that require high throughput and reliable storage, while the DE6500 is intended to offer high performance with simpler deployment and management. Lenovo also reported validation of the Cornelis CN5000 fabric as an interconnect option for demanding AI and HPC environments. Interconnects carry data between computing components, so their performance can affect workloads that move large volumes of data.
Services extend from deployment to operating capacity
Lenovo is adding GPU deployment, configuration and optimisation capabilities to its Hybrid AI Factory services. It also says the services now support multi-node environments using Red Hat AI Enterprise and NVIDIA AI Enterprise with NVIDIA Run:ai, giving customers options for building and managing AI systems across multiple servers.
For larger deployments, Lenovo is expanding gigafactory services, including NVIDIA Infrastructure Services expertise for deployment and cluster validation. The company describes these services as covering work from planning and build through deployment and optimisation.
Lenovo also says it can facilitate access to Digital Realty’s colocation facilities for organisations needing additional data-centre capacity. Its announcement describes a single point of contact for colocation space and related infrastructure services, including scalable capacity through Lenovo TruScale Infrastructure as a Service.
The portfolio’s significance is its system-level approach: it combines compute, cooling, storage, networking options and deployment support rather than focusing on processors alone. That is relevant as AI inference and scientific workloads place demands on both computation and the infrastructure that feeds and cools it. The announcement describes Lenovo’s intended capabilities; the SC755 performance figures are company-reported results from internal testing.