Lenovo Unveils New AI and HPC Servers Powered by AMD’s 6th Gen EPYC Processors

Lenovo has introduced a new portfolio of artificial intelligence and high-performance computing (HPC) infrastructure powered by AMD’s sixth-generation EPYC 9006 server processors. Announced on November 10, 2026, the portfolio includes new Lenovo ThinkSystem servers, high-speed storage systems, liquid-cooling technology and expanded infrastructure services designed for organisations building large-scale AI environments.

The announcement reflects the growing convergence of AI and traditional high-performance computing. As businesses and research institutions deploy more complex AI models, demand is increasing for servers capable of handling intensive calculations, large datasets and continuous inference workloads. Lenovo is positioning its latest infrastructure as a solution for these requirements, with a particular focus on computing performance, energy efficiency and data-centre scalability.

Lenovo Introduces New ThinkSystem Servers With AMD EPYC 9006

The centrepiece of Lenovo’s announcement is its expanded ThinkSystem portfolio powered by AMD’s sixth-generation EPYC 9006 SP7 processors.

The new processors are based on AMD’s Zen 6 architecture and are designed to support demanding workloads across enterprise computing, cloud infrastructure, scientific research and AI.

Lenovo introduced two key server platforms:

• Lenovo ThinkSystem SC755: A dual-socket, liquid-cooled server designed for high-performance computing and demanding AI workloads.
• Lenovo ThinkSystem SR755: A platform focused on high compute density and memory bandwidth, targeting engineering, financial modelling, large-scale virtualisation and agentic AI infrastructure.

The systems are designed to help organisations increase computing capacity while managing power consumption and infrastructure costs.

The launch also gives customers additional choices for building AI environments that combine conventional CPU computing with accelerated GPU infrastructure.

ThinkSystem SC755 Uses Lenovo Neptune Liquid Cooling

One of the most notable products in the portfolio is the ThinkSystem SC755, which uses Lenovo’s Neptune liquid-cooling technology.

High-performance processors generate substantial heat, particularly when operating under sustained workloads. Traditional air-cooling systems can become less practical as server power consumption and computing density increase.

Lenovo’s Neptune technology uses warm-water liquid cooling to remove heat directly from system components.

According to Lenovo, the SC755 can remove up to 100% of system heat through water cooling and deliver up to 40% higher sustained performance compared with similar air-cooled systems.

The company says this design can help organisations operate high-performance servers more efficiently, particularly in facilities where power availability and cooling capacity are major constraints.

The system is designed to support AMD’s sixth-generation EPYC processors with thermal design power of up to 600 watts.

The SC755 is intended for workloads such as scientific simulations, engineering research, AI processing and other computationally intensive applications.

These performance figures are based on Lenovo’s internal testing, and actual results will depend on system configuration, workload characteristics and data-centre conditions.

AMD’s 6th Generation EPYC Processors Target AI Infrastructure

AMD’s sixth-generation EPYC processors are an important part of Lenovo’s new server portfolio.

The processors, codenamed Venice, are based on the Zen 6 architecture and are designed to support the changing requirements of modern data centres.

AI infrastructure is no longer limited to training large models. Companies increasingly need computing resources for inference, data retrieval, software execution and AI agents that interact with multiple business applications.

These activities require a combination of CPU performance, memory bandwidth, networking and accelerated computing.

AMD’s new EPYC portfolio is designed to address different roles across AI infrastructure, including general-purpose enterprise services, GPU host systems and computing environments used to run AI agents.

The flagship EPYC 9996 processor is positioned for high-performance general-purpose server workloads. AMD says its performance and efficiency improvements can help customers run more applications and transactions within existing data-centre power and space limitations.

For Lenovo, integrating the new processors into ThinkSystem platforms provides customers with another option for deploying AI and HPC systems.

Lenovo Adds New High-Performance Storage Systems

Alongside the new servers, Lenovo announced two storage platforms designed to support data-intensive applications.

Product	Primary purpose
Lenovo ThinkSystem DE6800	High-throughput storage for demanding applications
Lenovo ThinkSystem DE6500	High-performance block storage with simplified deployment and management

The DE6800 is designed for workloads that require high data throughput and reliable storage performance.

The DE6500 targets organisations seeking high-performance storage with a more economical deployment model.

Storage has become an increasingly important part of AI infrastructure because models require access to large datasets during training, inference and other processing tasks.

A powerful server can still experience performance limitations if it cannot access data quickly enough. Lenovo’s expanded portfolio therefore addresses storage alongside computing and cooling rather than treating these components as separate infrastructure investments.

Cornelis CN5000 Adds Another Networking Option

Lenovo is also validating the Cornelis CN5000 fabric as a high-performance interconnect option for demanding AI and HPC environments.

Interconnect technology determines how quickly servers and computing nodes can exchange information.

This becomes especially important when workloads are distributed across multiple systems. Scientific simulations, large AI deployments and high-performance computing clusters may require substantial communication between processors and storage resources.

By adding another interconnect option, Lenovo aims to give customers greater flexibility when designing infrastructure for data-intensive workloads.

The company is positioning the networking, computing and storage components as parts of a broader system designed to reduce performance bottlenecks.

Lenovo Expands AI Factory and Infrastructure Services

The announcement also includes an expansion of Lenovo’s services portfolio.

Building an AI data centre involves more than purchasing servers and installing hardware. Organisations must configure computing clusters, integrate software platforms, manage security and optimise infrastructure for production workloads.

Lenovo is expanding its Hybrid AI Factory Services to help customers deploy and operate these environments.

The expanded offering includes GPU deployment, configuration and optimisation capabilities. Lenovo is also adding multi-node support across Red Hat AI Enterprise and NVIDIA AI Enterprise with NVIDIA Run:ai.

These capabilities are intended to help organisations deploy, govern and scale AI workloads across hybrid infrastructure.

For larger deployments, Lenovo is expanding its gigafactory services, including specialised NVIDIA Infrastructure Services expertise for deployment and cluster validation.

The company is also facilitating access to Digital Realty’s global colocation network. This is intended to help customers that need additional data-centre capacity without building and operating every facility themselves.

Through these services, Lenovo is seeking to position itself as an infrastructure and deployment partner rather than simply a supplier of server hardware.

AI and HPC Are Converging Across Research and Business

Lenovo’s announcement comes as AI becomes increasingly important in scientific research, engineering and industrial applications.

High-performance computing has traditionally been used for workloads such as climate modelling, physics simulations, financial calculations and engineering design.

AI is increasingly being integrated into these activities to accelerate analysis, identify patterns and improve simulation workflows.

The convergence creates demand for infrastructure capable of handling both traditional computational workloads and AI applications.

Lenovo highlighted several examples of organisations using its infrastructure.

At the University of Florida, the fourth generation of the HiPerGator supercomputer, built using Lenovo, NVIDIA and AMD technology, delivers 17.49 petaflops of computing performance. The system processed 33 million research requests over the previous year, according to Lenovo.

DreamWorks Animation also uses Lenovo Neptune liquid-cooled servers in its data centre. Lenovo said the infrastructure delivers 20% greater performance and faster rendering, helping the studio accelerate creative production.

These examples demonstrate the range of applications for AI and HPC infrastructure, from university research to commercial content production.

The Bigger Picture

Lenovo’s latest announcement highlights how competition in the AI infrastructure market is expanding beyond GPU accelerators. CPUs, storage, networking, cooling and software management all play important roles in determining how efficiently an AI data centre operates.

The integration of AMD’s sixth-generation EPYC processors also gives Lenovo customers another option as enterprises diversify their computing infrastructure. At the same time, liquid cooling and expanded deployment services address practical challenges that organisations face when scaling AI systems.

Looking Ahead

Lenovo’s new ThinkSystem servers and expanded infrastructure services are aimed at organisations preparing for more demanding AI and HPC workloads. The combination of AMD’s latest EPYC processors, Neptune liquid cooling, high-performance storage and additional networking options could help customers build more efficient computing environments. Adoption will depend on pricing, system availability, workload performance and the ability of organisations to integrate the platforms into existing data centres.

The announcement also reflects a broader shift in the technology industry towards full-stack AI infrastructure. As AI agents, scientific computing and continuous inference increase demand for computing resources, server manufacturers will compete on the performance of complete systems rather than processors alone. Lenovo’s expanded portfolio places it in that competition alongside AMD and other infrastructure technology providers.

Updated November 13, 2026: Added Lenovo’s original announcement and clarified the NVIDIA Run:ai product name.

Source: Lenovo’s November 10 infrastructure announcement. Performance comparisons are Lenovo claims based on internal testing or analysis; actual results vary.