Skip to main content
v2026.11,986 entries · CC-BY 4.0

Editorial · CASRAI · Research data infrastructure

Lenovo Introduces ThinkSystem SC755 V5 Liquid-Cooled Supercomputing Node at SC26

Lenovo introduced the ThinkSystem SC755 V5 supercomputing node at SC26, featuring Neptune warm-water cooling and 96 AMD EPYC CPUs per rack.

Published 16 Nov 2026· 7 minute read

Ask CASRAI · free to try

Ask about this story

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

CASRAI is the reference for research administration — bookmark it for the next question.

At the SC26 supercomputing conference in Chicago on November 15, 2026, Lenovo introduced the ThinkSystem SC755 V5, an ultra-dense, liquid-cooled supercomputing server tray engineered to bring high-performance computing efficiency and full warm-water thermal capture to scientific computing facilities and demanding artificial intelligence research clusters.

SC755 V5 Architecture and High-Density Supercomputing Launch

During the opening sessions of the SC26 supercomputing conference in Chicago on November 15, 2026, Lenovo officially unveiled its next-generation high-performance computing platform, the ThinkSystem SC755 V5. Built as an integral part of the Lenovo Neptune direct liquid-cooling portfolio, the server tray supports demanding computational workflows across academic research, national laboratories, and enterprise AI installations. The platform is designed around 6th Generation AMD EPYC processors, packing massive multi-threaded throughput into standard data center footprints without internal chassis fans.

Each SC755 V5 compute tray houses two dual-socket server nodes, delivering four 6th Generation AMD EPYC processors across a single vertical form factor. When deployed in the matching 13U ThinkSystem N1380 Neptune enclosure, eight compute trays slide vertically into the chassis. A standard 19-inch 42U rack accommodates three N1380 enclosures, achieving a density of 24 trays, 48 discrete nodes, and 96 AMD EPYC processors per rack. This architecture provides academic clusters with over 24,000 performance cores and up to 192TB of DDR5 memory per rack, meeting the needs of complex simulations, climate modeling, and distributed AI agent workflows.

Key Technical Specifications and N1380 Enclosure Design

The hardware architecture of the ThinkSystem SC755 V5 represents a generational shift in dense node engineering. Detailed in technical documentation published on Lenovo Press, each node supports two 6th Generation AMD EPYC processors with thermal design power ratings up to 600W per socket, enabling sustained peak frequencies without thermal throttling.

Dimension ThinkSystem SC755 V5 Specification Cluster Impact
Processor Sockets 2x 6th Gen AMD EPYC per node (4 CPUs per tray) Up to 256 cores per CPU at 600W TDP
Rack Compute Density 48 nodes / 96 CPUs across 3x N1380 enclosures Exceeds 24,000 cores in a single 19-inch rack
Cooling Mechanism 7th-Gen Neptune Direct Warm-Water Cooling 100% heat dissipation; up to 45°C inlet water
Chassis Interconnect Toolless blind-mate dripless quick-disconnects Eliminates manual fluid and power cabling
Memory Capacity Up to 32 TruDDR5 RDIMMs/MRDIMMs per tray Up to 192TB addressable DDR5 memory per rack
Storage and I/O Up to 6x E3.S NVMe SSDs; 2x PCIe Gen6 x16 Direct local scratch staging and 800G fabric

As documented in the Lenovo ThinkSystem SC755 V5 Datasheet, the N1380 chassis integrates up to four 15kW Titanium Power Conversion Stations, supplying centralized 48V direct-current power across a shared rear busbar. Trays slide into the chassis on guide rails and dock using dripless blind-mate quick-disconnect couplers for water, power, and management, eliminating manual fluid piping during servicing.

Neptune Warm-Water Cooling and Thermal Architecture

Thermal management in the ThinkSystem SC755 V5 is governed by the seventh iteration of Lenovo Neptune direct water cooling. Unlike hybrid systems that cool only the central processor while relying on fans for memory and power stages, the SC755 V5 tray captures 100% of generated heat directly into its liquid loop. Cold plates cover the AMD EPYC processors, TruDDR5 memory modules, PCIe networking cards, and onboard power conversion components.

A crucial operational capability of the Neptune architecture is its ability to operate with water inlet temperatures reaching up to 45°C (113°F). According to reporting by StorageReview, running at 45°C inlet water allows data centers to eliminate energy-intensive mechanical chillers, relying instead on outdoor dry coolers or cooling towers. Removing chassis fans cuts facility electrical draw by up to 40% and yields Power Usage Effectiveness metrics approaching 1.05. Furthermore, heated effluent above 50°C can be diverted into campus district heating networks, supporting institutional goals for carbon-aware computing.

Dynamic Partitioning and Compute Efficiency

A prominent architectural addition introduced with the SC755 V5 is Lenovo Dynamic Partitioning. Traditionally, high-density supercomputing blades are locked into fixed physical topologies, forcing dual-processor allocations that risk stranded resources during single-socket tasks. Dynamic Partitioning introduces a firmware-controlled capability that allows administrators to dynamically split the two dual-socket nodes on a compute tray into four independent single-socket systems.

This dynamic reconfiguration requires no physical re-cabling, board swaps, or chassis disassembly. Administrators can modify node topologies through the management interface based on queue demands in Slurm or other cluster orchestrators. This flexibility directly improves energy proportionality in computing by ensuring processor cores and memory channels operate at high utilization rather than idling in under-subscribed dual-socket configurations, allowing facilities to serve varied scientific workloads on shared hardware.

Background, Competitors, and Industry Reactions at SC26

The introduction of the SC755 V5 comes amidst rapid transformation across high-performance computing architectures. The underlying CPU silicon builds on the recent AMD EPYC Zen 6 processor launch, which introduced the SP7 socket and elevated performance core counts to 256 cores per socket. In the supercomputing tier, systems like the SC755 V5 share architectural lineage with leadership supercomputers such as El Capitan, where high-density liquid cooling and high-efficiency CPU designs proved essential to crossing exascale performance thresholds.

Industry analysts at SC26 noted that the SC755 V5 targets a key market transition. In an analysis of the SC26 announcements, Techaisle observed that while GPU acceleration captures substantial public attention, CPU-heavy infrastructure remains indispensable for data ingestion, graph analytics, agentic AI coordination, and scientific simulations. Techaisle emphasized that by packaging 24,000 cores into a single rack with warm-water cooling, Lenovo lowers the physical barrier to entry for mid-sized research facilities that lack custom exascale facility budgets. Representatives from Lenovo emphasized at the Chicago event that liquid cooling is transitioning from a specialized niche into a mandatory requirement for mainstream research clusters facing strict power caps.

Implications for Researchers, Universities, and Research Offices

For academic research institutions and campus research computing offices, the launch of the ThinkSystem SC755 V5 carries direct implications across procurement, data infrastructure, and grant administration:

  • Facility Retrofitting and Space Optimization: Many university data centers operate within legacy structures where floor space and air-cooling ductwork are exhausted. Because three N1380 enclosures deliver 96 sockets in a single 42U rack, universities can expand aggregate computational capacity without expanding data center square footage. The blind-mate dripless plumbing isolates liquid handling to the rack manifolds, reducing installation risk.
  • High-Throughput Research Data Pipelines: The inclusion of PCIe Gen6 expansion and high-speed E3.S NVMe storage on every compute tray accelerates data-intensive workflows. In fields such as cryo-electron microscopy, genomics, and astrophysics, staging intermediate datasets requires rapid I/O. Seamless integration with high-speed network fabrics improves data staging and transfer workflows to cluster storage, preventing compute nodes from idling during data ingest phases.
  • Compliance and Environmental Sustainability in Grant Proposals: Major funding agencies increasingly scrutinize the environmental footprint and energy efficiency of proposed computational infrastructure. Utilizing warm-water cooling that achieves PUE ratings below 1.10 provides research offices with concrete, verifiable metrics when preparing major research instrumentation proposals and environmental impact assessments.
  • Procurement and Operating Models: For universities that cannot immediately upgrade building plumbing, Lenovo is pairing the SC755 V5 with pre-configured colocation agreements and infrastructure-as-a-service options, allowing departments to deploy liquid-cooled hardware in certified third-party facilities under unified institutional oversight.

Deployment Timelines and Next Steps

Lenovo announced at SC26 that the ThinkSystem SC755 V5 compute tray and the ThinkSystem N1380 Neptune enclosure will begin customer shipments to early-access high-performance computing centers and research institutions in the first half of 2027. General availability across global channel partners, university procurement consortia, and enterprise colocation partners will follow as AMD scales volume availability of its 6th Generation EPYC processors.

Research computing directors planning cluster refreshes or facility upgrades can review architectural whitepapers and configuration tools through the Lenovo Press repository. Benchmarking studies evaluating Dynamic Partitioning performance across standard scientific workloads, including GROMACS, WRF, and OpenFOAM, are scheduled for publication alongside broader vendor hardware demonstrations during the first quarter of 2027.

Sources

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →