Skip to content
HomeSolutionsLenovoLenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL
Lenovo logo
LenovoNew

Lenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL

A four-H200 node sized for departmental fine-tuning and private inference.

Full manufacturer warrantyAuthorized channel48-hour quote

We help you choose, configure, and deliver the right system — no obligation.

Lenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL — Lenovo enterprise hardware
Lenovo logo
Lenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL hardware detail 1
Lenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL hardware detail 2
Lenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL hardware detail 3

Configuration at a Glance

GPU / Accelerator4x NVIDIA H200 NVL (141GB HBM3e each)
GPU InterconnectPCIe Gen5 + NVLink bridges, GPUDirect RDMA
CPU2x AMD EPYC 9004/9005 series
MemoryUp to 1.5TB DDR5-6400 (24 DIMM slots)

Tailored per engagement. Full technical overview below.

Configuration Options

Core specifications for this system. Every component is configurable to your workload — request a quote for a tailored build.

GPU / Accelerator

4x NVIDIA H200 NVL (141GB HBM3e each)

Processor

2x AMD EPYC 9004/9005 series

Memory

Up to 1.5TB DDR5-6400 (24 DIMM slots)

Storage

Up to 8x 2.5-inch NVMe SSD

Overview

This ThinkSystem SR675 V3 is configured with four NVIDIA H200 NVL GPUs, a right-sized entry into Hopper-class compute for fine-tuning and private inference without an eight-way footprint. Nexus Compute specifies, builds, and validates each system through Lenovo-authorized channels with Lenovo warranty and XClarity management.

Who This Solution Is For

Departments standing up private AI capacity
Teams fine-tuning models on owned hardware
Organizations piloting before scaling GPU fleets
Groups serving in-house inference securely

Business Benefits

Right-sized Hopper

Four H200 GPUs deliver serious capability at a more accessible entry point than eight-way nodes.

Data-private AI

On-premises H200 compute keeps sensitive training and inference data inside your perimeter.

Room to grow

The SR675 V3 platform leaves headroom to expand power, storage, and networking as needs increase.

Typical Business Use Cases

1

Domain-specific model fine-tuning

2

Private and regulated inference serving

3

Departmental AI development

4

Proof-of-concept before cluster scale-out

Industry Applications

AI & Machine LearningFinancial ServicesHealthcare & Life SciencesHigher Education & ResearchGovernment & Defense

Technical Overview

The SR675 V3 4-DW GPU model hosts four double-wide PCIe Gen5 NVIDIA H200 NVL GPUs on dual AMD EPYC processors with GPUDirect RDMA support. The 3U chassis provides eight 2.5-inch NVMe bays and redundant power for a balanced, expandable accelerated node.

GPU / Accelerator4x NVIDIA H200 NVL (141GB HBM3e each)
GPU InterconnectPCIe Gen5 + NVLink bridges, GPUDirect RDMA
CPU2x AMD EPYC 9004/9005 series
MemoryUp to 1.5TB DDR5-6400 (24 DIMM slots)
StorageUp to 8x 2.5-inch NVMe SSD
NetworkingOCP 3.0 + PCIe NICs (200/400GbE optional)
Form Factor3U rackmount
Power2-4x 1800/2400W PSU (redundant)
WarrantyLenovo warranty (configurable terms)

Specifications are indicative and configured to each engagement. Request a quote for a configuration tailored to your requirements.

Warranty, Support & Fulfillment

Every system ships from an authorized channel, configured and tested, with the documentation enterprise buyers need — backed by warranty and a dedicated account team.

Enterprise Warranty

Full manufacturer warranty with optional on-site, next-business-day support and extended coverage.

Authorized Channel

Sourced through Tier-1 distribution and OEM partners — never grey market. Asset & warranty records included.

Lead Time & Deployment

48-hour quotes, then configured, burn-in tested, and delivered on a committed schedule.

Nationwide Fulfillment

Coordinated logistics, rack-and-stack, and delivery wherever your infrastructure lives.

Frequently Asked Questions

Is four H200 GPUs enough to fine-tune large models?

With 141GB per GPU and NVLink bridging, a four-way node fine-tunes substantial models and serves large-context inference. We confirm fit against your target model sizes.

Can we expand this node to eight GPUs later?

The chassis is a four-GPU model; scaling beyond it typically means adding nodes. We design networking up front so growth is straightforward.

Why on-premises instead of cloud GPUs?

Owned H200 capacity gives predictable cost, data residency, and no egress fees for steady workloads. We help model the on-prem versus cloud trade-off.

Hardware Assistance

Configure the Lenovo ThinkSystem SR675 V3 — 4x NVIDIA H200 NVL with Nexus Compute

Tell us your requirements and a hardware specialist will help you specify, configure, and quote the right system — typically within two business days. No obligation.