LenovoThinkSystem SR650a V4
A 2U server with one or two Intel Xeon 6 processors and up to four 96 GB NVIDIA RTX PRO 6000 Blackwell GPUs at the front. It's Lenovo's server for starting with AI in your data centre without moving to an 8U machine.
Rack 2UWhat it supports
The SR650a V4 is configured to order: these are its limits according to Lenovo's guide. Lenovo Press product guide
- Power
- Two redundant power supplies of up to 3,200 W, Platinum or Titanium
- Cooling
- Air, with optional Neptune Core for the processors
- Management
- XClarity Controller 3
- Operating systems
- RHEL, SLES, Ubuntu, Windows Server and VMware ESXi
- Warranty
- 1 or 3 years depending on the model, with an on-site engineer next business day
Enterprise inference and RAG in 2U
This is the machine for an internal assistant, a RAG search over your company's documentation or an agent workflow serving several departments. A single RTX PRO 6000 holds a 70-billion-parameter model at 8 bits; with four you can serve several models at once, or one larger model split across them, over PCIe rather than NVLink.
What we do
We deliver it with RHEL, Ubuntu or SLES, the NVIDIA drivers and vLLM serving the model through an OpenAI-compatible API, so your applications and agents connect to it the same way they would to OpenAI. If you already run Kubernetes or OpenShift, we join it to the cluster with the GPU Operator. And if the project is RAG over your documents, we also set up the extraction and the index: we explain how in RAG with Docling.
Supported GPUs
| GPU | Memory | How many |
|---|---|---|
| NVIDIA RTX PRO 6000 Blackwell | 96 GB | 2 at 600 W or 4 at 450 W |
| NVIDIA H200 NVL | 141 GB | 2, linked by NVLink |
| NVIDIA L40S | 48 GB | 4 |
| NVIDIA RTX PRO 4500 Blackwell | 32 GB | 4 |
| NVIDIA L4 | 24 GB | 8 |
You can start with two RTX PRO 6000 cards and move to four later: Lenovo supports the change in the field.
What your data centre needs
It takes 2U and has two redundant power supplies, which Lenovo offers from 800 to 3,200 W. The SR650i V4, with two RTX PRO 6000 cards, uses two 3,200 W units at 230 V, and that's the reference for planning the sockets, each on a separate circuit. It's air-cooled with six fans, so a rack with a working cold aisle is enough, although it packs a lot of heat into little height and the thermal load is worth checking before you put several side by side.
SR650i V4: the same machine in a fixed configuration
Lenovo also sells the SR650i V4, a fixed configuration on this chassis that you order as a single package, with a three-year warranty. The processors, GPUs and memory can't be changed.
- Processor
- 2 Intel Xeon 6530P32 cores each
- GPU
- 2 NVIDIA RTX PRO 6000 Blackwell96 GB each, at 600 W
- Memory
- 512 GB8 × 64 GB DDR5 modules at 6400 MHz
- Storage
- 2 × 3.84 TB NVMePlus 2 × 960 GB M.2 in RAID for the OS
- Network
- 2 × 25 GbE portsBroadcom 57414 SFP28
- Power
- 2 × 3,200 W power suppliesTitanium, redundant
Compared with the rest of the family
| Model | What it has | When to choose it |
|---|---|---|
| ThinkSystem SR675 V3 | Up to 8 PCIe GPUs · AMD EPYC | Up to eight GPUs with AMD, for several models or teams at once. |
| ThinkSystem SR680a V4 | NVIDIA HGX B300 · 8 GPUs | Eight B300s linked by NVLink, for training and serving large models. |
| ThinkSystem SR680a V3 with B200 | NVIDIA HGX B200 · 8 GPUs | The previous HGX, also air-cooled, when lead time or budget comes first. |
| ThinkSystem SR780a V3 | NVIDIA HGX B200 · liquid-cooled | The same eight B200s in 5U, if your data centre has a water loop. |
| ThinkSystem SR665 V3 | 4 RTX PRO 4500 · AMD EPYC | Lenovo's entry point for inference, with the Hybrid AI 241 configuration. |
What the delivery includes
From the factory
- Server built to your configuration by Lenovo
- 1- or 3-year warranty depending on the model, with an on-site engineer next business day
From us
- GPU, memory and network sizing based on your model
- Rack mounting, XClarity and up-to-date firmware
- Operating system, NVIDIA drivers and the AI stack
- Testing with your workload before it goes into production
If you need it
- 24×7 on-call support with SLA
- Lenovo warranty upgrades: 4- or 2-hour response, or one or two extra years
How we build it
What runs on top of these servers, explained in detail:
Support and training for your team
Questions about the ThinkSystem SR650a V4
What's the difference between the SR650a V4 and the SR650 V4?
The SR650 V4 is Lenovo's general-purpose 2U server. The SR650a V4 uses the same Xeon 6 generation but is designed around GPUs: they sit at the front, with power and cooling built for four double-width cards. The SR650 V4 takes at most two double-width GPUs, so if the server is mainly for databases or virtualisation and AI is an add-on, it may be enough; if the machine is for AI, go for the SR650a.
How many GPUs should I order?
It depends on the model and the users. For an internal assistant running a model of around 70 billion parameters, two RTX PRO 6000 cards are usually enough, and they also run at 600 W. If you'll serve several models at once or a lot of people, four. We size it by testing your model before the order is placed.
What warranty does it come with?
Lenovo sells it with one or three years depending on the model, with an on-site engineer next business day. It can be upgraded to 4- or 2-hour response and extended by one or two years, and on top of that we can provide our 24×7 on-call support with SLA.
Is the SR650a V4 right for what you're building?
Tell us which model you want to run and for how many people, and we'll come back with the configuration, the price and a confirmed lead time.
Blog!
Our mission
SIXE ensures that your infrastructures based on IBM®, Red Hat®, Canonical®, and SUSE® technologies remain stable, secure, and optimized for as long as you need them.
At SIXE, we are committed to sustainability. That’s why we help you design, implement, and maintain durable and efficient technological solutions.
